key: cord-354143-p2ofapbd authors: hellferscee, orienka; tempia, stefano; walaza, sibongile; variava, ebrahim; dawood, halima; wolter, nicole; madhi, shabir a.; du plessis, mignon; cohen, cheryl; treurnicht, florette k. title: enterovirus genotypes among patients with severe acute respiratory illness, influenza‐like illness, and asymptomatic individuals in south africa, 2012‐2014 date: 2017-07-06 journal: j med virol doi: 10.1002/jmv.24869 sha: doc_id: 354143 cord_uid: p2ofapbd enteroviruses can cause outbreaks of severe acute respiratory illness (sari) and ev‐a, ‐b, ‐c, and ‐d species have different pathogenic profiles and circulation patterns. we aimed to characterize and determine the prevalence of enterovirus genotypes among south african patients with respiratory illness and controls during june 2012 to july 2014. syndromic sari and influenza‐like illness (ili) surveillance was performed at two sentinel sites. at each site nasopharyngeal/oropharyngeal specimens were collected from sari and ili patients as well as controls. specimens were tested for enterovirus by real‐time pcr. positive specimens were further genotyped by sequencing a region of the vp1 gene. the prevalence of enterovirus was 5.8% (87/1494), 3.4% (103/3079), and 3.4% (46/1367) among sari, ili, and controls, respectively (sari/controls, p = 0.002 and ili/control, p = 0.973). among the 101/236 (42.8%) enterovirus‐positive specimens that could be genotyped, we observed a high diversity of circulating enterovirus genotypes (a total of 33 genotypes) from all four human enterovirus species with high prevalence of enterovirus‐b (60.4%; 61/101) and enterovirus‐a (21.8%; 22/101) compared to enterovirus‐c (10.9%; 11/101) and enterovirus‐d (6.9%; 7/101) (p = 0.477). of the enterovirus genotypes identified, echovirus 30 (9.9%, 10/101), coxsackie virus b5 (7.9%, 8/101) and enterovirus‐d68 (6.9%, 7/101) were most prevalent. there was no difference in disease severity (sari or ili compared to controls) between the different enterovirus species (p = 0.167). we observed a high number of enterovirus genotypes in patients with respiratory illness and in controls from south africa with no disease association of ev species with disease severity. pneumonia in children also include enteroviruses, rhinovirus, human bocavirus, and human coronaviruses (229e, nl63, oc43, hku1), although this remains controversial. 3, 4 although influenza and respiratory syncytial viruses have been well described as causes of pneumonia in south africa, little is known about enterovirus prevalence and circulating genotypes. 5, 6 enteroviruses are members of the enterovirus genus in the family picornaviridae. 7 the capsid protein vp1, the most variable protein containing the majority of neutralization epitopes, is commonly used to characterize enteroviruses. [7] [8] [9] more than 100 enterovirus genotypes are currently classified into four species, hev-a, hev-b, hev-c, hev-d, which have different pathogenic profiles and circulation patterns. 7, 10 enteroviruses (ev) can cause symptoms similar to a mild cold, but have also been associated with severe respiratory infection requiring hospitalisation and may be fatal. 10 an outbreak of a re-emerging ev-d68 lineage causing severe acute respiratory illness (sari) occurred in the usa and canada in 2014, and these lineages were also detected in europe, asia, and south america from 2012 to 2013 samples. [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] though evs have been associated with sari and influenza-like illness (ili), 5 these viruses have been less well characterized in south africa. we identified high diversity among evs circulating in hospitalized south african patients, 25 however it is unknown whether different ev genotypes are associated with mild or severe respiratory illness in south africa. we aimed to characterize the ev genotypes circulating among south african patients with sari, ili, and asymptomatic individuals from june 2012 to july 2014. in addition we assessed the association of different ev species with mild (ili) or severe (sari) illness compared to asymptomatic individuals. an asymptomatic individual or control was defined as a person presenting at the same outpatient clinics with no history of fever, respiratory, or gastrointestinal symptoms during the 14 days preceding the visit. these individuals commonly presented to the clinics for visits such as dental procedures, family planning, well baby visits, voluntary hiv counseling and testing, or acute care for non-febrile illnesses. we aimed to enroll one hiv-infected and one hiv-uninfected control every week in each clinic within each of the following age categories: 0-1, 2-4, 5-14, 15-54, and ≥55 years. individuals were not followed-up to ensure they remained asymptomatic after enrollment. study staff completed case report forms for all enrolled sari and ili cases as well as controls. in addition, for sari cases, hospital records were reviewed to assess disease progression and outcome (ie, discharge, transfer, or in-hospital death). referral to hospital was recorded for all enrolled ili cases. ili cases that were referred to hospital were excluded from the analysis, as these patients did not adhere to the case definition of ili anymore. hiv results were obtained from a combination of two sources: (i) patient clinical records when available and (ii) for consenting patients, an anonymized linked dried blood spot was tested at the national institute for communicable diseases (nicd). when both results were available, the nicd result was used. nasopharyngeal aspirates for children <5 years of age and nasopharyngeal and oropharyngeal swabs from persons ≥5 years of age were collected from all enrolled participants and placed in universal transport medium (copan, murruieta, ca). all respiratory tract specimens were tested for the presence of 10 respiratory viruses (influenza virus a and b, respiratory syncytial virus, parainfluenza virus types 1, 2, and 3, adenovirus, human metapneumovirus, and rhinovirus) including ev, using a multiplex real-time reverse transcriptase polymerase chain reaction assay. 26 the real-time reverse transpriptase polymerase chain reaction used for ev testing, is a sensitive method for the routine detection of all members of the enterovirus genus. this method has a weak cross-reactivity to hightitre rhinovirus stocks, but not with rhinovirus-positive clinical samples. 27 all specimens testing positive for enterovirus were selected for molecular characterization. a 400 base pair region of the vp1 gene was amplified and sequenced as previously described using primer set 224/222 during the first round of amplification and primer set an89/an88 for the nested amplification. 8 this assay was originally designed to detect and identify ev, but also detects and identifies a subset of rhinoviruses. amplicons were purified using the exosap enzyme system the genetic diversity of each vp1 sequence was first determined by comparison with the reference strains in genbank (us national center for biotechnology information, ncbi, http://www.ncbi.nlm.nih.gov/, accessed 01 july 2016) using blast (basic local alignment search tool) and confirmed by phylogenetic analysis of the partial vp1 sequences. multiple sequence alignments were generated in the mafft (multiple alignment using fast fourier transform) multiple sequence alignment program 28 and analyzed in the bioedit sequence alignment editor software. 29 phylogenetic trees were generated using the neighbor-joining method and genetic distances were calculated with the kimura-2-parameter model using mega (molecular evolutionary genetics analysis) 6 software. 30 the statistical significance of the phylogenies was estimated by bootstrap analysis using 1000 pseudo replicates. sub analysis was similarly done for the three dominant ev species in this study. sequences of enterovirus partial vp1 genes generated in this study have been deposited in genbank with the following accession numbers: kx940982-kx941096. the chi-squared or fisher's exact test were used for comparison of categorical variables. unconditional exact logistic regression was used to assess the association of ev species with disease severity among patients with mild (ili) or severe illness (sari) using asymptomatic individuals as control group. exact logistic regression was used to account for the fact that no ev-d species was detected among controls. ev-b was used as the comparison group as it was the species most frequently detected. significance was assessed for p < 0.05. all models were adjusted for age, hiv serostatus and underlying medical conditions. the analysis was performed using stata 14 on multivariable logistic regression analysis using ev species b as the reference group and controlling for age, hiv status and underlying illness, no association with disease severity (sari or ili compared to controls) was identified for any of the ev species (supplementary table s1 ). enterovirus genotypes could be identified in 59% (51/87), 33% (34/103), and 35% (16/46) of enterovirus-positive samples among sari and ili cases and controls, respectively. we identified a total of 33 genotypes, distributed among all four ev species. the most prevalent genotype was e30 (9.9%, 10/101), followed by cvb5 (7.9%, 8/101) and ev-d68 (6.9%, 7/101). all other genotypes identified in the study were detected in <5% of genotyped samples (supplementary fig. s2 ). all e30 strains identified in this study displayed >98% nucleotide (nt) homology and formed a distinct bi-phyletic cluster with 79% bootstrap support (fig. 3) . in the bi-phyletic cluster, strains from edendale sari cases and klerksdorp sari and ili cases are distinctly located in each of the sub clusters with 92% bootstrap support, respectively. the e30 strains in our study differ from other e30 genotypes by mean nucleotide pairwise distance of 20.2% (15.6-27.5 ) and mean amino acid sequence distance of 6% (2.7-11.6) in amino acid sequence. all cvb5 strains identified in our study (six sari samples and two ili samples from edendale) were categorized as genogroup c and formed a distinct sub cluster with a bootstrap value of 91% (fig. 4) . the nucleotide the majority (71.4%, 5/7) of ev-d68 strains identified in 2013/2014 distinctly clustered in lineage b2 (89% bootstrap support) of which three were ili cases and two were sari cases, together with strains from one of the two co-circulating ev-d68 lineages that caused the large 2014 usa outbreak (fig. 5) . the nucleotide sequences for evd-68 strains were downloaded from genbank and sequences depicted in fig. 5 represent all known lineages. the lineage b ev-d68 strains in our study differed from other lineage b ev-d68 strains by 1.5% nucleotide and 1.4% amino acid sequence. these included two sari cases (one from klerksdorp and one from edendale) and three ili cases (one from klerksdorp and two from edendale). we describe the ev species circulating among sari and ili cases and asymptomatic controls in south africa. we observed a high diversity of circulating ev genotypes (a total of 33 genotypes) from ev species a-d with high circulation rates for ev-b and ev-a compared to ev-c and ev-d. from our systematic surveillance across all age groups, the majority of patients testing positive for ev were <5 years of age. we did not observe any difference in disease association due to different ev species since ev disease association cannot be conclusively ascribed since the analysis did not include coinfections with other pathogens. epidemics of human ev disease display a seasonal pattern, with infections more common in summer and early autumn in geographical regions with a temperate climate. 7, 33 most clusters of ev-d68 showed an atypical late seasonality compared to other evs, with a peak in autumn, instead of summer. 7 a common feature of e30 molecular epidemiology is the progression of circulating lineages within one prevalent genotype. 9, 31 the e30 strains identified in this study clustered together (designated genotype k) although no differences were observed between viruses from sari compared to ili cases. there is no report of a concurrent e30 aseptic meningitis outbreak at the same time the e30 np/op positives were detected. it has been reported that genotypes of e30 can become dominant for 3-4-year periods before they disappear from circulation, 34 and they are therefore known as epidemic strains. in our study, it seems that e30 genotype k became dominant and circulated in south africa during 2012-2014. phylogenetic studies with vp1 sequences indicate that e30 variants are continuously emerging and replacing the circulating e30 strains, largely due to the error prone ev rna polymerase 31 . cvb5 has been detected for over 50 years and, similar to e30, sporadic cases of aseptic meningitis, as well as outbreaks, have been reported, remaining one of the most predominant reported ev genotypes in a number of countries. 35 all of the cvb5 strains identified in this study formed a sub cluster in genogroup c. phylogenetic clustering by year of study enrolment was observed for the south african strains. different co-circulating strains of ev are often observed in ev disease outbreaks as ev genomes vary with respect to time but may also vary by geographical distribution. 36 in 2014, the largest outbreak to date of severe respiratory illness associated with ev-d68 occurred in the usa, and almost all of the case patients were children. 37 coinciding with the 2014 usa outbreak the mechanisms for emergence of ev-associated outbreaks are not known; however, a combination of virus-specific, population-level, and other external factors are likely to be involved. 7 our study has some limitations. previous studies observed a difference in the predominant ev genotypes among neonates and older children 39 but we were not powered for this analysis. some rhinovirus sequences were detected in samples that tested positive for ev on our in-house assay indicating that our ev detection assay is not 100% specific to detect only evs. this study was not designed to describe the role of co-infections with other respiratory pathogens. in addition the association with ev detection with mild and severe illness was not investigated. in conclusion, we showed that there was a high diversity in the vp1 sequences of the ev species circulating in south africa during 2012-2014 and most of these genotypes are associated with meningitis worldwide. ev was detected in outpatient and hospitalized patients with sari but was also detected in controls. we determined no disease association of ev species with disease severity; however, some genotypes (e30, cvb5, ev-d68) were more prevalent in symptomatic cases. further studies are needed to determine if other factors such as viral load or host interactions play a role in ev-associated disease as well as a robust spatiotemporal phylogenetic analysis based on complete vp1 sequences. we thank all members involved in sari and ili surveillance for the collection of specimens and data management. this work was supported by the national health laboratory service, south africa, and the united states centers for disease control and prevention, atlanta, georgia, usa (co-operative agreement number: 5u51ip000155). the findings and conclusions in this paper are those of the authors and do not necessarily represent the views of their affiliated institutions or the agencies funding the study. childhood pneumonia-progress and challenges who pneumonia fact sheet no 331 development and implementation of a molecular diagnostic platform for daily rapid detection of 15 respiratory viruses multiplex real-time pcr for detection of respiratory tract infections the role of influenza, rsv and other common respiratory viruses in severe acute respiratory infections and influenza-like illness in a population with a high hiv sero-prevalence epidemiology of influenza virus types and subtypes in south africa the epidemiology of nonpolio enteroviruses seminested pcr amplification of vp1 sequences for direct identification of all enterovirus serotypes from original clinical specimens typing of human enteroviruses by partial sequencing ofvp1 picornavirus and enterovirus diversity with associated human diseases molecular and epidemiological study of enterovirus d68 in taiwan coexistence of two clades of enterovirus d68 in pediatric swedish patients in the summer and fall of 2014 enterovirus d68-associated community-acquired pneumonia in children living in first enterovirus d68 (ev-d68) cases detected in hospitalised patients in a tertiary care university hospital in spain emergence of enterovirus d68 in denmark severe respiratory illness associated with a nationwide outbreak of enterovirus d68 in the usa (2014): a descriptive epidemiological investigation enterovirus d68: a focused review and clinical highlights from the epidemiology of enterovirus d68 in ontario the emergence of enterovirus d68 in a dutch university medical center and the necessity for routinely screening for respiratory viruses european surveillance for enterovirus d68 during the emerging north-american outbreak in 2014 low-level circulation of enterovirus d68-associated acute respiratory infections high frequency of enterovirus d68 in children hospitalised with respiratory illness in norway systematic community-and hospital-based surveillance for enterovirus-d68 in three canadian provinces enterovirus d68 infection enterovirus d68 and other enterovirus serotypes identified in south african patients with severe acute respiratory illness respiratory viral coinfections identified by a 10-plex real-time reverse-transcription polymerase chain reaction assay in patients hospitalized with severe acute respiratory illness-south africa rapid and sensitive routine detection of all members of the genus enterovirus in different clinical specimens by real-time pcr mafft multiple sequence alignment software version 7: improvements in performance and usability bioedit: a user-friendly biological sequence alignment editor and analysis program for windows 95/98/nt mega4: molecular evolutionary genetics analysis (mega) software version 4.0 phylogeography of circulating populations of human echovirus 30 over 50 years: nucleotide polymorphism and signature of purifying selection in the vp1 capsid protein gene worldwide emergence of multiple clades of enterovirus 68 phylogenetic patterns of human coxsackievirus b5 arise from population dynamics between two genogroups and reveal evolutionary factors of molecular adaptation and transmission molecular epidemiology of echoviruses 11 and 30 in russia: different properties of genotypes within an enterovirus serotype an outbreak of aseptic meningitis caused by a distinct lineage of coxsackievirus b5 in china genetic variation of coxsackie virus b5 strains associated with aseptic meningitis in greece severe respiratory illness associated with enterovirus d68-missouri and illinois enterovirus d-68 infection, prophylaxis, and vaccination in a novel permissive animal model, the cotton rat (sigmodon hispidus) presence of human non-polio enterovirus and parechovirus genotypes in an amsterdam hospital in 2007 to 2011 compared to national and international published surveillance data: a comprehensive review key: cord-338674-tnnd1s57 authors: yin, j kevin; lahra, monica m; iskander, mary; lambert, stephen b; heron, leon; nissen, michael d; rost, laura; murphy, jennifer; sloots, theo p; booy, robert title: pilot study of influenza vaccine effectiveness in urban australian children attending childcare date: 2011-06-10 journal: j paediatr child health doi: 10.1111/j.1440-1754.2011.02098.x sha: doc_id: 338674 cord_uid: tnnd1s57 background: influenza outbreaks in the childcare setting are a significant cause of excess winter morbidity. this study explored methods of follow up and sample collection for a proposed randomised controlled trial of influenza vaccination in children attending childcare. methods: the study was conducted in four sydney childcare centres during 2007. healthy children aged 6–59 months eligible for vaccination were recruited in two centres, with another two acting as controls. data on influenza‐like illness (ili: ≥37.8°c plus at least one respiratory symptom) occurrence were collected weekly. in those children with an ili, parents were asked to collect nasal swabs and send via surface mail for viral polymerase chain reaction. vaccine efficacy (ve) for ili was estimated overall and for subgroups aged 6–23 and 24–59 months using the formula ve = 1 − relative risk (rr). results: sixty‐three per cent (151/238) of eligible children had parents give consent. sixty‐three children received influenza vaccine and 88 participated as controls. of 26 specimens returned, a virus was detected in 18 (69%); none with influenza. two symptomatic children had positive near‐patient influenza tests in general practice (one a vaccine failure). the rr with 95% confidence interval in all children and those aged 6–23 months were less than one, 0.56 (0.32–1.02) and 0.46 (0.15–1.45), respectively. conclusions: this study demonstrated the feasibility and utility of parent‐collected and mailed respiratory specimens for ve research in the childcare setting. two‐thirds of parent‐collected swabs proved positive for at least one virus. finding ways to reduce reluctance of parents to submit samples could improve the representativeness of samples collected and the power of the study. no evidence was found for influenza ve, but point estimates were in the direction of protection. influenza is a seasonal, vaccine-preventable disease which causes excess morbidity and mortality during winter in temper-ate climates. the health and economic costs associated with childhood influenza are substantial. 1 for example, in australia during 2002-2005, there were reports of 25 433 hospitalisations and four deaths for influenza and pneumonia among children aged under 5 years. 2 the annual cost due to influenza-related diseases in australia is estimated to exceed $115 million. 3 the world health organization recommends annual influenza vaccination as the cornerstone for prevention and control. efficacious influenza vaccines have been available for over 50 years, and yet, routine use in childhood remains the what is already known on this topic 1 children in childcare are more likely to contract influenza and transmit infection to their siblings, parents, extended families and child-care workers. 2 usa, canada and western australia currently have a routine influenza vaccine policy in place that includes children 6 months of age and older. 3 evidence for the effectiveness of influenza vaccine in children aged less than 24 months is limited and high quality, appropriately powered, randomised controlled trials are needed. 1 it is feasible to follow children weekly for 3 months to obtain swabs for influenza-like illness. 2 two-thirds of parent-collected swabs were positive for at least one virus demonstrating the utility of this approach for future studies. reluctance of parents to submit swabs for analysis may be a limitation of this approach. exception in most countries. the effectiveness of influenza vaccine for children in childcare has been demonstrated for children aged ն24 months. [4] [5] [6] [7] [8] although the us advisory committee on immunisation practices has recommended children aged 6-23 months to be vaccinated with influenza vaccine since 2004, 9 there is ongoing debate about vaccine effectiveness in this age group. three recent systematic reviews [10] [11] [12] concluded that either influenza vaccine was not effective in children յ24 months of age or that there were insufficient data to form a conclusion. influenza is transmitted from person to person through contact and respiratory droplets; however, the droplets do not remain suspended in the air for long nor do they travel far. 13 transmission of influenza generally requires close contact with an infected person or contact with a contaminated surface or object. 14, 15 the childcare setting provides enhanced opportunities for transmission of infections including influenza as there is prolonged close interaction between young children and the sharing of toys and other objects. further to this, young children are particularly susceptible to infection as they are immunologically naïve to many viruses. commercial childcare in australia is available in two broad categories: daycare centres (dcc) for children aged 6 weeks until 6 years and pre-school centres (psc) for children aged 3 to 6 years. commercial childcare usage in australia is increasing. the median attendance time for australian children who use childcare is 10 h per week, but 13% attend 35 h a week or more. 16 children in childcare are known to be more likely to contract respiratory illnesses, including influenza, [17] [18] [19] [20] [21] and are considered to be major transmitters of influenza to their siblings, parents, extended families and care workers. 6, 8, 22, 23 the 2007 influenza season in australia ran from late may until october and notifications peaked during august. 24 australia witnessed antigenically drifted influenza virus (a/brisbane/ 59/2007 (h1n1)-like and a/brisbane/10/2007 (h3n2)-like), and it was the most severe influenza season since a national influenza reporting system was established in 2001. 25 with this study, the primary process issues of conducting influenza vaccine research in the childcare environment were evaluated, including recruitment, retention, vaccination and specimen handling. while this pilot study was not powered to assess an efficacy end point, preliminary vaccine efficacy (ve) data were also examined. from july to august 2007, children aged 6-59 months attending four childcare centres in new south wales were recruited for this study: two dcc caring for children aged 0-59 months and two psc caring for children aged 36-59 months. the four dcc were chosen by convenience (proximity to the children's hospital at westmead) with equal number of children between dcc and psc. one dcc and one psc were allocated to influenza vaccination, and one dcc and one psc were allocated to be controls. this study was approved by the royal alexandra hospital for children ethics committee, and informed parental consent for participation was obtained prior to study procedures. the par-ticipating children were evaluated in two age groups based on age at enrolment: 6-23 months and 24-59 months. the influenza vaccine administered was a 2007 southern hemisphere preparation, purified, inactivated, split vaccine (vaxi-grip junior, provided by sanofi pasteur, lyon, france), incorporating: children were administrated the vaccine according to the standard recommended dose and schedule for age. 26 as all children at the centres randomised to receive vaccine were influenza vaccine naïve, each received two doses of vaccine 1 month apart -0.25 ml intramuscular dose for those less than 36 months of age and 0.5 ml intramuscular dose for those aged 36-59 months at the time of their first dose. vaccines were administered between 11 july 2007 and 19 september 2007. we defined influenza-like illness (ili) as an illness with fever >37.8°c and with one or more respiratory symptoms (cough, blocked nose or runny nose) to maximise sensitivity. as a protective level of antibody is usually detectable within 2 weeks of the second dose of vaccine, 13,27 ili surveillance was commenced in vaccinated children at this time point. in control children, ili surveillance was arbitrarily commenced from the week ending 26 august 2007: at this time, just over half (32/62) of the children eventually fully vaccinated had received vaccine, and from that week, the ratio of child-weeks of follow up in vaccinated and unvaccinated children was similar (fig. 2) . parent education for ili surveillance was provided at study entry. households received a weekly email or telephone call from 30 july until 21 october 2007 (12 weeks) to monitor the study children for ili symptoms. parent training for the collection of nasal swabs was conducted by study nurses after the second immunisation. nasal swabs were collected using the virocult collection system (mw950) consisting of a rayon swab on a plastic shaft, with viral transport medium-soaked foam pad in the base of the transport tube (copan italia, brescia, italy), and were returned to the queensland paediatric infectious diseases laboratory using a pre-addressed, postage-paid envelope. returned specimens were tested using previously reported, real-time polymerase chain reaction assays with reverse transcription for rna viruses. a total of 16 viruses were investigated: human rhinoviruses (hrv), 28 influenza a, influenza b, rsv, adenoviruses, hmpv, parainfluenza viruses i, ii and iii, 29 bocavirus 30 hpyv-wu, hpyv-ki, 31 and human coronaviruses: oc43, 229e, nl63 32 and hku1. 33 while it was not part of the study protocol, some children had near-patient influenza tests performed by their general practitioners and these were reported by parents to study staff. ili incidence rates were calculated using child-weeks in the denominator. rate ratios (rr) and 95% confidence intervals were calculated comparing vaccinated and unvaccinated groups. these values were used to estimate ve using the formula ve = (1 -rr) ¥ 100%. comparisons were performed in three age groups, 6-59 months (all children), 6-23 months and 24-59 months. the average ages of the children in the vaccine and control centre were 43.4 (7.7-65.4) and 44.1 (7.9-66.0) months, respectively, while the proportions who were males were 50.8 and 58.0%. data on non-enrolled children were not recorded. there were 239 children in total attending the four childcare centres, and 151 children were enrolled giving a recruitment rate of 63%. complete information was available for analysis in 150 children, with one vaccinated child lost to follow up during the study period (fig. 1 ). there were 481 (62 children) and 792 (88 children) child-weeks of follow up in vaccine and control centres, respectively. a total of 59 ilis were identified in all study children, and weekly ili incidence rates are provided (fig. 2) . of the 26 swabs received during the study period, 18 (69%) had at least one virus identified, with bocavirus being the most common virus found in six swabs (table 1) , followed by hrv in five swabs. one swab contained three viruses (hrv, adenovirus and bocavirus). thirteen of these swabs were from 12 vaccinated children, with the other half from 13 controls. two positive near-patient influenza tests (one a vaccine failure) were reported by parents to the study staff: one was from an unvaccinated child in a control centre; the other child was vaccinated with the test done 14 days post-second vaccination ( table 1) . there were a total of 59 ilis identified with efficacy point estimates in the direction of protection for all age-groups but not significant ( table 2 ). the key findings of this pilot study were that a high recruitment rate could be achieved, that recruited families were tolerant of regular weekly follow up over an extended period (3 months) and that there was no evidence of protective efficacy, but point estimates of ve for the less-specific end point of ili were in the direction of protection. it is inevitable that ili would include non-influenza infections which cause respiratory signs and symptoms, especially as we used a sensitive definition (at the expense of specificity), so it is not surprising that a range of other viral pathogens were identified in our study. a population-based surveillance study showed that less than 10% of hospitalised children aged յ59 months with ili had confirmed influenza infection. 34 our study has some limitations. the childcare centres were not randomised. the commencement midway through an influenza season limited the number of influenza cases identified. less than half the episodes in children of ili (26 out of 59, 44%) had a respiratory sample sent. this reduced sampling is probably due to the added burden on parents of sample collection (and posting) while a child is ill. in addition, as the childcare centres were in the suburbs with relatively lower socio-economic indexes for areas 35 (and also involved larger families), this, too, may have limited parental cooperation. the participants were only followed from the 2nd half of august when the 2007 season was peaking, so some may have thought that sample collection in september or october was too late. given the limited data that were collected on symptomatology and the relatively small number of specimens, it was not possible to do an extensive analysis comparing symptoms by virus type to address if there are differences in symptoms among various viruses. furthermore, the demographic data (e.g. sex, age range) of those who did not participate in the study were not collected; therefore, it was not possible to identify if there was any recruitment bias. greater efforts are required for future studies in (i) improving the proportion of swabs collected and sent by initiating ili follow up before the influenza season starts and findings better ways to overcome parents' reluctance in submitting swabs; (ii) obtaining more detailed data on symptomatology of respiratory infection; and (iii) collecting de-identified demographic data on those who are not enrolled in the study. this study showed no evidence for influenza ve. there was only a suggestion of protection in that all the point estimates were in that direction. trivalent, live, cold-adapted influenza vaccine (caiv-t) may be a better option for young children and has been demonstrated to have significantly higher efficacy than inactivated vaccine among young children during moderate 36, 37 and high attack-rate influenza seasons. 38 caiv-t was also able to provide protection even when the circulating influenza virus was an antigenically distinct strain. 37 there were two children with positive results for influenza a from the near-patient test. this type of test is known to have only moderate sensitivity but high specificity, so a positive test is unlikely to be false. 39 week ending date (2007 older children (24-59 months) are more likely to have developed natural immunity through previous infection and are more likely to have developed better personal hygiene. for these reasons, older children are less susceptible to influenza as well as other non-influenza infections that may cause ili. this pilot study has shown the feasibility and value of parentcollected (and mailed) nasal sample for influenza ve research in childcare. two-third of parent-collected swabs proved positive for at least one virus. means of lessening reluctance of parents to submit respiratory samples from their children need to be found to improve the representativeness of samples collected and the power of the study. no evidence was found for influenza ve, but point estimates were all in the direction of protection. the cost of community-managed viral respiratory illnesses in a cohort of healthy preschool-aged children vaccine preventable diseases and vaccination coverage in australia influenza-related disease: the cost to the australian healthcare system influenza a vaccine decreases the incidence of otitis media in 6-to 30-month-old children in day care the effect of mass influenza immunization in children on the morbidity of the unvaccinated elderly effectiveness of influenza vaccination of day care children in reducing influenza-related morbidity among household contacts studies of the 1996-1997 inactivated influenza vaccine among children attending day care: immunologic response, protection against infection, and clinical effectiveness influenza burden of illness, diagnosis, treatment, and prevention: what is the evidence in children and where are the gaps? prevention and control of influenza: recommendations of the advisory committee on immunization practices (acip) vaccines for preventing influenza in healthy children the efficacy of influenza vaccine for healthy children: a meta-analysis evaluating potential sources of variation in efficacy estimates including study quality influenza vaccine in healthy children: a meta-analysis prevention and control of seasonal influenza with vaccines: recommendations of the advisory committee on immunization practices (acip) transmission of influenza a in human beings social contact networks for the spread of pandemic influenza in children and teenagers risk of respiratory illness associated with day-care attendance: a nationwide study form of day care and respiratory infections among finnish children lower respiratory illness, recurrent wheezing, and day care attendance day care centers and respiratory health frequency and severity of infections in day care: three-year follow-up influenza virus infections in seattle families, 1975-1979. ii. pattern of infection in invaded households and relation of age and prior antibody to occurrence of infection and related illness the seattle virus watch. iv. comparative epidemiologic observations of infections with influenza a and b viruses, 1965-1969, in families with young children contribution of influenza and respiratory syncytial virus to community cases ofinfluenza-like illness: an observational study annual report of the national influenza surveillance scheme australian government department of h, ageing c. national health and medical research council. the australian immunisation handbook amplification of rhinovirus specific nucleic acids from clinical samples using the polymerase chain reaction comparing nose-throat swabs and nasopharyngeal aspirates collected from children with symptoms for respiratory virus identification using real-time polymerase chain reaction detection of human bocavirus in respiratory, fecal, and blood samples by real-time pcr development and evaluation of real-time pcr assays for the detection of the newly identified ki and wu polyomaviruses real-time rt-pcr detection of 12 respiratory viral infections in four triplex reactions human coronavirus infections in rural thailand: a comprehensive study using real-time reverse-transcription polymerase chain reaction assays community epidemiology of human metapneumovirus, human coronavirus nl63, and other respiratory viruses in healthy preschool-aged children using parent-collected specimens census of population and housing: socio-economic indexes for areas (seifa), australia live attenuated versus inactivated influenza vaccine in infants and young children various factors associated with the manifestation of influenza-like illness trivalent live attenuated intranasal influenza vaccine administered during the 2003-2004 influenza type a (h3n2) outbreak provided immediate, direct, and indirect protection in children laboratory diagnosis of human seasonal and pandemic influenza virus infection key: cord-256943-71tnv4lp authors: santillana, mauricio; nsoesie, elaine o.; mekaru, sumiko r.; scales, david; brownstein, john s. title: using clinicians’ search query data to monitor influenza epidemics date: 2014-08-12 journal: clinical infectious diseases doi: 10.1093/cid/ciu647 sha: doc_id: 256943 cord_uid: 71tnv4lp search query information from a clinician's database, uptodate, is shown to predict influenza epidemics in the united states in a timely manner. our results show that digital disease surveillance tools based on experts' databases may be able to provide an alternative, reliable, and stable signal for accurate predictions of influenza outbreaks. the discovery of unusual outbreaks often depends on individual health practitioners who can promptly identify abnormal circumstances and then report those concerns to the greater community [1, 2] . although the impact of these reports cannot be overstated, recent developments in internet technologies have demonstrated the power of the crowd as well. for example, crowdsourcing approaches allow members of the public to complete tasks relevant to a larger goal [3] . search activity on diseases such as influenza and dengue has been shown to correlate with traditional surveillance data in multiple instances [4] [5] [6] [7] [8] . google flu trends (gft) demonstrated a link between influenzarelated search query data and the centers for disease control and prevention's (cdc) influenza-like illness (ili) index [5] . other examples include the use of search query data from yahoo! [9] and from baidu [8] to track influenza epidemics. internet search queries are available much earlier than data from validated traditional surveillance systems and have the potential to provide timely epidemiologic intelligence to inform prevention messaging and healthcare facility staffing decisions. the potential for the public's search activity to be influenced by anxiety, fears, and rumors raises concerns regarding reliability [10] [11] [12] [13] . although recent revisions to gft have shown that these concerns can be partially mitigated [13] [14] [15] , shifting internet-based surveillance from the entire public to subjectmatter experts may maintain timeliness while generating a more reliable and stable signal requiring much less data. a recent small retrospective study using data on queries to a finnish primary care guidelines database demonstrated, for example, that disease-specific queries for lyme disease, tularemia, and other infectious diseases correlated well with concurrent confirmed cases [16] . here, we show that uptodate (www.uptodate.com), a physician-authored clinical decision support internet resource that is used by 700 000 clinicians in 158 countries and almost 90% of academic medical centers in the united states, can be used for syndromic surveillance of influenza. specifically, we use upto-date's search query activity related to ili to design a timely sentinel of influenza incidence in the united states. uptodate is a professional database utilized by healthcare practitioners for point-of-care decisions. the information provided is rigorously authored and edited by experienced physicians. also, uptodate topics are accessed >18 million times monthly, and studies suggest that information provided through the site helps improve healthcare outcomes in hospitals [17] [18] [19] . in collaboration with uptodate, we obtained search volume of 23 search terms related to ili, as well as overall search activity from november 2011 to november 2013 for us accounts only. the search terms were as follows: influenza, haemophilus influenzae, flu, parainfluenza, h1n1, h7n9, h5n1, h3n2, grippe, gripe, adenovirus, rhinovirus, respiratory syncytial virus, metapneumovirus, coronavirus, bordetella pertussis, mycoplasma pneumoniae, pneumonia, bronchitis, h9n2, sinusitis, upper respiratory tract infection, and tamiflu. we obtained a weekly search fraction for each search term, at any given point in time, by dividing the number of searches for a given phrase by the total number of searches in the uptodate database, thus minimizing the effects of variation in the overall use of the uptodate database through time. we also obtained the national ili weekly index from the cdc for the same time period to use as a comparator (available at: http://www.cdc.gov/flu/weekly/pastreports.htm). we built a collection of multivariate linear models using the z scores of the aforementioned 23 search terms' weekly search fraction as explanatory variables and the cdc ili index as our dependent variable. the multiplicative coefficients associated with each search term in each multivariate linear model were updated weekly as the cdc ili index was updated. our multivariate models can be expressed as where i(t) is the percentage of national ili physician visits, q i (t) is the search fraction associated with term i at time t, α i (t) is the multiplicative coefficient associated with each term at time t, and e is the normally distributed error term. model selection was performed using a least absolute shrinkage and selection operator (lasso) technique [20] at every single week incorporating new cdc ili information as it became available. therefore, our approach recalibrated weekly the relevance of the search activity for each individual term according to its historical prediction ability. the lasso technique uses an optimization algorithm that favors models that minimize the mean squared error between the observations and predictions, while penalizing models containing many variables by simultaneously minimizing the sum of the absolute size of the regression coefficients. we produced real-time estimates of ili activity at time t, assuming that (1) we only had access to cdc-reported ili data up to 2 weeks prior (ie, up to t-2 weeks), and (2) assuming that we had access to the real-time (time = t) number of searches in the uptodate database. our dynamic approach is similar to the one presented in santillana et al [15] , and inspired by data assimilation techniques widely used in weather forecasting and oceanography [21, 22] and supervised machine-learning techniques [20] . our methodology was implemented in matlab version r2011a. the lasso routine was obtained from (available at: http://www. stanford.edu/~hastie/glmnet_matlab/) in november 2013 [23] . the training period for our first prediction comprised 26 weeks (5 november 2011-28 april 2012). thus, our first real-time figure 1 shows our real-time estimates and the cdc-reported ili visits. gft estimates are included for context. our estimates predict very well the cdc-reported ili visits and outperform gft estimates during the prediction period. moreover, our approach estimates accurately the peak of the 2012-2013 influenza season (in the week of 30 december 2012) and produces a slight overestimation of the influenza epidemic curve in the second week of january 2013 (overestimating the flu activity by approximately 25% in relative terms-ie, 5.6% of ili as opposed to the actual 4.5%). this overestimation is minimal when compared to the gft estimates (overestimating the influenza activity by 130% in relative terms-ie, 10.5% of ili as opposed to the actual 4.5%). our methodology has strong predictive power (pearson correlation of 0.972; a root mean square error [rmse] of 0.2829%) during the prediction period starting in the week of 12 may 2012 and ending in the last week of november 2013. although gft has a very high pearson correlation (0.9499) during this same time period, it clearly fails to produce reliable estimates for the peak of the 2012-2013 influenza season. this mismatch is better captured by the rmse, which shows that gft estimates are on average off by 1.4% of the national population (ie, almost 5 times larger than our rmse). in figure 2 we present a heatmap representing the relevance of each search term in predicting influenza activity as a function of time, during the validation time period. the term tamiflu is the strongest predictor, whereas sinusitis, influenza, h1n1, and coronavirus display relevance as predictors during different time periods. our findings demonstrate that combining a robust dynamic methodology and subject-matter experts' search activity more accurately predicts influenza activity than the well-established internet-based tool google flu trends. specifically, the model presented here has numerous strengths compared to gft. first, the model does not require expert supervision to adjust the search terms over the course of the influenza season. our clinicians' tamiflu search activity among clinicians is highly correlated with centers for disease control and prevention-reported influenza-like illness and thus is found to be the strongest predictor by our algorithm. sinusitis, influenza, h1n1, and coronavirus display significant relevance as predictors during different time periods. approach can also accommodate and identify changes in clinicians' selection of search terms over time while retaining the model's predictive power as demonstrated in figure 2 . not only does this strength address evolving medical vocabulary, it also avoids "model drift" (static models typically match the training data well; however, as time progresses its deviation from truth may cause its predictions to drift farther and farther from truth (as seen in cook et al [10] with gft). the success of our approach suggests that low volumes of queries (in the order of 100-10 000 seconds) in relevant subject-matter experts' databases, such as uptodate, provide a promising way to identify meaningful signals to track influenza activity. this will motivate the need for future research aimed at testing the accuracy of our methodology at state and city levels, and potentially in the prediction of other diseases. moreover, our findings in combination with those shown in jormanainen et al [16] suggest that data acquired from specialized databases may have an improved signal-to-noise ratio and may be less likely to be impacted by public disruption resulting from anxiety or media reports on increased morbidity and mortality during (novel) outbreaks of influenza. limitations in this data source include those inherent in most novel data sources advanced for monitoring infectious diseases. although timely, these data sources lack the specificity observed in traditional surveillance systems, which rely on hierarchical reporting procedures. these data streams therefore supplement traditional disease surveillance provided by organizations such as the cdc. finally, uptodate data is not publicly available and thus not ready to be used as an alternative disease detection sentinel. in this study, we demonstrate that search queries from the up-todate database in conjunction with a dynamic multivariate methodology can be successfully utilized to obtain real-time estimates of influenza incidence in the united states before the release of official reports. clinicians can use outcomes from the model to monitor estimated levels of influenza in the united states. we also discuss the potential usefulness and limitations of digital data sources for infectious disease surveillance based on search query data [5, 7, 8, [24] [25] [26] [27] [28] . future work may include analysis of smaller geographic units. promed-mail: an early warning system for emerging diseases evaluation of promedmail as an electronic early warning system for emerging animal diseases: 1996 to crowdsourcing-harnessing the masses to advance health and medicine, a systematic review using internet searches for influenza surveillance detecting influenza epidemics using search engine query data a new approach to monitoring dengue activity using web search query data to monitor dengue epidemics: a new model for neglected tropical disease surveillance monitoring influenza epidemics in china with search query from baidu use of prediction markets to forecast infectious disease activity assessing google flu trends performance in the united states during the 2009 influenza virus a (h1n1) pandemic when google got flu wrong reassessing google flu trends data for detection of seasonal and pandemic influenza: a comparative epidemiological study at three geographic scales the parable of google flu: traps in big data analysis google disease trends: an update what can digital disease detection learn from (an external revision to) google flu trends? physicians' database searches as a tool for early detection of epidemics the value of library and information services in patient care: canadian results from an international multisite study how doctors make use of online, point-of-care clinical decision support systems: a case study of uptodate association of a clinical knowledge support system with improved patient safety, reduced complications and shorter length of stay among medicare beneficiaries in acute care hospitals in the united states regression shrinkage and selection via the lasso data assimilation in meteorology and oceanography data assimilation and its applications glmnet for matlab influenza forecasting with google flu trends twitter catches the flu: detecting influenza epidemics using twitter separating fact from fear: tracking flu infections on twitter surveillance sans frontieres: internet-based emerging infectious disease intelligence and the healthmap project healthmap: global infectious disease monitoring through automated classification and visualization of internet media reports acknowledgments. we thank the analytics team at uptodate for sharing the data used in this manuscript.disclaimer. the funder had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.financial support. financial support for this study was provided by the national library of medicine (grant number r01 lm010812-04).potential conflicts of interest. all authors: no potential conflicts of interest.all authors have submitted the icmje form for disclosure of potential conflicts of interest. conflicts that the editors consider relevant to the content of the manuscript have been disclosed. key: cord-252884-miptf6od authors: jeffery, diana d.; cohen, martin; brooks, arnold; linton, andrea; gromadzki, richard; hunter, christine title: impact of the 2009 influenza (h1n1) pandemic on the united states military health care system date: 2013-06-17 journal: mil med doi: 10.7205/milmed-d-12-00345 sha: doc_id: 252884 cord_uid: miptf6od background: during public health emergencies, the military health system experiences challenges similar to those across the u.s. public and private health systems. this study explored how 1 such event, the 2009/2010 influenza (h1n1) pandemic, impacted health care utilization and associated costs in the military health system. methods: data from the military data repository were used to examine diagnoses, claims data, and dates of services with respect to military or civilian care during 2004–2009/2010 influenza seasons. comparison analysis was conducted through two-tailed t-tests and regression models. results: there was a significant increase in inpatient and outpatient health care utilization during the 2009/2010 h1n1 pandemic year, most markedly for emergency department visits. the 2009/2010 h1n1 pandemic cost the department of defense $100 million compared to influenza-related health care costs incurred in previous influenza seasons. highest health care utilization costs were found in children less than age 5. the greatest cost burden was attributed to immunizations for active duty personnel delivered at military facilities. conclusion: annual trend analysis of costs and health care utilization would be helpful to plan and resource emerging influenza pandemics and to identify subgroups at greatest risk for contracting influenza. public health emergencies have long been an area of concern for military health care systems because of the increased demand for health care services. over the last decade, there have been several instances where outbreaks of infectious disease, as public health emergencies, have repeatedly challenged military policies for preparedness. in addition to localized outbreaks, international pandemics have become a focal concern because of their impact on available resources in both civilian and military health care systems. since the 1918 influenza, the united states has helped mitigate the spread of disease during pandemics, but recent global epidemics have shifted the focus to strengthened preparedness for such public health emergencies. 1 the growing number of emerging and re-emerging infectious diseases, like the severe acute respiratory syndromeassociated coronavirus in 2003 and the influenza a virus subtype h1n1 pandemic in 2009, have highlighted the rapid progression of localized epidemics into global pandemics. 2 available treatment options with novel pathogens are typically limited during such global public health emergencies, thus necessitating greater reliance on emergency outpatient or inpatient supportive care. from a health systems management perspective, the large influx of patients being treated for either an unknown virulent etiology or supportive therapy can easily overburden health care facilities. the u.s. department of defense (dod) has long recognized potential threats posed by global pandemics to the readiness of combat forces, and the health and well-being of the nation they protect. the military health system (mhs), which maintains health care services for u.s. armed services personnel and their families, has extensive experience with influenza disease epidemics, such as the 1976 outbreak of h1n1 that was detected after the death of an army recruit in fort dix, new jersey. 3 this outbreak resulted in a mass vaccination program across the united states and initiated a collaborative approach between champus (now tricare, the system of health care plans for the u.s. armed forces and beneficiaries), the centers for disease control and prevention, and local/ regional health departments. twenty years after the fort dix incident, an outbreak of influenza a (h3n2) on a navy ship led to the discovery of an antigenically distinct h3n2 5 other types of communicable diseases have similarly challenged military preparedness policies such as outbreaks of pertussis, 6 occupational exposures to blood-borne pathogens or enteroviruses, 7 and food-borne infectious diarrhea. 8 federal experiences identifying and controlling new strains of influenza are critical for military preparedness and the health care of dod beneficiaries. 9 as part of national pandemic preparedness, dod policy now authorizes individual military base commanders to plan and coordinate response activities with local public health officials based on local declaration of public health emergencies. 10 dod also contributes to national pandemic control by assisting in diagnostic testing of laboratory samples and reporting of influenza prevalence data through the dod electronic surveillance system for the early notification of community-based epidemics, published online at the force health protection pandemic influenza watch board. 10, 11 during the 2009 h1n1 pandemic, the world health organization reported 477,593 laboratory positive cases and 17,919 fatal cases from 214 countries. 12 in the united states, the centers for disease control and prevention laboratory confirmed 2,117 fatal cases and 41,821 reported hospitalizations. 13 although laboratory-confirmed cases provide real-time case counts during a public health emergency, estimates are likely to be greater because of underreporting. the dod implemented a preparedness policy during this 2009 pandemic, including a stringent vaccination program and guidelines to providing health care during public health emergencies. antiviral therapy and pandemic (h1n1) 2009 vaccination for active duty (ad) service members, including high risk beneficiaries, became a priority and this then served as the primary approach against this pandemic. 10 tricare beneficiaries received the h1n1 vaccine under another program, one with the same vaccine accessibility provided to civilian populations but with prioritization for vulnerable beneficiary populations. 14 with the implementation of the vaccination program during the peak of the pandemic, military treatment facilities (mtfs) had to cope with the influx of ad personnel and beneficiaries who were seen in emergency departments, inpatient wards, or outpatient clinics during the pandemic. here, we report on the results of a retrospective study to assess the impact of the 2009 h1n1 pandemic on the mhs with the overall goal of assessing and improving the effectiveness of pandemic preparedness in the dod. this study exam-ines the burden of disease caused by the 2009 h1n1 pandemic strain on the mhs with respect to health care utilization and associated costs and provides insight into dod resource management of pandemics compared to previous nonpandemic influenza seasons. we used the mhs data repository (mdr) to assess health care utilization, prescription claims, and estimated costs of h1n1 and influenza-like illnesses (ilis). the mdr captured immunizations rendered to patients in a military health care setting only. influenza immunization rendered in civilian settings were not tracked as non-ad beneficiaries may obtain immunization through multiple sources that may not be entered as mdr claims data (e.g., employment, free public health programs). our analytical dataset consisted of outpatient encounters, inpatient stays, and medication records rendered through mtfs and civilian care, october 2004 through april 2010. we included encounter records if they had specific and nonspecific codes associated with ili, influenza immunization, and associated treatments. individuals with multiple influenza-related encounters were only counted once per season. each record contained the month/year of the service encounter, the patient pseudoidentifier, sex, age, beneficiary category, branch of service, state of residence, and type of tricare enrollment plan (mtf prime, managed care support contractor prime, and non-prime). the final dataset was aggregated to person-level records to enable characterization of the 2009 h1n1 influenza patient. age was grouped by 0-4, 5-17, 18-24, 25-34, 35-44, 45-64, and 65 years or older. beneficiary categories were ad, family members of ad, military retirees and family members of retirees age 65, and military retirees and family members age 65 or older. cost data included costs associated with outpatient and emergency department visits, inpatient stays, influenza vaccine administration, and antiviral prescriptions rendered in the military and civilian settings from october 2004 to january 2010 (the last month when full cost data were available at the time of analysis). costs also included $15.9 million spent by dod to purchase h1n1 vaccine although this expenditure could not be allocated to the patient level. for each ili service delivered from october 2004 through january 2010, we used the full cost associated with services delivered in mtfs (direct care); for services delivered by civilian providers (care purchased in the tricare network or from non-network providers), we tabulated the actual amounts paid by dod using claim-level data. the study was reviewed by the tricare management activities human research protection program oversight office and determined to be exempt from full review by the dod institutional review board. demographics (number of cases by age group, gender, beneficiary category, branch of service, and tricare enrollment plan) for 2008/2009 influenza season and 2009/2010 pandemic season were examined using 2-tailed correlation analyses to identify differences between the two seasons. cost data for the 2009/2010 h1n1 pandemic was derived by comparing the total ili-related costs for the 2009/2010 influenza season to the ilirelated costs associated with the five preceding seasons. the expected costs were then compared with the actual costs for the 2009/2010 season to isolate the incremental cost of h1n1. the ili-related costs excluded costs associated with services rendered outside a health care facility; costs borne by dhhs for h1n1 immunizations administered in civilian facilities; costs associated with lost productivity because of ili; and costs of h1n1 surveillance, reporting, and response activities. a series of linear regression models were then used to estimate the expected ili costs during the 2009/2010 influenza season as if it had been a typical influenza season, and to ferret out those factors that contributed to costs. eight regression models were specified, with separate models for military (direct) care and civilian (purchased) care as well as for four types of services: inpatient, outpatient nonpharmacy, outpatient pharmacy, and immunizations. the models controlled for the number of beneficiaries, age mix, beneficiary category, seasonality, and the underlying time trend. the time trend was modeled based on an explicit, independent variable that increased by one unit per month from october 2004 (month 1) through january 2010 (month 64). the model implicitly assumed that the underlying historical trend, controlling for the other independent variables, would continue. this underlying historical trend accounted for medical inflation and underlying trends in utilization per capita. the total and mean per-patient number of outpatient visits, emergency department visits, inpatient stays, and antiviral prescription fills by each type of ili service are shown in table i . relative to the preceding influenza season, utilization of 2009-2010 health care services for ili increased at both military and civilian-contracted facilities, with the largest increases observed for emergency department visits (854.5% in military and 524.3% in civilian facilities). ili office visits increased 532.1% and 324.6% at military and civilian facilities, respectively, whereas inpatient dispositions increased 513.0% and 284.2% at military and civilian facilities, respectively. in both tables ii and iii show the hi1n1 cost impact by age group and type/source of care, respectively. for the estimated $84 million h1n1 impact that could be allocated to the patient level (table ii) , 61% of the cost increase was attributable to care rendered to beneficiaries aged 0 to 24 years, whereas 72% of the cost increase was due to care rendered to ad service members and their family members. table iii indicates where the total of $156.7m was spent on ili from july 2009 to january 2010. this total was approximately $100m more than projected for these months during a normal influenza season. table iii includes $15.9m for dod-purchased h1n1 vaccine that could not be allocated in the patient-level cost data. treatment costs accounted for 58% of the $100m overall cost impact, whereas immunization-related costs, including the dod purchase of vaccine, accounted for 42%. the regression models showed that there was a high cost burden for civilian inpatient hospitalization and civilian outpatient services across the period of analysis. the highest health care utilization costs were found for children ages 5 to 17. the overall greatest cost burden, however, was incurred for immunizations delivered at mtfs. the estimated costs from the regression models and the actual costs for each influenza season by month between october 2004 and january 2010 are presented in figure 1 . in 2009/2010, the mhs realized an estimated $100 million increase in ili-related costs compared to prior influenza seasons. the dod experience with the 2009/2010 h1n1 pandemic is remarkably similar to experiences found in the nation as a whole, at least among populations with access to health care. our findings closely mirror those based on u.s. civilian populations, i.e., heavy reliance on emergency departments 15 and high rates of ili among beneficiaries less than age 18. 16 the increased rates of health care utilization and costs observed in the dod pediatric population is particularly concerning, given a new report that concludes that the nation's capacity to care for pediatric populations during the 2009/2010 influenza season was marginally adequate; had the 2009/2010 h1n1 pandemic been more virulent, tertiary pediatric facilities would have been overwhelmed. 17 further, in contrast to our findings, the u.s. civilian population saw a heavier burden of ili among young adults. 18 most young adults included in our dod population were ad personnel, who, unlike their civilian counterparts, were required to be vaccinated for h1n1. based on our findings and those based on civilian reports, 19 we recommend that dod, as well as the local communities where mtfs reside, develops specialized preparedness procedures for pediatric populations in anticipation of future influenza pandemics or epidemics. one strategy, based on estimated models when immunization resources are scarce and the influenza incidence is declining, is to consider immunizing school-age children as well as high-risk adults. 20 although dod has clear requirements for vaccinating ad personnel, there are no such directives for vaccinating family members. another strategy is to consider closing elementary schools on dod installations, the length of closure dependent on pandemic severity as modeled elsewhere using civilian schools. 21 in addition, to assist fiscal and resource planning, we recommend time trends that compare rates and costs to previous influenza seasons be conducted annually for each mtf. finally, we recommend that future study be directed to evaluating effective models of triaging individuals who present with ili symptoms, particularly for those presenting in emergency departments. our study has several limitations. our inclusion of nonspecific vaccine administration codes in the mdr may have captured vaccines unrelated to influenza. only vaccines administered in the health care setting are captured in the mdr, but some immunizations are administered to service members in the field; thus, total vaccine-related costs for dod are underestimated in our study. cost allocation data, particularly for care rendered in the military hospitals, were not specific to immunization clinics or different types of immunizations. immunizations or treatment for ili may have occurred as part of a broader health care visit, and ili costs may not have been identified. moreover, our cost estimation methodology may not have generated an accurate estimate for all events because we only had cost data through january 2010. to some extent, comparison with prior influenza seasons mitigates the impact of this limitation, assuming that health care practices persisted across the 2004-2009 influenza seasons. yet another limitation, out of concern for patient privacy, is that we did not conduct analysis at the mtf level. therefore, we neither examined the prevalence of h1n1 cases, health care utilization, or costs by individual mtfs nor consider mtf facility size, the number of available beds, and size of outpatient clinics in relation to the number of cases received at a particular facility. finally, while the cost of care provided in civilian settings is directly calculated from government payments/costs at a claim level, military care costs are reported in aggregate (by clinic) and then fitted and allocated to the detail "visit" level; hence, military care costs are inexact comparisons to civilian costs. the focus of this study was to examine the impact of the 2009 h1n1 pandemic on the mhs and inform military commanders and policy makers about the burden of a public health emergency on health care systems used by dod beneficiaries. this information should assist in review of dod's response to this most recent pandemic and help strategize for preparedness response to similar events. in particular, review and evaluation of strategies that address our capacity to care for the dod pediatric population during an influenza pandemic is warranted, as is the need to manage increased demand on emergency departments during such events. pandemic influenza preparedness and community resiliency a novel coronavirus associated with severe acute respiratory syndrome extent of spread and duration of the outbreak outbreak of influenza in highly vaccinated crew of u.s. navy ship the national strategy for pandemic influenza, implementation plan (nspiip) armed forces health surveillance center (afhsc): pertussis diagnoses among service members and other beneficiaries of the u.s. military health system armed forces health surveillance center (afhsc): viral meningitis, active and reserve components diarrhea outbreak during u.s. military training in el salvador effectiveness of seasonal influenza vaccines against influenza-associated illnesses among us military personnel in 2010-11: a case-control approach department of defense instruction. number 6200.03. public health emergency management within the department of defense available at pandemic influenza watchboard pandemic (h1n1) 2009-update 98 h1n1 flu u.s. situation update defense officials prepare for h1n1 flu syndromic surveillance during pandemic (h1n1) 2009 outbreak hospitalized patients with 2009 h1n1 influenza in the united states inpatient capacity at children's hospitals during pandemic (h1n1) 2009 outbreak, united states estimates of the prevalence of pandemic (h1n1) 2009, united states an hhs retrospective on the 2009 h1n1 influenza pandemic to advance all hazards preparedness available at rsif .royalsocietypublishing.org simulating school closure policies for cost effective pandemic decision making key: cord-348061-ssjl2w7l authors: chamberlain, samuel d; singh, inder; ariza, carlos a; daitch, amy l; philips, patrick b; dalziel, benjamin d title: real-time detection of covid-19 epicenters within the united states using a network of smart thermometers date: 2020-04-10 journal: nan doi: 10.1101/2020.04.06.20039909 sha: doc_id: 348061 cord_uid: ssjl2w7l containing outbreaks of infectious disease requires rapid identification of transmission hotspots, as the covid-19 pandemic demonstrates. focusing limited public health resources on transmission hotspots can contain spread, thus reducing morbidity and mortality, but rapid data on community-level disease dynamics is often unavailable. here, we demonstrate an approach to identify anomalously elevated levels of influenza-like illness (ili) in real-time, at the scale of us counties. leveraging data from a geospatial network of thermometers encompassing more than one million users across the us, we identify anomalies by generating accurate, county-specific forecasts of seasonal ili from a point prior to a potential outbreak and comparing real-time data to these expectations. anomalies are strongly correlated with covid-19 case counts and may provide an early-warning system to locate outbreak epicenters. epidemic dynamics of the coronavirus disease (covid-19) have varied widely across the globe, since the outbreak emerged from wuhan, china in december 2019 (1) . countries with proactive management and widespread testing have more effectively limited spread (i.e. 'flattening the curve'), whereas countries with limited response, such as the united states, have experienced steeper growth rates in cases (2) . syndromic monitoring systems-collecting information about symptoms often prior to, or instead of, interaction with the formal healthcare system-have played a pivotal role in successfully mitigating covid-19 spread in countries such as taiwan and south korea (3) . rapid syndromic detection also played a role in curtailing the [2014] [2015] [2016] ebola outbreak in west africa (4) , where speed of detection was important in reducing outbreak intensity (5) . networks of geolocated, user-generated physiological measurements hold the potential for improved tracking and prediction of outbreak epicenters (6, 7) . data from these networks are typically less specific, but more sensitive, than formal surveillance, and are often available more rapidly, because formal surveillance is constrained by testing speed (5) and/or time for record aggregation (8). data from syndromic monitoring networks can be physiologically grounded in disease processes (e.g. fever), in contrast with internet-based proxies, such as search engine logs, that are often used as a substitute for real-time syndromic data (9) . physiological data from connected devices, such as thermometer data, or elevated heart-rate data (10) , are thus a potentially valuable resource for outbreak response, and, at present, may be able to detect covid-19 symptoms in the absence of widespread and rapid testing in the us. . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint here, we outline a method to identify illness incidence anomalies using a geospatial network of smart thermometers, where county-scale anomalies are flagged in real-time. we then use the method to identify covid-19 outbreaks in the continental united states. our anomaly detection method follows three core steps: 1) generate county-specific forecasts of influenza-like illness (ili) from a time point prior to a potential outbreak, 2) compare real-time thermometer-derived ili to forecast expectations when new data is aggregated daily, and 3) flag anomalous ili values by evaluating the probability that the current signal is driven by regular seasonal influenza. in this case, we flag ili values that exceed the 97.5% percentile of expected influenza trajectories. we track real-time ili and create ili forecasts using data collected from a network of smartphone-connected personal thermometers managed by kinsa, inc. we track ili in the network by classifying users with elevated temperature readings over multiple days as having influenza-like illness (see methods). this sensor network records the temperature and approximate geo-location when a user takes their temperature (e.g. during an illness episode). these recordings store temperature readings, via kinsa's smartphone app, and locations are logged via gps location or ip address. readings are aggregated to county-scale ili and anonymized. the temperature readings are used to construct an ili signal that is highly correlated to center for disease control and prevention (cdc) ili nationally (r > 0.95) and across cdc regions (r range 0.70-0.94), and these signals have been demonstrated to improve regional ili surveillance and forecasting (6, 7) . we construct the ili signal at the county-scale, allowing identification of anomalous ili incidence at the scale of individual cities. more information on constructing illness incidence from the thermometer network can be found in the supplemental methods. . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint 5 our forecast builds upon the findings that individual cities have predictable, distinct epidemic intensity patterns governed by both population size and climate (11, 12) . as population size and climate change relatively slowly from year-to-year, this allows the use of the county-specific, daily network data to learn the typical seasonal flu transmission patterns of a region. from this we construct local ili forecasts that form the basis for real-time anomaly detection. following (13), we estimate a county's daily reproductive number (r t ), the average number of secondary cases that each infected individual would infect if the conditions remained as they were, at time t by solving the following equation: where i t is kinsa-derived ili incidence, and w k is the infectivity profile for influenza, representing the probability that an individual who becomes infected on day t acquired the infection from an individual who became infected on day t-k. we approximate w k by a gamma distribution with a mean of 2.5 and variance of 0.7 days (14) . the summation term in equation (1) . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint 6 we compare the real-time ili to these influenza trajectories to identify localized anomalies, daily. we flag anomalous ili when the daily, real-time data exceeds the 97.5% percentile of the expected influenza forecast ensemble. we define anomaly incidence as the difference between the real-time thermometer ili and the 97.5% percentile drawn from the influenza forecast ensemble. to validate this method, we apply our approach to detect covid-19 anomalies by forecasting expected influenza starting on march 1 st , 2020, before widespread outbreaks were underway (2). we assess detection quality by comparing anomaly fever counts to state-and county-level covid-19 confirmed cases aggregated from a variety of government sources (15) . anomaly fever counts are calculated from cumulative anomaly incidence multiplied by the estimated user base of a region. the skill of the ili forecast in the anomaly detection system compares favorably with the carnegie mellon (cmu) epicast and stat models (16) at cdc region and national scales (fig. 1a ). to calculate these forecast error rates, we compare cmu forecasts to cdc ili, while comparing anomaly detection forecasts to thermometer-based ili, given that these models are trained on different data sources. in contrast to other ili forecasts, our models do not need to be initialized with now-cast predictions, (i.e. inferred initial conditions) as thermometer-based ili, upon which the model is trained, is available in real-time. additionally, our forecast errors do not appear to increase much past the 5-week forecast horizon, suggesting the ili forecasts in the anomaly detection system are stable at long-time horizons (fig. 1a) . forecast error rates display seasonality, where forecasting errors are higher in winter months during periods of increased illness incidence, and lower in the spring and summer (fig. s1 ). . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint our detected anomalies correlate strongly to positive covid-19 cases at both county (fig. 1b) and state scales (fig. 1c) , validating this method for the rapid detection of covid-related illness anomalies. specifically, we find significant correlations between total confirmed covid-19 cases and total fever anomalies at the county (fig. 1b: r = 0 .54, p < 0.0001) and state scales ( fig. 1c : r = 0.55, p < 0.0001). we restrict this analysis to counties with at least one confirmed covid-19 case, given that a lack of confirmed cases could be due to either true covid-19 patterns or an absence of testing and reporting. we do not observe anomalous fevers in three states where covid-19 cases are confirmed; minnesota, wisconsin, and south dakota. countylevel total anomalous fevers also correspond to the spatial distribution of major known outbreak centers, with hotspots in the seattle area, san francisco bay area, new york metro area, and florida (fig. 1d ). we provide an example for how this method is applied in real-time for brooklyn, ny, usa (fig. 1e ). here, we forecast expected influenza trends from march 1 st , 2020 before large-scale outbreaks were occurring in the new york metro area (2), and we observe a strong divergence from expected influenza trajectories (fig. 1e) . however, we also observe an inflection back toward declining ili shortly after social distancing efforts were enacted (fig. 1e) , beginning with school closures on march 16 th and a later 'stay-at-home' order on march 20 th , 2020. we observe even sharper declines in ili post-social distancing in santa clara, ca, where a 'shelterin-place' order was enacted on march 16 th , 2020 (fig. 1f) . . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint an . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint 9 our estimates of anomaly incidence are likely impacted by social distancing in these cases, as distancing should also reduce influenza transmission. we are thus likely underestimating anomaly incidence after social distancing is enacted, given that our model assumes typical seasonal influenza transmission patterns. we therefore explore the sensitivity of our method by reducing r t values used in the influenza forecast for all days after social distancing efforts were implemented. in brooklyn ny ( fig. 2a) , dropping future r t values by 25% reduces the trajectory of seasonal influenza, leading to an increased magnitude of anomaly incidence ( fig. 2a ). in santa clara county, without accounting for social distancing we observe only a short period of detected anomalies, where real-time ili falls below the expected range of seasonal influenza (fig. 1f) . here, reducing r t following the 'shelter-in-place' order leads to additional anomaly detections post-social distancing (fig. 2b) . similarly, following school closures in miami-dade county, real-time ili values fell below anomalous levels by late march (fig. s2) , though accounting for reduced influenza transmission leads to more anomaly detections (fig. 2c ). in these case examples, we assumed r reduced by 25% given social distancing directives. while the true value cannot be estimated at present, it likely varies with both policy implementation and adherence. these findings suggest social distancing plays an important role in the quality of anomaly detection, and future research should address the impact of social distancing on influenza transmission rates. . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint of . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint 11 working in synergy with the formal healthcare system, high-throughput signals from distributed syndromic monitoring networks, such as we describe here, could play a decisive role in managing local outbreaks, and alter the trajectories of pandemics. these networks may be of particular use for situations where rapid testing and isolation are key to pandemic control (3, 4) , such as the current covid-19 outbreak, where the absence of widespread and rapid testing have led to a massive growth in cases worldwide. . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint covid-19) -situation report -51 response to covid-19 in taiwan: big data analytics, new technology, and proactive testing ebola surveillance -guinea the role of rapid diagnositics in managing ebola epidemics a smartphone-driven thermometer application for real-time population-and individual-level influenza surveillance improving state-level influenza surveillance by incorporating real-time smartphone-connected thermometer readings across different geographic domains accurate estimation of influenza epidemics using google search data and argo harnessing wearable device data to improve state-level real-time surveillance of influenza-like illness in the usa: a population-based study. the lancet digital health urbanization and humidity shape the intensity of influenza epidemics in us cities human mobility patterns predict divergent epidemic dynamics among cities a new framework and software to estimate time-varying reproductive numbers during epidemics time lines of infection and disease in human influenza: a review of volunteer challenge studies author/funder, who has granted medrxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) the copyright holder for this preprint the anomaly detection and forecasting code can be accessed on github, and all examples here are reproducible (https://github.com/kinsahealth/therm_anomaly_detection). we thank c. jessica metcalf for her input on an initial draft of this manuscript. author contributions sdc conceived and implemented the anomaly detection approach with feedback from bdd and pp. is conceived of and designed the kinsa products, including its use as syndromic early warning and monitoring systems. pp, ca, ad, and sdc designed and managed the real-time illness signal and geospatial platform. sdc wrote the manuscript with feedback from all authors.competing interests sdc, is, pp, ad and ca are employees of and shareholders in kinsa, inc.is conceived of and designed kinsa products to track the spread of infectious disease. bdd has no competing financial interests. all authors have completed the icmje uniform disclosure form at www.icmje.org/coi_disclosure.pdf . cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity.is the (which was not peer-reviewed) the copyright holder for this preprint this work is not a clinical trial and is instead a population-level observational study, where all user information is anonymized. more specifically, this study uses aggregated temperature readings to estimate the prevalence of influenza-like illness at the scale of united states counties.raw measures are temperature, time of reading, and approximate geolocation from smart thermometer readings, where user information is anonymized and aggregated to whole population influenza-like illness at the scale of us counties. it therefore does not use individual, private or personally identifiable information. given the nature of this data, it does not require irb review.. cc-by-nc 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity.is the (which was not peer-reviewed) the copyright holder for this preprint . https://doi.org/10.1101/2020.04.06.20039909 doi: medrxiv preprint key: cord-310956-qwe4ndvb authors: qian, yan‐hua; su, jing; shi, ping; he, en‐qi; shao, jie; sun, na; zu, rong‐qiang; yu, rong‐bin title: attempted early detection of influenza a (h1n1) pandemic with surveillance data of influenza‐like illness and unexplained pneumonia date: 2011-04-18 journal: influenza other respir viruses doi: 10.1111/j.1750-2659.2011.00248.x sha: doc_id: 310956 cord_uid: qwe4ndvb please cite this paper as: qian et al. (2011) attempted early detection of influenza a (h1n1) pandemic with surveillance data of influenza‐like illness and unexplained pneumonia. influenza and other respiratory viruses 5(6), e479–e486. background to collect disease information and provide data for early detection of epidemics, two surveillance systems were established for influenza‐like illness (ili) and unexplained pneumonia (up) in wuxi, people’s republic of china. objectives the current study aims to describe the performance of these surveillance systems during 2004–2009 and to evaluate the value of surveillance data in detection of influenza epidemics. methods two national ili sentinel hospitals and three up sentinel hospitals provided data to the surveillance systems. the surveillance data from hospital‐based outpatient clinics and emergency rooms were compared by year. the ili data of 2009 were further modeled based on previous data using both a control chart method and a moving average regression method. alarms of potential epidemics would be raised when the input surveillance data surpassed a threshold. results in 2009, the proportions of ili and respiratory illness with fever (one surveillance syndrome of the up system) to total patient visits (3·40% and 11·76%, respectively) were higher than the previous years. the surveillance data of both systems also showed developing trends similar to the influenza a (h1n1) pandemic in 2009. when the surveillance data of 2009 were fitted in the two detection models, alarms were produced on the occurrence of the first local case of influenza a (h1n1), outbreaks in schools and in general populations. conclusions the results indicated the potential for using ili and up surveillance data as syndromic indicators to detect and provide an early warning for influenza epidemics. influenza, commonly known as the flu, is a severe infectious disease accounting for between 250 000 and 500 000 deaths worldwide every year. millions of people died in the influenza pandemic of 1918, one of the worst epidemic disasters in recorded human history. 1 the life style of modern society, including overcrowded cities and rapid global transportation, has resulted in dissemination of influenza and generation of novel virus strains. in the past year, the world health organization (who) announced a level-6 alert for the influenza a (h1n1) pandemic, the highest level of disease outbreak, because of the sustained spread of this new virus on a global scale. 2 the world was fortunate not to have experienced another devastating influenza epidemic, because the fatality rate of the new virus strain was similar to seasonal influenza virus. 3 however, there remains an inevitability of a new influenza pandemic, which could be highly infectious and as deadly as the 1918 influenza pandemic. thus, it is necessary, although challenging to detect the onset of a new influenza epidemic, to allow the government to make timely and proper interventions to reduce the social and economic impact from influenza epidemics. the influenza surveillance system long has been playing an essential role in collecting disease information and providing evidence for prevention and control strategies and measures. two surveillance systems were established in wuxi for influenza-like illness (ili) and unexplained pneumonia (up) after the severe acute respiratory syndrome (sars) outbreak. the ili system was implemented for influenza surveillance, and the up system was designed to track lower respiratory illness with pneumonia symptoms. although the disease surveillance systems appeared decades later than in developed countries, their performance is by no means inferior to define the distribution of circulating strains in the community and detect disease aberration. here we described and compared the accumulated ili data from 2004 to 2009 and up data from 2007 to 2009. to further evaluate the effectiveness of these surveillance systems in early warning of influenza epidemics, we monitored ili data between 2004 and 2008 by both a control chart method and the serfling method and tested goodness of fit using influenza a (h1n1) data of 2009. wuxi is a prefecture-level city of jiangsu province, located in eastern china, with a population of 6ae2 million. wuxi has seven districts and two county-level cities (jiangyin and yixing). wuxi has a typical subtropical monsoon climate and seasonal influenza usually peaks once a year. 4 there are two surveillance systems in wuxi. one is ili surveillance system and the other is up surveillance systems. both systems collected data from sentinel hospital-based outpatient clinics and emergency rooms. all the selected sentinel hospitals are general hospitals and available to every resident in wuxi city including jiangyin and yixing county. the participating doctors in referral sentinel hospitals were provided with detailed diagnosis instructions for sample selection. as required by the chinese center for disease control and prevention (cdc), two hospitals (children's hospital and the no. 3 people's hospital of wuxi) were selected as the national sentinel hospitals for ili surveillance. the number of patient visits ranged from 8000 to 12 000 ⁄ week. ili was defined as follows: fever ‡38°c, either cough or sore throat, and lack of evidence of other laboratory-confirmed diagnosis. pharyngeal swab specimens collected from patients with ili were sent to cdc laboratories for subsequent isolation and identification. the proportion of patient visits for ili was calculated as the number of patients with ili divided by the total number of patients and was reported through the influenza surveillance system of china. the up surveillance was performed in three hospitals (the no. 2 people's hospital of wuxi, the people's hospital of jiangyin county, and the people's hospital of yixing county). this system was built with special attention to lower respiratory illness, especially pneumonia caused by unknown reagent like sars. routinely collected data included respiratory illness with fever (riwf, also known as acute respiratory illness) and classified pneumonia. emergency measures would be taken in case of any up. pneumonia was diagnosed according to international classification of diseases (10th version, j00-j99). unexplained pneumonia was defined as follows: fever ‡38°c, pneumonia-like characteristics in diagnostic imaging, decreased white blood cell count or decreased lymphocyte differential count in early clinical stage, and no improvement in or even worsening of patient's condition after regular antibiotic treatment for 3-5 days. the total number of patient visits ranged from 11 000 to 14 000 ⁄ week, among whom an average of 3ae5-3ae7% were diagnosed as classified pneumonia. weekly numbers of riwf and classified pneumonia were reported to wuxi center for disease prevention and control. we retrospectively studied ili data from 2004 to 2009 and up data from 2007 to 2009, respectively. all the data were organized by microsoft office excel 2007. statistical analysis was performed by using spss software (version 17.0, spss inc., chicago, il, usa). the kolmogorov-smirnov test was used to check the normality of the data. the chisquare test was used for the comparison of different ratios and proportions. the ili data during 2004-2008 were used in statistical modeling for influenza detection. both a control chart method 5-7 and a moving average regression method 8, 9 were applied to produce the threshold value. alarms of potential epidemics were raised when the input surveillance data of 2009 surpassed the threshold. the control chart method was originally developed to determine and control the situation for manufacturing processes and was later implicated in surveillance analysis. 10 in this study, a long-term average weekly proportion of ili (p) and its standard deviation (sp) were calibrated using all the surveillance data. the data higher than p + 2sp were referred to as being out of statistical control and were excluded. such data were usually observed in epidemics of seasonal flu and could not generally represent a stable and predictable situation over time. adjusted average weekly proportion of ili (p') and standard deviation (sp') were calculated with those data in the state of statistical control. the threshold was defined as p' + 2sp'. the moving average regression method was originally proposed by serfling and was used in our study to generate a time-varying threshold for influenza. 8, [11] [12] [13] [14] a linear secular trend was estimated from the surveillance data through 2004-2008 by using the regression analysis tool of excel. the secular trend was then removed from the original surveillance data, so the adjusted data represented only seasonal change and could be compared by year. the time-varying threshold was calibrated as the 5-year moving average plus 1ae6 sd (standard deviation) (p = 0ae05) or 2ae0 sd (p = 0ae01) from the mean. the first imported case of the influenza a (h1n1) was reported on june 25, 2009 (the 26th week of the year). the first local case was reported on august 16 (week 33). the first school case was reported on september 3 (week 36), and transmission spread quickly when students came back to school in september. the influenza epidemic spread widely during september to december in the general population, with 132 confirmed cases in september, 89 cases in october, 146 cases in november, and 113 cases in december. the last identified patient was reported on february 12, 2010. the epidemic timeline is shown in figure 1 . the surveillance data collected from the two sentinel hospitals showed an earlier peak of the proportion of ili in 2009 when compared with the same period in 2007 and 2008 ( figure 2 ). the ili proportion aberrantly increased to 3ae72% in the 22nd week of 2009, higher than 3ae61% in 2007 and 2ae74% in 2008. this increase in ili proportion appeared 4 weeks earlier before the first imported case of influenza a (h1n1) and started to decline after week 28. simultaneously with the appearance of the first local case, the proportion of ili started to increase rapidly beginning in week 34, indicating the early stage of the influenza epi-demic. the proportion of ili peaked in the general population at week 39 (4ae74%), when the schools also experienced a peak of illness. figure 3 ). the proportion of riwf to total patient visits (11ae76%) was also significantly higher than that in the previous 2 years (9ae91% and 10ae14%, respectively), while the proportion of pneumonia to riwf was similar in all 3 years (3ae53-3ae69%) ( table 1 and figure 3 ). the proportion of riwf began to increase in week 32, around the time of the first imported influenza a (h1n1) case. when we examined the 2009 epidemic season, the trend of up surveillance data was similar to the trend of ili data (figures 2 and 3) . therefore, the data observed from up surveillance system may also reflect the epidemic situation of influenza. the basic design of outbreak detection is that an alert is generated when the current data surpass a threshold. therefore, determining a threshold is essential and crucial for the performance of an early warning system. to obtain the optimal sensitivity and specificity, both control chart method and moving average regression method were used to calibrate the threshold. the control chart method was used to determine an influenza threshold by week. a centerline threshold value was drawn based on the weekly proportion of ili from 2004 to 2008. when the 2009 data were fitted into the chart, alarms of aberrant higher proportions of illness appeared in week 18, week 20 to week 29, week 31 to week 41, week 45 to week 47, and week 49 (figure 4) . these results were consistent with the true observations, including the first imported case in week 26, the first local case in week 33, the outbreak in school beginning in week 36, and the outbreak peak in general population in week 45. in the moving average method, a time-varying threshold for influenza was generated. the ili counts for 2009 were higher than the moving average baseline in week 18, week 20 to week 21, week 24, and week 32 to week 50 ( figure 5 ). more precisely, the ili counts were higher than mean + 1ae6 sd in week 34 to week 37, week 39, and week 45 to week 47, and the periods from week 34 to week 37 and week 46 to week 47 showed the highest ili counts (above mean + 2ae0 sd). the aberration points were consistent with the reported date of the first local case, the outbreak in schools, and the peak in the general population. the early detection of disease outbreaks has long been important to public health. the importance of a surveillance network at the global level has been unprecedentedly emphasized because of the emergence of newly infectious diseases, pandemics, and the threats of bioterrorism. 15 the monitoring of surveillance data plays an essential role in detecting signals for a potential outbreak. for diseases or illnesses with enormous social impact, timeliness of detection is extremely valuable. with these types of diseases or illnesses, indicators like hospital consultations rather than mortality or morbidity are used in modeling, because they tend to provide earlier outbreak indication. 16, 17 for influenza, ili is commonly used as an indicator to predict the occurrence of epidemics. 13, 18, 19 unlike other diseases, the value of routinely collected ili data for influenza detection has been carefully studied and confirmed. 17, 18, 20 in this current study, weekly counts of consultations for ili and up were monitored for influenza detection. the distribution of sentinel ili data in 2009 ( figure 2) were consistent with the real spread of the pandemic and showed aberrant increase close to or earlier than the reference data. these results confirmed the capability of ili surveillance system to detect influenza. although the up surveillance system was designed to detect sars instead of influenza, the data are useful because ili and pneumonia usually have similar symptoms such as fever, cough, and breathing difficulty. compared with the reference data (figure 1 ), the sentinel up data (table 1 and figure 3 ) also showed similar epidemic trend. however, the efficiency of early detection was not high using the up surveillance system as significantly increased proportion of riwf was observed in sentinel hospitals only after the first local case was reported. a reasonable explanation might be because of the diagnostic criteria. ili was used specifically to indicate influenza, while riwf includes pneumonia and many other kinds of respiratory illnesses besides ili. the nature of the up system could not distinguish between the respiratory complaints related to influenza and those caused by other pathogens. the population at risk could be another reason for not being able to detect outbreaks because the ili sentinel sites and up sentinel sites cover different areas. nevertheless, the sentinel up data provide additional information to detect influenza epidemic and can add confidence to declare an epidemic alarm. at the beginning of the 2009 pandemic flu period, asia was in especially high alert of epidemic influenza and china implemented aggressive policies trying to delay illness spread by early identification and isolation of every confirmed h1n1 patient. the increased proportion of ili and riwf before the first imported h1n1 case could be explained as a surveillance artifact because of the heightened concern of individuals. with the development of the epidemic, control strategies were changed to intensive treatment of the seriously ill and self-isolation of mild cases. people who had mild flu symptoms were encouraged to take anti-viral medicines and stay at home instead of seeking medical care in hospitals. all the sentinel hospitals collected data as usual without any strengthened effort. the higher proportion of ili and riwf observed after week 29 was more likely due to the high incidence of h1n1 influenza in general population rather than surveillance artifact. a variety of statistical methods have been developed to monitor surveillance data, including time series, regression, cumulative sum (cusum), and hidden markov models. 9, 11, 14, [21] [22] [23] [24] the performance of these models is usually evaluated from the following characteristics: sensitivity, specificity, and timeliness. 21, 25 sensitivity is the ability to find a real outbreak alert; specificity is the ability to exclude false alarm; timeliness is the ability to raise an alarm within the shortest amount of time from the onset of the peak of the season. most approaches of early detection produce alerts based on the theory of signal detection. 25 the optimal sensitivity and specificity depend on a proper level of threshold, but there is always a trade-off between a high threshold for good specificity and a low threshold for good sensitivity. the calibration of a threshold is a complex determination for each system. both the control chart method and the serfling method are time series approaches and have their own advantages in reducing the variance caused by epidemic data or secular trend. 14, 21 the control chart method optimizes the threshold calibration by excluding the unexpected data that are beyond the controlled scope. 6 the serfling method provides reasonably accurate estimates by adjusting the threshold line with secular regression trend. 8 the us centers for disease control and prevention (atlanta, ga, usa) developed an early aberration reporting system (ears) based on cusum method, which can use only recent data and is supposed to incorporate the threshold calculation by regression method. 21, 22 however, a study in hong kong showed that the regression method and the time series method are superior to the cusum method in a small city with fewer sentinel sites, more variable data, and a more abrupt peak in activity. 14 as a result, the cusum method was not used in our study. the control chart method detected all four remarkable signals, namely the occurrence of the first imported case, the first local case, outbreaks in schools, and the peak in the general population ( figure 4) . it also produced two false-positive alarms, one in the beginning and the other at the end of the epidemic period. when using mean + 1-ae6 sd as the threshold, the serfling method detected the outbreaks in schools and the peak in the general population, with a false-positive signal in the interval; when raising the threshold to mean + 2ae0 sd, only the outbreaks in schools could be detected in a timely manner, and the second alarm was 1 week after the peak in the general population ( figure 5 ). the comparison of the two methods indicated that the control chart method had better sensitivity and timeliness and the serfling method had better specificity. nevertheless, considering that the false-positive alerts of the control method appeared outside the epidemic period, this fact might be taken in favor of a more sensitive detection. the same opinion was seen in gault's paper. 18 a final decision probably should be made combining the results of both methods. the control chart method can generate sensitive and timely alerts, and serfling method can offer secular trend to increase specificity and avoid wasting medical resources. in developing countries like china, surveillance systems like the ili and up systems have only recently been implemented. we performed this study after the 2009 pandemic flu in attempt to make full use of these routinely collected background data. the results are theoretically helpful for healthcare workers to take prevention and control measures when influenza epidemic occur, including preparing antiviral drugs and other medical supply needs, and sending health education messages. so far no such actions, which we already took during the epidemic period, have been applied based on our study. future studies will be carried out to improve the performance of these surveillance systems, such as adding more sentinel sites, covering larger population, improving the quality of indicators, and trying different detection algorithms. studies of decision making are also needed before these approaches could be used in practice. the decision theory provides mathematical methods to estimate the benefits of true alarms and the costs of false alarms for getting an optimal threshold. 25 this analysis was beyond the scope of our current study, and we will investigate such methods in our future work. in conclusion, this study described the surveillance data of ili and up during 2007-2009 and the epidemic progress of influenza a (h1n1) pandemic in wuxi, china. the surveillance data of both systems were correlated with the influenza epidemic and therefore valuable in monitoring influenza activity and generating early warning methods. the control chart method and serfling method had their respective merits and faults in data analysis. further studies could be considered to optimize the threshold of warning. the principle is to improve performance of surveillance systems and allow timely precautionary measures to be implemented in vulnerable populations. emerging infections: pandemic influenza world now at the start of 2009 influenza pandemic mortality from pandemic a ⁄ h1n1 2009 influenza in england: public health surveillance study guideline for prevention and control of influenza statistical quality control methods in infection control and hospital epidemiology, part ii: chart use, statistical properties, and research issues statistical quality control methods in infection control and hospital epidemiology, part i: introduction and basic theory study on early warning based on influenza surveillance data in guangdong province methods for current statistical analysis of excess pneumonia-influenza deaths time-series forecasting use and interpretation of statistical quality control charts influenza-associated hospitalizations in the united states impact of influenza vaccination on seasonal mortality in the us elderly population can syndromic thresholds provide early warning of national influenza outbreaks? methods for monitoring influenza surveillance data trends and directions of global public health surveillance the impact of influenza epidemics on hospitalizations value of icd-9-coded chief complaints for detection of epidemics performance of a syndromic system for influenza based on the activity of general practitioners, france situational uses of syndromic surveillance sensitivity, specificity and predictive values of health service based indicators for the surveillance of influenza a epidemics comparing aberration detection methods with simulated data a simulation model for assessing aberration detection methods used in public health surveillance for systems with limited baselines review of an influenza surveillance system, beijing, people's republic of china disease surveillance using a hidden markov model the emerging science of very early detection of disease outbreaks key: cord-002753-lvlbwcl0 authors: rogers, kimberly b.; roohi, shahrokh; uyeki, timothy m.; montgomery, david; parker, jayme; fowler, nisha h.; xu, xiyan; ingram, deandra j.; fearey, donna; williams, steve m.; tarling, grant; brown, clive m.; cohen, nicole j. title: laboratory-based respiratory virus surveillance pilot project on select cruise ships in alaska, 2013–15() date: 2017-10-06 journal: journal of travel medicine doi: 10.1093/jtm/tax069 sha: doc_id: 2753 cord_uid: lvlbwcl0 background: influenza outbreaks can occur among passengers and crews during the alaska summertime cruise season. ill travellers represent a potential source for introduction of novel or antigenically drifted influenza virus strains to the united states. from may to september 2013–2015, the alaska division of public health, the centers for disease control and prevention (cdc), and two cruise lines implemented a laboratory-based public health surveillance project to detect influenza and other respiratory viruses among ill crew members and passengers on select cruise ships in alaska. methods: cruise ship medical staff collected 2–3 nasopharyngeal swab specimens per week from passengers and crew members presenting to the ship infirmary with acute respiratory illness (ari). specimens were tested for respiratory viruses at the alaska state virology laboratory (asvl); a subset of specimens positive for influenza virus were sent to cdc for further antigenic characterization. results: of 410 nasopharyngeal specimens, 83% tested positive for at least one respiratory virus; 71% tested positive for influenza a or b virus. antigenic characterization of pilot project specimens identified strains matching predominant circulating seasonal influenza virus strains, which were included in the northern or southern hemisphere influenza vaccines during those years. results were relatively consistent across age groups, recent travel history, and influenza vaccination status. onset dates of illness relative to date of boarding differed between northbound (occurring later in the voyage) and southbound (occurring within the first days of the voyage) cruises. conclusions: the high yield of positive results indicated that influenza was common among passengers and crews sampled with ari. this finding reinforces the need to bolster influenza prevention and control activities on cruise ships. laboratory-based influenza surveillance on cruise ships may augment inland influenza surveillance and inform control activities. however, these benefits should be weighed against the costs and operational limitations of instituting laboratory-based surveillance programs on ships. approximately 1 million cruise ship passengers visit the us state of alaska each summer. the inbound, summertime tourist population is 1.5 times alaska's wintertime resident population. 1 some cruise ships in the region carry upward of 2500 passengers and 1000 crew members originating from many countries. these factors, along with close quarters and prolonged contact among travellers on ships and during land-based tours before embarkation, increase the risk of communicable disease transmission. influenza outbreaks among cruise ship passengers and crew members are relatively common, and may be prolonged and challenging to control as new, susceptible passengers embark at frequent intervals. 2, 3 cruise ships destined for us ports of entry are required by federal regulations to report to the centers for disease control and prevention (cdc) all deaths on board, and certain signs and symptoms suggestive of infectious disease in passengers and crew members. 4 cruise ships are required to report gastrointestinal illness to cdc's vessel sanitation program at least 24 h before a ship's arrival at a us port. 5 in contrast, cruise ships are not required to conduct respiratory illness surveillance. however, cdc's division of global migration and quarantine asks cruise ships to report outbreaks of influenza-like illness (ili) and to complete end-of-voyage cumulative ili reports. ships voluntarily report the total number of crew members and passengers with ili either electronically or by phone to the cdc quarantine station with jurisdiction over the port of call. if a ship's ili incidence rate exceeds the cdc-defined outbreak threshold of 1.380 cases per 1000 traveller days, 6 cdc requests enhanced data collection and, in coordination with the state health department, provides consultation on influenza testing. if influenza viruses are detected in crew members or passengers, recommendations are made regarding control measures, including antiviral treatment and chemoprophylaxis, isolation of ill individuals, monitoring of exposed individuals, and vaccination of crew members. during low influenza activity periods such as summer, influenza outbreaks in the usa and canada have been linked to influenza virus transmission among tourists travelling on combined land-sea tours or cruise ship voyages, particularly in alaska. 3, 7 cruise ship passengers who are at increased risk for more severe disease, such as the elderly or immunocompromised, may develop severe complications from influenza such as pneumonia, which may result in disembarkation for hospitalization. the value of cruise ship influenza surveillance is highlighted by outbreaks of seasonal and pandemic influenza among passengers and crews, 2,7-9 and the potential for introduction of antigenically drifted seasonal influenza virus strains. 3 for example, in 1997, an antigenically drifted h3n2 virus strain identified in ill australian cruise ship passengers became the predominant virus during the subsequent 1997-1998 influenza season in north america. 10 cruise ship ili reporting to cdc is aggregate and does not include a laboratory component. integration of syndromic and virologic surveillance could improve the ability to detect and characterize influenza virus strains and other respiratory viruses circulating among cruise ship travellers. from may through september, 2013-2015, the alaska department of health and social services, the alaska state virology laboratory (asvl), cdc influenza division, cdc anchorage quarantine station and two cruise lines initiated a pilot laboratory-based surveillance project to detect influenza and other respiratory viruses among ill crew members and passengers on a few cruise ships in alaska. the pilot project aimed to characterize respiratory viruses circulating among cruise ship travellers in the state, to correlate the timing of cruise ship ili incidence peaks with identification of respiratory viruses in alaska, and to compare influenza viruses identified in ill passengers and crew members on ships with influenza viruses identified at land-based influenza surveillance sites. specimens were collected from passengers and crew members reporting to the ship infirmary on participating cruise ships in alaska from may-september, 2013-2015, and tested for respiratory viruses. four ships from two cruise lines participated in 2013 and 2014, and two additional ships from the same cruise lines were added for a total of six ships in 2015. the ships were relatively representative of those sailing in the region during the time period of the pilot project;~2000-2500 passengers, and 750-1000 crew were on board each ship. participating ships had voyage lengths varying from 7 to 14 days. throughout the season, asvl provided specimen-collection kits to cruise ships, including nasopharyngeal swabs, universal transport medium, laboratory submission slips and packaging for shipping specimens. the case definition for inclusion in year 1 (2013) was ili (temperature of ≥100°f and either cough or sore throat in the absence of a known cause other than influenza); in years 2 and 3 (2014 and 2015), the definition was expanded to include signs of an acute respiratory illness (ari) without fever, if the clinician suspected an infectious aetiology. ship staff obtained verbal consent from the individual (if an adult) or permission of a parent or guardian (for those under 18 years old). each participating ship was asked to collect 2-3 specimens per week during the course of the pilot project. cdc approved the surveillance project with a determination that it did not meet the definition of human subjects research under 45 code of federal regulations part 46. nasopharyngeal specimens were collected using flocked swabs (puritan ® ), placed in transport medium (puritan ® ), and refrigerated. specimens were kept on ice packs during their transit from the cruise ship to asvl. upon receipt at the laboratory, 200 μl of each specimen was extracted (biomerieux nuclisens ® easymag ® ) to produce a final elution of 60 μl of total nucleic acid. the genmark esensor ® 14-target respiratory virus panel was used per manufacturer guidelines to determine the presence or absence of influenza and other respiratory viruses. a random subset (~1 of every 4) of specimens testing positive for influenza viruses was shipped to the cdc influenza division laboratory for cell culture and antigenic characterization by using hemagglutination inhibition (hi) methods. 11 for viruses that did not yield sufficient titre for hi testing, genetic characterization (sequencing) was conducted to determine the influenza type, subtype or genetic group. 12 in addition to personal identifiers collected on the asvl requisition form per standard protocol for clinical specimens, ship medical staff provided asvl demographic, clinical and epidemiological information for each specimen. this information included whether the individual was a crew member or passenger, age, sex, country of residence, recent travel history outside the usa or canada, date and port of embarkation, influenza vaccination history (self-report), illness onset date and symptoms, receipt of antiviral treatment prior to specimen collection, and whether other travelers on board had respiratory illness. asvl coded and de-identified the data and laboratory results and shared that information with the cdc anchorage quarantine station. each month throughout the project period, the quarantine station compiled and issued reports to each cruise line. descriptive analyses were conducted on laboratory, demographic and epidemiological data from the 3-year pilot. while symptom history was analyzed, results were unremarkable and are not reported. in order to compare the pilot data to existing alaska state influenza surveillance data, asvl provided cdc with inland state surveillance data representing the same time period of the pilot project (may-september, 2013-2015). of the 410 nasopharyngeal specimens tested from 2013, 2014 and 2015, 274 (67%) were from passengers and 136 (33%) from crew members with ili or ari. three hundred and forty (83%) specimens tested positive for at least one virus on the asvl target panel; 292 (71%) tested positive for influenza a or b (table 1) , representing 75% of sampled passengers and 64% of crew members (tables 1 and 2) . throughout the 3-year pilot period, 13 respiratory viruses were identified (table 3) . fifteen passengers and 4 crew members tested positive for 2 viruses; 9 different co-infections were identified from these 19 individuals. asvl sent a subset of samples (56, 19%) positive for influenza a or b viruses to cdc for further antigenic characterization; results are shown in table 4 . beyond this subset, 19 additional influenza a(h3n2) viruses were genetically characterized at cdc; these viruses did not yield a sufficient titre for the hi test and thus are not presented in table 4 . however, all 19 belonged to the same genetic group as the predominant a(h3n2) virus strains circulating during the 2014-2015 influenza season. of the 2479 samples collected by instate providers for alaska's statewide inland influenza surveillance program from may through september in 2013-2015, 368 (15%) tested positive for influenza a or b virus. while both the state inland surveillance program and the alaska cruise ship respiratory surveillance pilot project primarily identified influenza a(h3) among all individuals testing positive for influenza viruses, state inland surveillance results yielded a higher percentage of influenza b virus but fewer influenza a(h1n1)pdm09 viruses (table 5 ). participating passengers and crew members on all ships represented 31 countries. seventy-one (52%) of crew members submitting specimens during the pilot project period were permanent residents of the philippines or india, whereas most passengers were from the united states (216, or 79%). the median age of passengers submitting specimens was 66 years (range: 5-90); the median age of crew members submitting specimens was 33 years (range: 20-59). results were relatively consistent across age groups, although most (25, or 83%) influenza h1n1pdm09positive specimens throughout the pilot came from individuals younger than 40. six percent (25) of crew members and passengers submitting specimens had been outside the usa or canada in the days (10 days in 2013, 5 days in 2014 and 2015) before symptom onset; countries included australia (8), japan (9), india (2), taiwan (1), south africa (1), china (1), israel (1), singapore (1) and trinidad and tobago (1). aggregate virologic results from these individuals did not deviate markedly from those with no recent travel outside the us or canada. self-reported influenza vaccination coverage for passengers and crews varied greatly between participating ships and voyages. more than half of the passengers (158, or 58%) submitting specimens reported having been vaccinated for influenza during the previous year. in contrast, fewer than half (61, or 45%) of the crew members reported the same. one of the cruise lines reported that of those crew vaccinated, all had received the northern hemisphere influenza vaccine, regardless of ship's origin or the crew member's origin. aggregate data for all 3 years showed that positive test results for influenza a or b were similar for passengers and crew members who reported recent influenza vaccination (73 and 64% respectively) and those who did not (77 and 64% respectively). no participants received antivirals before specimen collection. analysis of the timing of participants' illness onset relative to embarkation revealed a peak for northbound ships (originating in vancouver, canada or seattle, usa, en route to alaska) on day 3 of the voyage (range: 1 day pre-embarkation to 5 days post-embarkation) and for southbound ships (which originate in alaska and sail toward the us pacific northwest) on day 1 of the voyage (range: 3 days pre-embarkation to 5 days post-embarkation) (figure 1 ). although onset dates differed from north vs. southbound cruises, no marked differences in influenza subtypes between individuals on north vs. southbound cruises were identified. the alaska cruise ship respiratory surveillance pilot project represents the first civilian sentinel respiratory virus surveillance program conducted on cruise ships in north america. although the number of specimens collected and participating ships were limited, the high yield of positive results (83% for at least one respiratory virus and 71% for influenza viruses) indicates that influenza was common among sampled passengers and crew members with ili or ari. this finding highlights the value of conducting such surveillance among cruise ship passengers and crews and reinforces the need to have robust influenza prevention and control activities on ships. influenza virus strains and respiratory viruses identified by the pilot were similar to findings of the state surveillance. the latter is conducted year-round in alaska, including in healthcare facilities in port cities where passengers disembark. data collected in the pilot project may aid in determining specific voyages with higher potential for influenza virus transmission. as southbound voyages originating in alaska had peak onset dates on the day before the start of the voyage, most influenza virus transmission may have occurred before embarkation. southbound voyages are generally associated with land-based tours immediately preceding the voyage involving buses and trains, which are additional semi-closed and close-contact environments that can potentially facilitate the transmission of influenza viruses. conversely, northbound voyages (which often have their land-based tours post-voyage) had peak onset dates occurring mid-voyage, suggesting that most influenza virus transmission occurred on board. additional surveillance is needed to better understand the role of land-based tours and potential differences in influenza virus transmission by voyage direction and route. adding a laboratory component to routine cruise ship respiratory surveillance in alaska, such as sending specimens from port to state laboratories, may better inform resource allocations and anticipate needs for cruise ship populations, which are typically disproportionately older. the average age of an alaska cruise ship passenger is 65 years, and influenza in these passengers may be of more concern due to comorbidities in contrast to crew members, who are on average 30 or younger and generally healthy. crew members may seek healthcare for no cost on board ship, but passengers without international or travel healthcare insurance coverage may incur medical costs for infirmary care on international waters. thus, cost could be a disincentive for passengers to seek care until they are severely ill or need to be medically evacuated to a hospital. 3 this disincentive, in addition to participating cruise ship physicians' accuracy in targeting ill individuals for pilot project inclusion, may have contributed to the pilot's relatively high percent positive yield on respiratory pathogen and influenza testing. ili outbreaks can have economic and human resource costs to both cruise ships and public health agencies. additional staff are required to implement active and passive surveillance and coordinate testing; crew time is lost during illness or isolation; and antiviral agents for treatment and chemoprophylaxis of influenza can be costly. in some circumstances, land-based medical assistance may be required for follow-up and treatment of complications. influenza vaccination of crews may help reduce the impact of influenza virus transmission on cruise ships and serve as a cost-saving measure. both participating cruise lines have a target goal of vaccinating 80% of crew members, particularly in guest-facing departments. the findings of ship-based influenza surveillance can augment or supplement land-based surveillance. in alaska, influenza virus transmission during summertime is driven to a degree by the annual influx of tourists. states that experience off-season (summer) influenza virus transmission may benefit from extended summertime surveillance in order to guide recommendations on the timing of influenza vaccination and clinical decisions regarding testing and antiviral treatment, as well as to evaluate the match between circulating influenza virus strains and vaccine strains. these potential benefits should be weighed against the operational limitations in instituting such a surveillance program. while~30-32 cruise ships dock in alaska ports during the summer tourist season, only four to six participated in the alaska cruise ship respiratory surveillance pilot project each year. the small number, in addition to the relatively small number of specimens collected, does not allow generalization of results to all cruise ships in alaska or beyond. the results are also not generalizable to the entire cruise ship, as only ill individuals visited the ship medical centre for testing. as such, no true attack rate for influenza can be established. inclusion of only individuals reporting to the infirmary with respiratory illness may have biased selection of passengers and crew members to those with more severe symptoms or those more likely to seek healthcare. clinicians on participating ships were limited to collecting 2-3 specimens per week on average; thus, they may have chosen more overtly ill individuals to test. additionally, the case definition for participant inclusion changed between years 1 (2013) and 2 (2014). influenza vaccination history was obtained by self-report and may have been subject to recall bias, and the length of time between selfreported vaccination and illness onset was largely unknown. operational issues with specimen storage and shipping resulted iñ 10% (45/455, table 1 ) being lost in transport or deemed unsuitable for testing at the laboratory, and delays in shipping some specimens may have caused some falsely negative results. the alaska cruise ship respiratory surveillance pilot project served as a feasibility test for laboratory-based cruise ship surveillance in alaska. the pilot proved such surveillance is possible and thus may be helpful in informing public health officials and the cruise ship industry of the utility of surveillance to prevent and control outbreaks. in order to apply results to the whole alaska cruise ship population, additional ships should be included and sampling should be more representative of all ill persons. the program was operationally resource-intensive for all stakeholders, particularly cruise ship medical departments in collecting information on the ill individual, and the cdc anchorage quarantine station in equipment shipping and data management. subsequent similar laboratory-based surveillance projects may increase efficiency and lower the resource cost per sample. program evaluation is needed to weigh the time and operational limitations against the utility of this added layer of lab-based respiratory surveillance. despite its limitations, the pilot project's findings highlighted the usefulness of augmenting routine disease reporting and cumulative ili surveillance on ships in alaska. strengthening cumulative ili reporting on cruise ships to the level of required gastrointestinal illness reporting-an illness that generally is less severe than influenza-may be useful for improving the health of the travelling public. cruise ships should participate in requested cumulative ili reporting, and ship healthcare providers could collect recent travel history along with medical history during patient intake. cdc should be notified immediately of ili outbreaks on board, and its published outbreak management guidance 13 should be followed, including specimen collection for respiratory virus identification and implementation of prevention and control measures. conducting limited laboratorybased surveillance during times of respiratory disease outbreaks or at the beginning and end of the cruise ship season, when detection of antigenic variants from ships repositioning to alaska is a higher possibility, should also be considered. cdc provided support to the alaska section of epidemiology through its existing epidemiology and laboratory capacity for infectious diseases (elc) grant for this pilot project. state of alaska department of commerce, community, and economic development influenza outbreaks among passengers and crew on two cruise ships: a recent account of preparedness and response to an ever-present challenge alaska/yukon territory respiratory outbreak investigation team. large summertime influenza a outbreak among tourists in alaska and the yukon territory outbreak updates for international cruise ships how to calculate the influenza or influenza-like illness (ili) case outbreak threshold for cumulative reports outbreak of influenza-like illness in a tour group-alaska outbreaks of pandemic (h1n1) 2009 and seasonal influenza a (h3n2) on cruise ship shipsan trainet project. the decision making process on public health measures related to passenger ships: the example of influenza pandemic cruise ships: high-risk passengers and the global spread of new influenza viruses world health organization. manual for the laboratory diagnosis and virological surveillance of influenza single-reaction genomic amplification accelerates sequencing and vaccine production for classical and swine origin human influenza a viruses guidance for cruise ships on influenza-like illness (ili) management we are grateful to the participating cruise ship clinicians, alaska maritime shipping agents, and dr sue gerber and dr dean erdman from the cdc division of viral diseases for their support during this project. the findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the centers for disease control and prevention. key: cord-336335-spap39b7 authors: silva, denise r; viana, vinícius p; müller, alice m; livi, fernando p; dalcin, paulo de tarso r title: respiratory viral infections and effects of meteorological parameters and air pollution in adults with respiratory symptoms admitted to the emergency room date: 2013-08-26 journal: influenza other respir viruses doi: 10.1111/irv.12158 sha: doc_id: 336335 cord_uid: spap39b7 background: respiratory viral infections (rvis) are the most common causes of respiratory infections. the prevalence of respiratory viruses in adults is underestimated. meteorological variations and air pollution are likely to play a role in these infections. objectives: the objectives of this study were to determine the number of emergency visits for influenza-like illness (ili) and severe acute respiratory infection (sari) and to evaluate the association between ili/sari, rvi prevalence, and meteorological factors/air pollution, in the city of porto alegre, brazil, from november 2008 to october 2010. methods: eleven thousand nine hundred and fifty-three hospitalizations (adults and children) for respiratory symptoms were correlated with meteorological parameters and air pollutants. in a subset of adults, nasopharyngeal aspirates were collected and analyzed through ifi test. the data were analyzed using time-series analysis. results: influenza-like illness and sari were diagnosed in 3698 (30·9%) and 2063 (17·7%) patients, respectively. thirty-seven (9·0%) samples were positive by ifi and 93 of 410 (22·7%) were ifi and/or pcr positive. in a multivariate logistic regression model, ifi positivity was statistically associated with absolute humidity, use of air conditioning, and presence of mold in home. sunshine duration was significantly associated with the frequency of ili cases. for sari cases, the variables mean temperature, sunshine duration, relative humidity, and mean concentration of pollutants were singnificant. conclusions: at least 22% of infections in adult patients admitted to er with respiratory complaints were caused by rvi. the correlations among meteorological variables, air pollution, ili/sari cases, and respiratory viruses demonstrated the relevance of climate factors as significant underlying contributors to the prevalence of rvi. respiratory tract infections are the most common causes of infection, and viruses account for the majority of these infections, leading to significant levels of morbidity and mortality. 1 emergency rooms (ers) serve as the frontline for patients at highest risk for respiratory infection diseases, especially because of the acute nature of these illnesses. 2 the prevalence of respiratory viral infection (rvi) in adults admitted to the er is largely unexplored, as most relevant data concern infants and children. 3, 4 rvi can be severe in elderly patients, especially in those with underlying respiratory or cardiac disease. during winter months, rvi can account for many of the admissions to hospitals. 5, 6 the etiology of respiratory infections in adults remains undetermined in more than 50% of cases. 7, 8 in a study with 510 adults hospitalized with pulmonary diseases, an overall prevalence of respiratory viruses (rvs) in the lower respiratory tract was of 42á2%, with rhinoviruses and influenza a virus being the most common. 9 in adults with acute asthma admitted to the er, a prevalence of 12á2% of rvi was found. 10 in another study, with adults admitted to hospital with respiratory symptoms, viruses accounted for 15% of hospital admissions for respiratory infections. 11 seasonality of certain acute respiratory tract infection pathogens can be explained by meteorological variations. in a study, temperature was highly inversely correlated with respiratory syncytial virus (rsv), influenza a, and adenovirus frequency; rhinovirus was also associated with relative humidity (rh). climatic factors may influence the interaction among the host, pathogen, and environment, increasing the probability of exposure, susceptibility, and infection. 12 in addition, experimental data have shown that air pollutants affect lung immune responses and inflammatory reactions and that these effects may underlie the increased risk for respiratory infections. 13, 14 because of the large impact respiratory virus infections have on morbidity and even mortality, it is important to understand whether and how meteorological factors and exposure to air pollutants could influence respiratory virus infections. the aims of this study were to determine the number of emergency room visits for influenza-like illness (ili) and severe acute respiratory infection (sari) and to evaluate the association between ili/sari frequency, respiratory virus prevalence, and meteorological factors/air pollution, especially in adult population, in a humid subtropical climate. the present study was divided into two parts: in the first one, we characterized the symptomatic respiratory subjects attending the er at hospital de cl ınicas de porto alegre (hcpa), during 1 year (november 2008 to october 2009). in the second part, we included patients with respiratory symptoms ≤5 days to determine the prevalence of rvi; this part was conducted during 2 years (november 2008 to october 2010). in the first year of the study, we also collected climate and air pollution data. the study was conducted at hcpa, in the city of porto alegre, southern brazil. the hcpa is a general, tertiary care, university-affiliated hospital with 750 beds and approximately 30 000 hospitalizations/year. with a population of 1 360 590 inhabitants, porto alegre is surrounded by a metropolitan area that encompasses 31 municipalities (3 717 430 inhabitants). porto alegre has a humid subtropical climate, with the hallmark of the great variability (classification cfa in k€ oppen-geiger). 15 the ethics committee at hcpa has approved access to patient records. all subjects selected for the study gave written informed consent to participate. in the first part of the study, the clinical records of all daily visits (adults ≥18 years and children) to the er were reviewed by the research team. patients with respiratory symptoms (cough, coryza, nasal obstruction, odynophagia, dyspnea, chest pain, dysphonia, wheezing, and fever), regardless of the onset time, were included in the study. patients presenting only with fever were not included. demographic, clinical, and laboratorial characteristics were registered in a standardized questionnaire: sex, age, race, years of schooling, smoking status, symptoms at admission, duration of symptoms, presence of comorbidities, admission vital signs (temperature, heart rate, respiratory rate, and peripheral oxygen saturation measured by a digital oximeter), breath sounds, radiological findings, length of hospital stay, intensive care unit (icu) admission, hospitalization outcome (death or discharge). in the second part, adults ≥18 years with respiratory symptoms ≤5 days were included, and nasopharyngeal aspirates were collected. every day, a daily shift (morning, afternoon, or evening) was randomized for inclusion of patients. the patients were interviewed by a member of the research team, and the following data were registered in a standardized questionnaire (in addition to the data already registered in the first part): family income, influenza vaccine, previous antibiotic use, and occupational exposure, flu symptoms in the family, presence of smokers at home, family history of respiratory disease, air conditioning at home or at work, use of wood stoves, mold in home. the total cases of ili (fever >38°c and cough or sore throat) and sari (fever >38°c and cough or sore throat and shortness of breath or difficulty breathing) were registered. nasopharyngeal aspirates were obtained according to a standard protocol for the second part of study. an indirect immunofluorescence assay (ifi) was carried out using the respiratory daily meteorological parameters, like temperature (maximum, average, and minimum, in°c), relative (%) and absolute (g/kg) humidity, rainfall (mm), and sunshine duration (number of sunshine hours per day), were obtained from the climate laboratory at the federal university of rio grande do sul. also, the mean concentration of pollutants (lm/m 3 ) was recorded. the methodology used by this laboratory informs automatically the pollutant that reached the highest concentration in the last 24 hours. in the city of porto alegre, it is probable that the pollutant that is responsible for more than 90% of the days is ozone, and in the rest of days, particulate matter of <10 lm dynamic diameter (pm 10 ). data were presented as number of cases, mean ae standard deviation (sd), or median with interquartile range (iqr). categorical comparisons were carried out by chi-square test using yates's correction if indicated or by fisher's exact test. continuous variables were compared using the t-test or wilcoxon test. odds ratios (ors) and nominal 95% confidence intervals (ci) were presented. a two-sided p value <0á05 was considered significant for all analyses. data analysis was performed using spss 18á0 (statistical package for the social sciences, chicago, il, usa) and the free statistical software r (http://www.r-project.org). the data were analyzed using time-series analysis, through a generalized linear model (glm) to examine the association between ili or sari and air pollution/meteorological variables, using logistic regression. an autoregressive integrated moving average (arima) model was developed for ili or sari. pollution and meteorological parameters were inserted as explanatory variables into these arima models. for the second part of the study, further models were explored to include ifi and/or pcr results as outcome variables. volatility of the mean model variance error(s) was addressed using autoregressive conditional heteroscedasticity (arch) models within an arima modeling framework. an arima was developed for ili and sari. arima (p, d, q) are useful tools for analyzing time-series data containing non-stationary common trends, and these models were proposed by box and jenkins in 1976. the arima (p, d, q) allow to make predictions from their parties ar (autoregressive) and ma (moving average). for both sg and sars, we made the identification of the order of the parts and autoregressive moving average following the box-jenkins approach. first we identified the need for differentiation, for ili and sari, the graphical analysis of time-series and autocorrelation functions, as well as for testing the unit root (dickey-fuller) who did not reject the hypothesis of non-stationarity for ili and sari. in a second moment, it was noted that there was no seasonality for ili and sari, by analyzing the periodograms and autocorrelation functions and partial autocorrelation. subsequently, variable selection models arima (p, d, q) for ili and sari followed the approach backward. the akaike information criterion (aic) which acts to penalize the number of parameters in the modeland the schwarz information criterion (sic) (or bayesian information criterion, bic) were also used to select models. sample size requirements were estimated from the literature review. 10, 11 using a prevalence of rvi in adults of 15%, with a significance level of 99%, and total confidence interval amplitude of 0á10, we calculated that 339 patients would be needed in the second part of the study. during the 12-month study period, there were 37 059 admissions to the er (24 189 adults and 12 870 children), of which 11 953 (32á3%) presented with respiratory symptoms. the most common symptoms were cough (73á4%), fever (56á1%), dyspnea (40á9%), chest pain (24á5%), and coryza (20á9%). the median duration of symptoms before admission was 3 days (iqr: 1-6 days). a total of 2205 (18á5%) patients admitted to er needed to be hospitalized; of these patients, 242 (2á0%) required icu admission. ili and sari were diagnosed in 3698 (30á9%) and 2063 (17á7%) patients, respectively. the overall mortality rate among all study participants was 280 of 11 953 (2á3%). demographic and clinical characteristics of the study population are shown in table 1 . according to the selection criteria of the study, 425 patients met the inclusion criteria and were invited to participate, and 15 patients declined to participate. then, 410 adults were enrolled for virological investigation, 255 in the first year and 155 in the second one. there were no differences in age, sex, race, years of schooling, and symptoms between selected sample and all adults admitted to the er in the same period. however, as expected, the median duration of symptoms was lower in the selected seven samples considered to be unsatisfactory for ifi were positive by pcr. another 50 samples that were negative in ifi were positive by pcr. of the 37 samples positive in ifi, 18 were sent for pcr too, and in nine, the result was positive. in seven of these cases, the results of ifi and pcr were concordant. in one patient, adenovirus was identified on ifi and influenza a was found in pcr. in another case, ifi detected piv type 2, and pcr identified piv types 1 and 3. the characteristics of patients with pcr positive and negative were shown in table 3 . figure 1 shows timesseries graph for virus percent positive by ifi and by pcr by month. table 4 shows the descriptive statistics corresponding to the environmental variables considered in this study. figure 2 shows the modeled and observed values for ili and sari cases. figures 3 and 4 show the daily number of patients with ili and sari and meteorological parameters. we checked the autocorrelation between the covariates of climate data and observations from previous days, and we considered the following lags (in days) for each variable: average temperature (5), rainfall (2), sunshine duration (6), relative humidity (5), mean concentration of pollutants (3), and absolute humidity (2) . the number of ili and sari cases tends to be higher between july 5, 2009, and august 22, 2009 . in this period, the mean temperatures (tmin: 9á52 ae 3á89°c and tmax: 18á9 ae 4á92°c) and the sunshine duration (4á13 ae 3á31 hours of sun per day) were lower, as expected in winter. the rainfall tends to be higher than the median calculated for the entire year (0 mm, iqr: 0-5á4 mm). in addition, the absolute humidity (ah; 7á8 ae 2á1 g/kg) and mean concentration of pollutants (20á0 ae 7á0 lm/m 3 ) were lower in this period compared with the annual values. table 5 shows multivariate logistic regression model for ifi and pcr. we included 255 patients in logistic regression for ifi and 180 patients in logistic regression for pcr, because we have climate data only for the first year of study. ifi positivity was statistically associated with ah (or: 0á72; 95% ci 0á59-0á86), use of air conditioning (or: 4á16; 95% ci 1á45-11á83), and presence of mold in home (or: 2á95; 95% ci 1á10-8á29). on the other hand, pcr positivity was statistically associated with use of air conditioning (or: 2á27; 95% ci 1á04-4á97), average temperature (or: 0á92; 95% ci 0á86-0á98), and mean concentration of pollutants (or: 1á04; 95% ci 1á00-1á08). the multivariate time-series models for ili and sari cases are summarized in table 6 . sunshine duration was the only independent covariate that was significantly associated with the frequency of ili cases. the b-coefficient for this parameter was negative, indicating increasing ili frequency with decreasing sunshine duration. in the model for sari cases, the following variables proved to be significant: mean temperature (b = 0á399; p = 0á025), sunshine duration (b = à0á392; p = 0á007), rh (b = à0á098; p = 0á05), and mean concentration of pollutants (b = à0á079; p = 0á018). acute rvis are responsible for causing significant levels of morbidity and mortality. the most common respiratory syndrome caused by these pathogens is ili. a more severe presentation, named sari, was also related to some rvs. 16, 17 in this study, we have examined the relationship between ili and sari cases, meteorological variables, and air pollution using multivariate time-series analyses. we found that ili cases were inversely correlated with sunshine duration. in addition, sari cases were significantly associated with mean temperature, sunshine duration, rh, and concentration of pollutants. seasonal cycles of infectious diseases have been attributed to changes in atmospheric variables, the prevalence or virulence of the pathogen, or the behavior of the host. 18 earlier investigations have demonstrated that lower temperatures and sunshine duration, conditions usually encountered in winter, were associated with admissions for rvi. 12, 19 temperature was found to be highly inversely correlated with rsv, influenza a, and adenovirus frequency. 12 interestingly, we found a positive correlation between temperature and sari cases. one possible explanation is that it was demonstrated that for every one degree celsius rise in temperature, the risk of premature death and acute morbidity especially among respiratory patients is up to six times higher than in the rest of the population. second, evidence is emerging that increasing temperature is associated with increases in air pollution, especially ground-level ozone, and can amplify the adverse effects of poor air quality. 20 taking this evidence into account, we could expect that higher temperatures may have increased concentration of pollutants, leading to more sari cases. however, our data showed a decrease in air pollution during the months with a higher prevalence of sari. the third hypothesis to explain the relationship between higher temperatures and sari cases was related to el niño southern oscillation (enso) phenomenon. enso undergoes cycles between warm phases (el niño episodes) and reverse cold phases (la niña episodes). in the southern region of brazil, this phenomenon is associated with elevated temperatures and rainfall, especially in spring and in the period between may and july. previous reports have determined that el niño events were associated with increased hospitalizations and more severe influenza epidemics. 21, 22 severe acute respiratory infection cases were found to be negatively related to rh in our study. previous studies have demonstrated that higher rh decreases the survival of lipidenveloped virus, like influenza a, influenza b, rsv, and piv. [23] [24] [25] the use of indoor heating in winter lowers the rh; breathing dry air could cause desiccation of the nasal mucosa, epithelial damage, and reduced mucociliary clearance, increasing the host susceptibility to rvis. 19 however, even in tropical regions with humid climate (rh >70%), a higher activity of influenza can be found. this observation could be explained by the variation of viral stability in different rh levels. the stability of aerosolized influenza virions is maximal at lower rh (20-40%), moderate at higher rh (60-80%), and minimum at a mid-range rh (50%). 23 in a multivariate logistic regression model for ifi-positive patients, we found that ah was a protect factor for rvi. a recent study suggested that ah may better correlate with influenza virus survival and transmission. unlike rh, ah measures the actual water vapor content of air irrespective of temperature and has a prominent wintertime low, both indoor and outdoor. such findings suggest that humidification measures could be helpful decreasing survival and transmissibility of influenza. 26 air pollution has been associated with adverse health outcomes. studies have suggested acute effects causing respiratory symptoms, cardiovascular events, hospital admissions, and mortality. although the available evidences indicate associations between exposure to pollutants and increased risk of rvi, potential mechanisms mediating these effects are largely unexplored. 27, 28 surprisingly, our results showed that sari cases were associated with a decrease in mean concentration of pollutants. in fact, this could be a reflection of higher rainfall in the same period, as rain acts washing out or scattering pollutants from atmosphere. 29 on the other hand, we cannot exclude an effect of indoor pollution. in the last years, indoor pollution has been recognized as an emerging health problem, as about 90% of our time is spent indoors where we are exposed to chemical and biological contaminants. 30 we estimated indoor pollution indirectly in our study, questioning patients about the use of wood stoves and air conditioning, and the presence of mold in home. our findings suggested that ifi-positive patients were more prone to live in a residence with mold growth. dampness and mold are two important sources of indoor pollution, consistently associated with respiratory symptoms. home dampness may be a marker for mold growth, dust mites, endotoxins, and reduced ventilation, which could increase concentrations of indoor pollutants. 31 cough, wheezing, and upper respiratory symptoms were associated with dampness and mold in a metaanalysis. 32 according to these results, the prevalence of cough and wheezing was higher in patients with mold in home and ifi positive. air conditioning was also positively related to ifi test in this study. air conditioning use was associated with fewer hospital admissions for cardiovascular diseases, chronic obstructive pulmonary disease, and pneumonia on days with high concentrations of pm 10 , 33 as individuals are less exposed to outdoor pollutants. nevertheless, the majority of virus transmission occurs within indoor, air-conditioned (i.e., cooler, lower humidity) environments that favor airborne virus survival and transmission. 24, 25 in hot and humid conditions, indoor transmission in air conditioning environments may account for most of the transmission. 34 we found a prevalence of 22% of rv, which is higher than that previous studies have demonstrated (between 12% and 15% in adults). 12, 13 moreover, the length of stay was lower in our ifi-and/or pcr-positive patients. this finding is consistent with existing knowledge that virus identification allows the prompt initiation of therapy when indicated and avoids the unnecessary use of antibiotics, decreasing the length of hospital stay. the present study has some limitations. first, it was based on data collected from a single center, which may have potential biases because of the characteristics of the catchment population, like vaccination coverage. second, it is also important to note that this investigation was performed in a group of hospitalized patients, which is a bias toward the most severe disease cases. additionally, we do not have the concentrations of individual air pollutants, but it is implausible to reliably separate the effects of air pollutants because they frequently react with each other, sometimes potentiating individual effects. 10, 35 the short study period should also be considered a limitation. finally, the use of molecular techniques (pcr) in all study patients could be useful, increasing the number of viruses detected, as limited sensitivity of ifi method is well known. 10, 36 despite these limitations, this is the first study, to our knowledge, to analyze the relationship between rv, meteorological parameters, and air pollution in an adult population. in conclusion, we found that in adult patients admitted to er with respiratory complaints, at least 22% of infections were caused by rv. the correlations found among meteorological variables, air pollution, ili/sari cases, and rv demonstrated the relevance of climate factors as significant underlying contributors to the prevalence of rvi in a temperate region. there is still a need of additional investigations to clarify and confirm these data, perhaps using longer time-series observations. fundo de incentivo a pesquisa -sbpt; fipe-hcpa; fapergs. communicable respiratory threats in the ed: tuberculosis, influenza, sars, and other aerosolized infections role of respiratory viruses in acute upper and lower respiratory tract illness in the first year of life: a birth cohort study detection and typing by molecular techniques of respiratory viruses in children hospitalized for acute respiratory infection in rome, italy the contribution of influenza to combined acute respiratory infections, hospital admissions, and deaths in winter excess hospital admissions for pneumonia and influenza in persons > or = 65 years associated with influenza epidemics in three english health districts: 1987-95 community-acquired pneumonia prospective epidemiologic survey of patients with community-acquired pneumonia requiring hospitalization in switzerland frequency of detection of respiratory viruses in the lower respiratory tract of hospitalized adults incidência de infecc ßão viral do trato respirat orio em asma aguda atendida em sala de emergência surveillance of respiratory virus infections in adult hospital admissions using rapid methods are meteorological parameters associated with acute respiratory tract infections? increased susceptibility to rsv infection by exposure to inhaled diesel engine emissions ultrafine carbon black particles enhance respiratory syncitial virus-induced airway reactivity, pulmonary inflammation, and chemokine expression updated world map of the k€ oppen-geiger climate classification influenza activity-united states and worldwide, 2007-08 season influenza pandêmica (h1n1) 2009 -an alise da situac ßão epidemiol ogica e da resposta no ano de seasonal variation in host susceptibility and cycles of certain infectious diseases influenza virus transmission is dependent on relative humidity and temperature climate change and respiratory disease: european respiratory society position statement association of normal weather periods and el nino events with hospitalization for viral pneumonia in females: california, 1983-1998 respiratory virus and the effects of climate ª 2013 the authors. influenza and other respiratory viruses published by survival of airborne influenza virus: effects of propagating host, relative humidity, and composition of spray fluids the effect of environmental parameters on the survival of airborne infectious agents aerosol transmission of influenza a virus: a review of new studies absolute humidity modulates influenza survival, transmission, and seasonality air pollution and health ambient air pollution and health climate change and human health the index project: executive summary of a european union project on indoor air pollutants variations of formaldehyde and voc levels during 3 years in new and older homes meta-analyses of the associations of respiratory health effects with dampness and mold in homes air conditioning and source-specific particles as modifiers of the effect of pm(10) on hospital admissions for heart and lung disease incidence of common respiratory viral infections related to climate factors in hospitalized children in hong kong air pollution and cardiovascular disease: a statement for healthcare professionals from the expert panel on population and prevention science of the american heart association advances in the laboratory diagnosis of viral respiratory disease the authors have no competing interests to disclose. key: cord-264140-5cxzc3z8 authors: tam, clarence c.; anderson, kathryn b.; offeddu, vittoria; weg, alden; macareo, louis r.; ellison, damon w.; rangsin, ram; fernandez, stefan; gibbons, robert v.; yoon, in-kyu; simasathien, sriluck title: epidemiology and transmission of respiratory infections in thai army recruits: a prospective cohort study date: 2018-09-04 journal: am j trop med hyg doi: 10.4269/ajtmh.18-0219 sha: doc_id: 264140 cord_uid: 5cxzc3z8 military recruits are at high risk of respiratory infections. however, limited data exist on military populations in tropical settings, where the epidemiology of respiratory infections differs substantially from temperate settings. we enrolled recruits undertaking a 10-week military training at two royal thai army barracks between may 2014 and july 2015. we used a multiplex respiratory panel to analyze nose and throat swabs collected at the start and end of the training period, and from participants experiencing respiratory symptoms during follow-up. paired sera were tested for influenza seroconversion using a hemagglutinin inhibition assay. overall rates of upper respiratory illness and influenza-like illness were 3.1 and 2.0 episodes per 100 person-weeks, respectively. a pathogen was detected in 96% of samples. the most commonly detected microbes were haemophilus influenzae type b (62.7%) or non–type b (58.2%) and rhinovirus (22.4%). at baseline, bacterial colonization was high and included h. influenzae type b (82.3%), h. influenzae non–type b (31.5%), klebsiella pneumoniae (14.6%), staphylococcus aureus (8.5%), and streptococcus pneumoniae (8.5%). at the end of follow-up, colonization with h. influenzae non–type b had increased to 74.1%, and s. pneumoniae to 33.6%. in the serology subset, the rate of influenza infection was 3.4 per 100 person-months; 58% of influenza infections resulted in clinical disease. our study provides key data on the epidemiology and transmission of respiratory pathogens in tropical settings. our results emphasize the need for improved infection prevention and control in military environments, given the high burden of illness and potential for intense transmission of respiratory pathogens. military recruits are at high risk of respiratory infections. 1 the congregation of individuals from diverse geographic locations in semi-closed settings, together with high levels of close contact, provide conditions that favor the introduction and transmission of respiratory pathogens. 2 studies among military recruits have found high rates of illness and infection with respiratory viruses. a longitudinal study in singapore found that 39.4% of military personnel had evidence of seroconversion against influenza a/h1n1 during the 2009 pandemic, compared with 13.5% of individuals in the community and 6.5% of hospital staff. 3 numerous outbreaks of respiratory infections among military recruits have been reported in the literature, including outbreaks of influenza, adenovirus, 4-6 and pertussis. 7 resumption of vaccination against adenovirus types 4 and 7 in the u.s. military is estimated to have resulted in a 7-fold decrease in acute respiratory disease 8 and a shift in the predominant adenovirus types from types 4, 3, and 7 to types 1 and 2. the increasing availability of multiplex molecular diagnostic assays can yield data on multiple pathogens that would not routinely be detected. 9, 10 studies in military populations can provide valuable information regarding the epidemiology and transmission of respiratory infections in adults because of the availability of well-defined populations that can be followed up over time. despite this, data from tropical settings, in which the epidemiology of influenza and other respiratory infections differs substantially from temperate settings, are limited. increased understanding of the etiology of respiratory infections in these settings is important to identify opportunities for disease control in high-risk military populations and to identify risks of pathogen emergence with potential for spread into the wider community. we present the results of a longitudinal study of the burden and etiology of respiratory infections in thai military recruits. the study setting and procedures have been previously described. 11 between may 2014 and may 2015, we enrolled participants from consecutive cohorts of recruits undertaking basic military training at two royal thai army (rta) barracks in bangkok. trainees entered the camps at the start of may and november each year and remained in the camps for 10 weeks. the camps consist of large, common sleeping quarters with beds arranged foot-to-foot in long rows. trainees share meals in a common canteen and train in a large field in the center of the barracks and a number of ancillary buildings. each camp has its own medical unit. individuals were eligible for enrollment if they were aged ³ 18 years and were entering one of the two army barracks involved in the study. suspected tuberculosis cases or individuals with immune deficiencies, such as acquired immune deficiency syndrome, leukemia, or lymphoma, were excluded. individuals providing informed consent were enrolled in the study and a demographic and clinical questionnaire was administered. participants were then followed up for symptoms of respiratory illness until the end of the 10-week training period. at one camp, nasal and throat swabs were obtained at the time of enrolment and at the end of the follow-up period (10 weeks later) from consenting participants. participants experiencing respiratory symptoms were asked to consult the medical unit, where medical staff took a history, conducted a medical examination, and recorded symptoms of upper respiratory illness (uri) or influenza-like illness (ili). upper respiratory illness was defined as an illness with at least two of the following: 1) runny nose or sneezing; 2) nasal congestion; 3) sore throat, hoarseness, or difficulty swallowing; 4) cough; 5) swollen or tender glands in the neck; and 6) fever (oral temperature > 38°c) at the time of presentation. influenza-like illness was defined as a respiratory illness with acute onset presenting with fever and cough or sore throat. all ili cases also met the case definition for uri. non-ili cases were defined as uri cases not meeting the case definition for ili. additional nasal and throat swabs were requested from participants presenting with acute ili or uri. laboratory investigations. nasal and throat swabs were placed in viral transport media and stored at −20°c until transfer to the armed forces research institute of medical sciences for further testing. we tested acute swabs for influenza virus using reverse transcription polymerase chain reaction (rt-pcr) following the u.s. centers for disease control protocol. 12, 13 we also tested acute samples (from both camps) and the routine enrolment and follow-up specimens (from one camp) using a multiplex real-time pcr assay comprising 33 bacterial, viral, and fungal targets (ftd33 kit; fast track diagnostics, esch-sur-alzette, luxembourg). a cycle threshold value of < 33 was considered a positive result. previous studies have shown high agreement between ftd33 and other commercial multiplex assays with specificities > 95% for most viral targets. 14,15 sensitivity for influenza a and b, and human coronaviruses is high (> 90%), but is lower and more variable for other viral targets, including human bocavirus (80-92%), rhinovirus (61-75%), respiratory syncytial virus (50-72%), and human metapneumovirus (74-100%). 14, 15 in a subset of recruits, we measured influenza antibody titers in paired sera taken at enrollment and at the end of followup using a hemagglutinin inhibition assay. seroconversion was defined as a 4-fold rise in influenza subtype-specific antibody titers. statistical analysis. rates of illness. individuals were considered at risk from the date they entered the camp for a period of 10 weeks. we computed the rates of uri and ili as the number of cases of each outcome divided by the total person-time at risk, using robust standard errors in the calculation of 95% confidence intervals (cis) to allow for clustering of illness episodes by camp and cohort. disease etiology. we determined the percentage of uri and ili cases positive for each target pathogen, the percentage positive for more than one pathogen, and the percentage with no pathogen identified. influenza seroconversion. we calculated the percentage of recruits who seroconverted, overall and by influenza subtype, based on a 4-fold rise in antibody titer between baseline and end of follow-up samples. bacterial colonization. we determined the percentage of recruits with a positive identification of a bacterial pathogen at recruitment, the overall change in colonization prevalence between baseline and end of follow-up for each bacterial species, and the risk of acquisition of bacterial pathogens over the follow-up period among those initially negative. antibiotic use. we calculated the percentage of uri and ili cases that were prescribed antibiotics and investigated clinical signs and symptoms associated with antibiotic prescription using logistic regression. among those prescribed antibiotics, we determined the fraction of cases with a likely viral etiology based on the available fast track multiplex pcr results. analyses were conducted using stata 11 (statacorp, college station, tx) and r 3.2.2. ethical approval. the study was approved by the institutional review boards of the rta in bangkok, thailand, the walter reed army institute of research, and the london school of hygiene & tropical medicine. all participants provided written informed consent. three cohorts of army recruits undertook basic military training at the two camps during the study period, in may 2014-july 2014, november 2014-january 2015, and may 2015-july 2015. in total, 933 recruits undertook training during these three periods. of these, 867 (93%) were enrolled in the study. of the 867 participants, five participants withdrew from the study and 44 participants recruited in november 2014 were transferred out of the camp to complete their training elsewhere. the remaining 818 (94%) completed the 10-week follow-up and were included in the analysis. all participants were male and the median age was 21 years (range: 20-29 years). details of recruitment and follow-up completion for each cohort are given in supplemental figure 1 . rates of illness. there were a total of 248 episodes of uri and 8,063 person-weeks of follow-up, yielding an overall uri rate of 3.1 per 100 person-weeks. upper respiratory illness rates ranged from 2.6 to 4.6 per 100 person-weeks across cohorts, with the exception of the camp a may 2014 cohort, in which no uri episodes were reported ( figure 1 ). there were 160 ili episodes, giving an overall ili rate of 2.0 per 100 person-weeks, ranging from 0 to 4.1 per 100 person-weeks across cohorts. symptoms and time off work. fever was present in nearly two-thirds of uri episodes. cough and sore throat occurred in > 90% of uri episodes, nasal congestion and headache in > 80% of episodes, and malaise and breathing difficulty in > 60% of episodes. all these symptoms were significantly more common in ili compared with non-ili episodes ( table 1) . influenza-like illness cases were also more likely to take time off work than non-ili cases (57.5% versus 13.6%, p < 0.001), although the length of time taken off work was short (median 1 day). there was one hospitalization in an individual diagnosed with pneumonia who was off work for 7 days. disease etiology. among the 134 uri and ili samples tested by multiplex pcr, a pathogen was detected in 128. a single pathogen was detected in 26% of samples, two pathogens in 31%, three pathogens in 26%, and four pathogens in 12%. among uri pathogens, the most common viruses detected were rhinoviruses (22.4%), influenza b (11.9%), coronavirus 229 (5.2%), adenovirus (3.0%), and influenza a/h3 (3.0%). other viruses accounted for < 3% of uri cases each. of 133 samples tested for influenza by both rt-pcr and the fast track multiplex assay, 20 (15%) tested positive by at least one assay, and all of these were also positive by rt-pcr. of the 20 rt-pcr positives, four were positive for influenza a/h3, of which one was also positive by the multiplex assay. of the remaining 16 positive for influenza b by rt-pcr, 14 were also positive by the multiplex assay. there were no samples that tested positive by the multiplex assay and negative by rt-pcr. among the bacteria, the most commonly detected species were haemophilus influenzae type b (62.7%), h. influenzae non-type b (58.2%), streptococcus pneumoniae (16.4%), and klebsiella pneumoniae (16.4%). however, these were commonly found in combination with other pathogens. when considering only samples in which a single organism was found, h. influenzae type b was found in 13.4% of samples, h. influenzae non-type b in 9.0%, and s. pneumoniae, k. pneumoniae, and legionella pneumophila in 0.7% each. there was little evidence to suggest that etiological agents differed between ili and non-ili cases, with the exception of h. influenzae non-type b and coronavirus 229, which occurred more frequently in non-ili cases, and influenza a/h3 and coronavirus hku1, which occurred more frequently in ili cases ( table 2) . analysis of uri cases by time since the start of follow-up indicated strong temporal clustering of cases, with each cluster involving several respiratory pathogens (figure 2 ). for example, the single cluster in camp a during the november 2014 cohort resulted in identification of both rhinovirus and coronavirus 63, as well as numerous colonizing bacteria. in other clusters, rhinovirus was also identified in combination with parainfluenza viruses, adenovirus, and influenza b virus ( figure 2) . influenza vaccination and seroconversion. at enrollment, 2.4% of recruits reported having received influenza vaccination in the previous 12 months. in the serology subset, table 3) . thirteen of 169 (8%) recruits seroconverted over the 10-week training period. seroconversion was most common against influenza b/massachusetts (n = 7) and influenza a/h3 (n = 3) (table 3) colonization. at baseline, 82.3% of recruits were colonized with h. influenzae type b and 31.5% with h. influenzae non-type b. other common colonizing bacteria were k. pneumoniae (14.6%), staphylococcus aureus (8.5%), and s. pneumoniae (8.5%). at the end of follow-up, colonization with h. influenzae non-type b had increased to 74.1%, and s. pneumoniae to 33.6%, whereas colonization with h. influenzae type b and k. pneumoniae had decreased to 67.2% and 8.4%, respectively. in addition, moraxella catarrhalis was found in four (3.1%) and l. pneumophila in two (1.5%) end of follow-up samples. no change in s. aureus colonization was found. in a subset of paired baseline and end of follow-up samples initially negative for h. influenzae non-type b (n = 89), the risk of h. influenzae non-type b acquisition was 70.1% (95% ci: 54.4-90.6%). among those negative for s. pneumoniae at baseline (n = 119), the risk of acquiring s. pneumoniae during the follow-up period was 31.9% (95% ci: 22.6-43.8%). use of medications. of 248 uri episodes, 39.1% were prescribed antibiotics. of these, all but one were prescribed amoxicillin or azithromycin. in multivariable logistic regression, cases were more likely to be prescribed antibiotics if they presented with tonsillitis (or = 5.10, 95% ci: 2.35-11.10), chills (or = 3.91, 95% ci: 1.44-10.62), and swollen lymph nodes (or = 2.54, 95% ci: 1.07-6.04). antibiotic prescription was less likely if patients presented with nasal congestion (or = 0.26, 95% ci: 0.12-0.54). there was no difference in prescribing between ili and non-ili patients. of 35 uri cases prescribed antibiotics and with available multiplex pcr results, 16 (45.6%) had viral pathogens detected and eight (22.9%) had a bacterial target detected in the absence of other pathogens. we found a high incidence of uri in thai army recruits, indicating that approximately 30% of recruits experienced a uri episode over the 10-week training period. comparison with other military cohorts is not straightforward because of differences between settings in terms of vaccination policies and seasonality of key pathogens such as influenza. the uri and ili rates in our study are nonetheless comparable with those reported by other studies in the united states. [16] [17] [18] despite this, we found low rates of clinical influenza a in this minimally vaccinated population. possible reasons for this could be exposure to circulating influenza strains before recruits entering the training camps. thai national influenza surveillance data indicate high influenza activity in the first half of 2014, with influenza a/h1n1 and influenza b predominating. 19 concomitantly, our data indicated that 55% of recruits with available serological information had baseline influenza a/h1n1 titers ³ 1:40. titers greater than this value have been shown to be associated with a 50% reduction in risk of clinical influenza from homologous strains, for both influenza a and b viruses. [20] [21] [22] [23] of interest, 86% of recruits also had baseline titers against influenza b/massachusetts greater than this level. the higher baseline titers against influenza b/massachusetts are consistent with the frequent detection of influenza b isolates from the yamagata lineage in clinical specimens from thailand in the first half of 2014. 19 in our study, the fast track multiplex assay appeared to have lower sensitivity for influenza compared with the cdc rt-pcr assay. we did not specifically investigate reasons for these discrepancies, although it should be noted that the two assays were not performed at the same time, as the multiplex assays were all performed at the end of data collection. we therefore cannot discount the possibility that sample degradation might have influenced the relative performance of the fast track assay. although influenza incidence was low, pathogen detection with sensitive multiplex diagnostics indicated high frequencies of respiratory pathogens, with co-circulation of multiple viruses including rhinovirus, adenovirus, and coronaviruses as well as influenza. in addition, we found high levels of colonization with numerous bacterial species, including both type b and non-type b h. influenzae, k. pneumoniae, and s. pneumoniae. although we did not specifically study transmission chains or epidemiologic links between cases of illness, in the context of our study, recruits did not leave the camps during the 10-week training periods, such that the occurrence of respiratory pathogens is most likely to have resulted from transmission within the camps rather than from external sources. intense transmission of bacterial species was apparent, as the risks of acquiring h. influenzae non-type b and s. pneumoniae among initially non-colonized individuals were 70% and 31%, respectively. bacterial species were also commonly found in samples from uri cases. the colonizing nature of these bacterial species makes it difficult to determine the clinical relevance of these detections. however, h. influenzae type b and non-type b were identified in 13% and 9% of clinical samples in the absence of any other pathogens, suggesting that these pathogens could be responsible for illness in a fraction of adult uri cases. interestingly, h. influenzae type b has not been previously associated with a significant burden of uri in adults. 24 it is possible, however, that these cases were caused by other pathogens not included in the multiplex assay. high levels of colonization with bacterial species also point to the need for judicious use of antibiotics to treat uri in military populations. a quarter of uri cases in our study were prescribed broad-spectrum antibiotics, of whom 50% were more likely to have infections caused by viral pathogens based on multiplex pcr results. although we lacked additional information to determine whether this resulted in adverse antibiotic resistance patterns, high levels of antibiotic use in a population with a high background of bacterial colonization presents a risk for development of antibiotic resistance, and highlights the importance of improved rapid diagnostics to rule out viral etiologies to aid clinical management. in addition, it points to the need for increased antibiotic stewardship and health-care provider education to reduce unnecessary use of antibiotics in military settings. our study also emphasizes the importance of systematic microbiological surveillance of uri in military populations, to allow prompt detection and control of outbreaks and early identification of emerging pathogens that may be circulating or could be introduced in the wider population. this is particularly important as infection control options in such closed settings are limited and vaccination coverage for key pathogens is low. limited resources meant it was not possible to conduct multiplex diagnostics for all baseline, acute, and end of followup samples, as well as influenza serology. as a result, there was limited power to investigate whether colonization influenced risk of illness, or factors associated with colonization and acquisition of bacteria. in addition, co-circulation of multiple pathogens makes it difficult to establish chains of transmission, which would require more detailed microbial sequencing information. clinical signs and symptoms were available at presentation only, and we did not have information regarding the duration of individual symptoms or additional clinical investigations conducted longitudinally during the course of illness, as these were not routinely performed. the specific barracks setting and population characteristics limit the generalizability of our findings to other military settings with similar conditions. nevertheless, our study provides key data regarding the epidemiology and transmission of respiratory pathogens in military populations in tropical settings. our results emphasize the need for improved infection prevention and control strategies in military settings, given the potential for intense transmission of a wide range of respiratory pathogens. financial support: this work was supported by the united states department of defense-global emerging infectious disease surveillance (dod-geis), protocol 989a. disclosures: the authors declare that they have no competing interests. material has been reviewed by the walter reed army institute of research. there is no objection to its presentation and/or publication. the opinions or assertions contained herein are the private views of the author, and are not to be construed as official, or as reflecting true views of the department of the army or the department of defense. the investigators have adhered to the policies for protection of human subjects as prescribed in ar 70-25. respiratory tract infections in the military environment environmental factors, immune changes and respiratory diseases in troops during military activities 2009 influenza a(h1n1) seroconversion rates and risk factors among distinct adult cohorts in singapore epidemiology and transmission characteristics of human adenovirus type 7 caused acute respiratory disease outbreak in military trainees in east china febrile respiratory illness associated with human adenovirus type 55 in south korea military outbreak of acute respiratory disease caused by human adenovirus type 7 in a military training camp in shaanxi pertussis outbreak among soldiers during basic training: the need for updated protocols acute respiratory disease in us army trainees 3 years after reintroduction of adenovirus vaccine the increasing application of multiplex nucleic acid detection tests to the diagnosis of syndromic infections advances in multiparametric molecular diagnostics technologies for respiratory tract infections elevated transmission of upper respiratory illness among new recruits in military barracks in thailand cdc protocol of realtime rtpcr for influenza a(h1n1) cdc realtime rt-pcr (rrtpcr) protocol for detection and characterization of influenza evaluation of four commercial multiplex molecular tests for the diagnosis of acute respiratory infections comparison of fast-track diagnostics respiratory pathogens multiplex real-time rt-pcr assay with in-house singleplex assays for comprehensive detection of human respiratory viruses reduction in acute respiratory infection among military trainees: secondary effects of a hygiene-based cluster-randomized trial for skin and soft-tissue infection prevention comparison of the effectiveness of trivalent inactivated influenza vaccine and live, attenuated influenza vaccine in preventing influenza-like illness among us military service members handwashing and respiratory illness among young adults in military training influenza activity in thailand and occurrence in different climates the role of serum haemagglutination-inhibiting antibody in protection against challenge infection with influenza a2 and b viruses determinants of immunity to influenza infection in man estimation of the association between antibody titers and protection against confirmed influenza virus infection in children relationship between haemagglutination-inhibiting antibody titres and clinical protection against influenza: development and application of a bayesian random-effects model haemophilus influenzae infections in the h. influenzae type b conjugate vaccine era acknowledgments: we are grateful to the military recruits who participated in the study, the royal thai army, and the clinical, laboratory, and administrative personnel at afrims for their support. this is an open-access article distributed under the terms of the creative commons attribution license, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. key: cord-254556-1zthrgy1 authors: taylor, sylvia; lopez, pio; weckx, lily; borja-tabora, charissa; ulloa-gutierrez, rolando; lazcano-ponce, eduardo; kerdpanich, angkool; angel rodriguez weber, miguel; mascareñas de los santos, abiel; tinoco, juan-carlos; safadi, marco aurelio p.; lim, fong seng; hernandez-de mezerville, marcela; faingezicht, idis; cruz-valdez, aurelio; feng, yang; li, ping; durviaux, serge; haars, gerco; roy-ghanta, sumita; vaughn, david w.; nolan, terry title: respiratory viruses and influenza-like illness: epidemiology and outcomes in children aged 6 months to 10 years in a multi-country population sample date: 2016-09-22 journal: j infect doi: 10.1016/j.jinf.2016.09.003 sha: doc_id: 254556 cord_uid: 1zthrgy1 background: better population data on respiratory viruses in children in tropical and southern hemisphere countries is needed. methods: the epidemiology of respiratory viruses among healthy children (6 months to <10 years) with influenza-like illness (ili) was determined in a population sample derived from an influenza vaccine trial (nct01051661) in 17 centers in eight countries (australia, south east asia and latin america). active surveillance for ili was conducted for approximately 1 year (between february 2010 and august 2011), with pcr analysis of nasal and throat swabs. results: 6266 children were included, of whom 2421 experienced 3717 ili episodes. rhinovirus/enterovirus had the highest prevalence (41.5%), followed by influenza (15.8%), adenovirus (9.8%), parainfluenza and respiratory syncytial virus (rsv) (both 9.7%), coronavirus (5.6%), human metapneumovirus (5.5%) and human bocavirus (hbov) (2.0%). corresponding incidence per 100 person-years was 29.78, 11.34, 7.03, 6.96, 6.94, 4.00, 3.98 and 1.41. except for influenza, respiratory virus prevalence declined with age. the incidence of medically-attended ili associated with viral infection ranged from 1.03 (hbov) to 23.69 (rhinovirus/enterovirus). the percentage of children missing school or daycare ranged from 21.4% (hbov) to 52.1% (influenza). conclusions: active surveillance of healthy children provided evidence of respiratory illness burden associated with several viruses, with a substantial burden in older children. respiratory viruses and influenza-like illness: epidemiology and outcomes in children aged 6 months to 10 years in a multi-country population sample introduction acute respiratory tract infections (artis) comprise the most common illnesses worldwide, and children often experience several episodes a year. 1 in children, clinical presentation can range from mild, uncomplicated upper respiratory tract illness to severe lower respiratory tract infection (lrti) including pneumonia, bronchiolitis, croup and exacerbations of asthma or wheezing. 2 the world health organization estimated that 1.9 million children died from arti in 2000, 70% in africa and south east asia. 3 in 2011, pneumonia alone led to 1.3 million deaths worldwide in children less than 5 years of age. 4 most artis are caused by viruses, but until the advent of multiplex polymerase chain reaction (pcr) techniques, it was often not possible to identify accurately specific viral infections in clinical cases. 5, 6 the pathogens considered responsible for most arti are respiratory syncytial virus (rsv), influenza a and b, parainfluenza viruses types 1, 2 and 3, and adenovirus. 2 several new pathogens associated with arti have been identified more recently, including human metapneumovirus (hmpv), rhinovirus, coronavirus, human bocavirus (hbov) and parainfluenza 4. 2, 7 vaccine efficacy trials provide intensive, active followup of a well-defined population and can be used to evaluate viral epidemiology. as part of a trial of pandemic influenza vaccines, which included 1 year of prospective, active, community-based surveillance for influenza-like illness (ili) in 17 centers in eight countries, 8 we evaluated the prevalence and incidence of respiratory viruses in children 6 months to less than 10 years of age at first vaccination. samples were obtained from an efficacy trial of two pandemic influenza h1n1 vaccines (nct01051661 sponsored by gsk vaccines). 8 an analysis estimating the prevalence of rsv in ili has been previously reported. 9 here, we report data on all respiratory viruses evaluated, a secondary analysis objective. healthy children 6 months to <10 years of age were enrolled in a randomized, observer-blind, parallel group, multi-country trial of as03-adjuvanted versus nonadjuvanted monovalent pandemic h1n1 vaccines. 8 the trial was conducted in 17 centers in australia, brazil, colombia, costa rica, mexico, the philippines, singapore, and thailand between 15 february 2010 and 19 august 2011; enrollment took up to 6 months and timing varied by country. the trial was approved by an institutional review board for each center and written informed consent was obtained from parents/guardians. parents were instructed to contact the study center within 24 h if the child became ill. active surveillance via scripted telephone contact was conducted from 2 weeks after first vaccination, and contact made every 1e2 weeks to day 385 for each child, regardless of time of enrollment. ili was defined as temperature !38.0 c by any route and at least one of: new/ worsening cough, sore throat, stuffy nose, or runny nose. study staff visited the child's home to collect one anterior nasal swab and one throat swab, ideally within 24 h of ili onset, and at most within 7 days. collection could also take place at a study center or hospital if necessary. swabs were transported in a single tube of m4rt transport medium, stored at à70 c and maintained on dry ice during transport. a 7-day symptom-free period was required between new ili episodes. sample testing was performed by standard multiplex pcr techniques. 8, 10 analysis of viral epidemiology the viruses evaluated were influenza (subtypes a-h1, a-h3 and b), parainfluenza (subtypes 1, 2, 3 and 4), rsv (subtypes a and b), hmpv, rhinovirus/enterovirus (the assay did not distinguish between them), adenovirus, coronavirus (subtypes 229e, oc43, nl63 and hku1) and hbov. first analyzed separately, influenza, parainfluenza and coronavirus subtypes were grouped in a post-hoc analysis. the main outcome variable was pcr-confirmed infection with the stated viruses in nasal/throat swabs in children with ili. this included infection with the virus under consideration alone (single infection) or with the virus under consideration plus one or more of the other viruses (co-infection). single and co-infections were also recorded separately. clinical characteristics of ili episodes were reported by the parents of the children, and any hospitalization or medical attendance (by a doctor or other healthcare professional, not including sample collection by study staff) were recorded. pneumonia was defined as acute illness (one or more of fever !38 c, new or worsening cough, dyspnea, consistent auscultation findings [rales or diminished breath sounds], pain in the chest or abdomen when breathing, or purulent or blood-stained sputum production) and radiologic findings consistent with pneumonia. the total cohort included all children enrolled in the randomized trial. the total cohort with ili episodes tested by multiplex pcr included all children enrolled who experienced an ili and had an adequate nasal/throat sample tested by multiplex pcr. the analysis of prevalence of respiratory viruses in ili and clinical characteristics associated with ili was performed in the total cohort with ili episodes tested by multiplex pcr. the analysis of overall incidence of ili, medically-attended ili and hospitalized ili in which respiratory viruses were detected was performed in the total cohort. the prevalence of respiratory viruses among ili episodes was calculated as: where x is the number of ili episodes with nasal/throat samples positive for the virus and n is the total number of ili episodes with samples collected within 7 days and tested. as there was at least 7 days between two ili episodes, it was assumed that each episode was independent. exact 95% confidence intervals (ci) were computed. 11 prevalence was stratified according to country, age at the time of the ili episode (6e11, 12e23, 24e35, 36e59, 60þ months) and whether the child was medically attended or hospitalized. the incidence per 100 person-years (py) of virusassociated ili in the study population was calculated as: where n is the total number of children enrolled in the trial, ε i is the total number of virus-positive ili episodes for subject i, and d i is the follow-up period for subject i. incidence rates were stratified according to country and age group at the time of the ili episode. exact 95% poisson cis were calculated. 12 observations with incomplete data for the outcome variable and ili episodes for which no nasal/ throat sample was taken were removed from the analysis. missing data were accounted for by calculating the missing proportion for each country and age group, then multiplying the py by (1 minus missing proportion). the trial included 6266 children (total cohort). after excluding children with no ili or inadequate samples, 2421 children experienced 3717 ili episodes (total cohort with ili episodes tested by multiplex pcr). participant flow is shown in supplement fig. 1 . demographics were similar in both cohorts, except that children in the total cohort were older (median 55 versus 42 months) (supplement table 1 ). a respiratory virus was detected in 2958 of 3717 ili episodes (79.6%). rhinovirus/enterovirus had the highest overall prevalence (41.5%), followed by influenza (15.8%), adenovirus (9.8%), parainfluenza and rsv (both 9.7%), coronavirus (5.6%), hmpv (5.5%) and hbov (2.0%) ( table 3 ]). co-infection was detected more often with adenovirus and hbov (fig. 1 ). single infections with parainfluenza, hmpv and coronavirus were identified at approximately the same frequency as co-infections ( fig. 1 ). however, parainfluenza 4 and coronavirus 229e were identified more often as co-infections, whilst coronavirus nl63 was identified more often as a single infection (supplement table 3 ). in all countries, rhinovirus/enterovirus was the most prevalent (36.9e59.2%), whilst hbov was least prevalent (0.8e4.7%) ( table 1) . influenza prevalence ranged from 6.1% to 18.5%; parainfluenza prevalence was approximately 10% except in brazil (5.8%) and singapore (6.1%) ( table 1) . (table 1) . influenza was most prevalent (21.3%) in the oldest children (60þ months), followed by 36e59 months (15.6%) and the other age groups (9.8e12.3%) ( table 2 ). all other viruses were least prevalent in the oldest group (table 2) . there was a less obvious pattern in the other age groups, but, in general, prevalence declined with age except for influenza ( table 2) . the incidence of detected respiratory viruses associated with ili reflected their prevalence. the overall incidence per 100 py (total cohort, all children randomized) was 29.78 for rhinovirus/enterovirus, 11.34 for influenza, 7.03 for adenovirus, 6.96 for parainfluenza, 6.94 for rsv, 4.00 for coronavirus, 3.98 for hmpv and 1.41 for hbov (table 3) . australia had the highest incidence of hmpv (5.08) and the second highest of rsv (7.03), but low incidence of the other viruses relative to other countries ( table 3 ). the philippines, singapore and thailand also had low incidences of most viruses in ili relative to the latin american countries ( table 3) . detection of the respiratory viruses at different times during the year was highly variable across countries (fig. 2aeh) . the overall incidence of medically-attended ili associated with viral infection per 100 py (total cohort, all children randomized) ranged from 1.03 for hbov to 23.69 for rhinovirus/enterovirus (table 3) . corresponding values for incidence of hospitalized ili associated with viral infection were 0 for hbov and 0.81 for rhinovirus/enterovirus (table 3) . clinical characteristics of ili episodes associated with a single respiratory virus (i.e. no co-infection) are shown in table 4 . median duration of ili episodes ranged from 8.9 to 13.4 days. few children were hospitalized but most were medically attended outside study procedures. the percentage of children missing school or daycare was highest with influenza-associated ili (52.1%), followed by hmpv (41.5%), adenovirus (39.0%), rhinovirus/enterovirus (37.6%), coronavirus (31.1%), rsv (30.2%), parainfluenza (28.1%) and hbov (21.4%) ( table 4 ). sore throat was experienced by 25e52% of children, cough by 62e97%, stuffy nose by 40e62%, and runny nose by 66e84% (table 4 ). fever was part of the ili definition and therefore experienced by all children. cough was reported in almost all children with influenza, parainfluenza, rsv, hmpv and coronavirus infections, but only in 60e70% of children with rhinovirus/ enterovirus, adenovirus and hbov infections. there were no medically important differences in clinical characteristics between children with a single viral infection compared with children with multiple infections (supplement table 4 ). a total of 58 pneumonia cases were identified among the 6266 children enrolled in the overall clinical trial, corresponding to a detection rate of 0.9%. of the 58 cases, 32 met the definition of ili and were therefore eligible for sample collection as per the clinical trial protocol. a sample was collected within 7 days of onset of ili symptoms for 20 of these 32 cases: one case in thailand, three in the philippines, five in brazil, five in mexico and six in colombia (table 5) . no virus was detected in three cases, a single infection was detected in 10 cases (four rhinovirus/enterovirus, two parainfluenza, one influenza, two rsv and one hmpv), and co-infection was detected in seven cases (table 5) . nine children were hospitalized. rhinovirus/enterovirus had the highest prevalence and incidence in ili of all respiratory viruses tested in all countries, followed by influenza, adenovirus, parainfluenza and rsv, coronavirus, hmpv and hbov. the burden of ili associated with respiratory viruses was considerable, with a high proportion of children being seen by a medical professional and many missing school or daycare. our analysis benefited from being part of a clinical trial, as previously described. 9 most importantly, we conducted 1 year of prospective, active community surveillance of healthy children in tropical and southern hemisphere countries where prospective data are lacking. most studies of viral epidemiology use hospital-based surveillance because community-based surveillance is difficult and expensive. however, hospital-based surveillance tends to capture only the most severe illness and many cases are missed in developing countries because of limited hospital access. our analysis avoided these limitations and allowed us to capture the burden of virus-associated ili in communities. understanding community epidemiology is essential to implement effective control measures. other advantages of being part of a clinical trial included a wellcharacterized population, wide age range up to 10 years, samples taken from a high proportion of children, consistent methodology between countries, and use of sensitive and validated pcr assays. the trial was conducted in eight countries encompassing australia, south east asia and latin america. the exact timing of enrollment varied somewhat between countries, but was planned so that data collection was performed during the peak 2010e2011 influenza season for each individual country. as stated in the methods, all children were followed for 385 days, with the complete period of surveillance for the study occurring between 15 february 2010 and 19 august 2011. this allowed us to compare the distribution of viruses across the different countries. there was considerable variation in the incidence and prevalence of the viruses by country, although rhinovirus/enterovirus had by far the highest incidence and prevalence in all countries. hbov had consistently the lowest incidence and prevalence. several other studies have evaluated the prevalence of viruses in children with respiratory illness. the relative prevalence of the different circulating viruses varied by study. however, the main circulating viruses were similar between studies and with our study, and included picornaviruses (including rhinovirus), adenovirus, rsv, bocavirus, pivs, hmpv, influenza and coronavirus. 13e17 rhinoviruses are classified in the picornavirus family, of the enterovirus genus. 7 a high prevalence of this family has been reported in other studies in different settings. 13,14,18e20 as in our study, an australian study with active community-based surveillance of healthy preschool-age children with arti found that picornaviruses (including rhinoviruses) were the most frequently detected (41.3%). 13 however, other viruses were detected less frequently than in our study: rsv (6.6%), parainfluenza (4.1%), influenza a and hmpv (both 3.7%), adenovirus (3.1%) and coronavirus nl63 (1.5%). 13 in another prospective australian study in children aged 6 months to 3 years reporting ili, rhinovirus was again the most commonly detected. 15 however, in contrast to our results, adenovirus was detected at the same frequency as rhinovirus, followed by parainfluenza 3, polyomavirus, hmpv and hbov. 15 influenza (a/h1n1) and rsv were relatively uncommon; approximately 40% of children were fully or partially vaccinated against influenza. rhinovirus is not always the most commonly detected virus in children with respiratory disease. in children under 5 years of age hospitalized for lrti in thailand, the most commonly detected viruses were rsv (19.5%), rhinovirus (18.7%), hbov (12.8%) and influenza (8.2%). 16 a study of children aged <3 years hospitalized for lrti in brazil found that rsv was most prevalent (53.5% of episodes), followed by hmpv (32.3%), rhinovirus (20.8%), influenza (12.7%), hbov (10.4%), parainfluenza and adenovirus (both 6.5%) and coronavirus (1.2%). 17 in our analysis, influenza prevalence increased with age. the other viruses showed the opposite trend, with the lowest prevalence observed in the oldest children (60þ months). there was a less obvious pattern in younger ages, but, in general, prevalence of all viruses except influenza declined with age. despite this, the burden of illness remained considerable in older children. there was a clear seasonal pattern for influenza, rsv and hmpv in most countries, and to a lesser extent for rhinovirus/ enterovirus. a previous study found that, although there was no clear seasonal peak for rhinovirus/enterovirus, onset seemed to correspond with the start of the school year in the usa. 18 a limited one year analysis of human rhinoviruses and enteroviruses in ili in latin america showed a year-round temporal distribution throughout central and south america. 21 however, human rhinovirus c species displayed opposite seasonal trends on either side of the equator, accounting for a higher percentage of ili cases north of the equator between september and january, while south of the equator detection increased between april and july. 21 as part of the study, all children received a monovalent influenza a/h1n1 pandemic vaccine; one or two doses of an as03-adjuvanted vaccine were administered or two doses of an unadjuvanted vaccine. trivalent seasonal influenza vaccination rate in the present study was approximately 18%. influenza a subtype h1 was not isolated in any children; influenza a subtype h3 was isolated in 9.0% of children; and influenza b was isolated in 5.7% of children. no difference between the study vaccine groups was observed. cases associated with influenza were least likely to be co-infected with other respiratory viruses. rhinovirus/ enterovirus was also more common as a single infection. adenovirus and hbov were found more often as a coinfection. bacterial co-infection was not measured as part of this study. in the us-based influenza incidence surveillance project, which evaluated the most commonly detected viruses in outpatients with arti or ili, threequarters of all co-infections involved adenovirus and rhinovirus/enterovirus. 18 a uk-based analysis found negative associations between influenza a and hmpv, and between influenza a and rhinovirus. 22 positive associations were found between parainfluenza and rhinovirus, rsv and rhinovirus, adenovirus and rhinovirus, and parainfluenza and rsv. 22 no correlation was found between co-infection and clinical severity in a study in brazil evaluating children under 5 years who sought medical care for respiratory tract infections. 23 more research is needed to understand the interaction of respiratory viruses, and the host response to infection. there were no clear differences between viruses in the severity of illness. most ili episodes were medically attended. ili associated with influenza resulted in the highest proportion of children missing school or daycare (52%), although 20e40% of children infected with the other viruses also missed school or daycare. there was no difference between viruses in the proportion of children hospitalized. clinical features were variable depending upon the viral infection associated with the ili episode. study limitations have been described previously. 9 only healthy children participated in the trial, limiting generalizability. in addition, our study did not include any children aged <6 months and only a limited number of children aged 6e11 months, so our findings are mainly relevant to older children. fever was part of the ili definition, and therefore we would have missed cases in children with no fever. to put this into perspective, in the influenza incidence surveillance project, 34% and 43% of cases among children aged 1e4 years and 5e17 years, respectively, met the arti definition which did not require fever, but did not meet the ili definition which did require fever. 18 however, our definition of ili was somewhat broader (except in children under 2 years of age) and us data may not be generalizable to the tropical and southern hemisphere countries in our study. the inclusion of only healthy children in the study and the exclusion of cases with no fever would have underestimated the burden. we also could not discriminate between rhinovirus and enterovirus by pcr, therefore the exact prevalence and incidence of each one could not be determined. finally, our study included only a small number of pneumonia cases (n z 20), limiting the conclusions that can be drawn regarding the distribution of viral infection in these cases. the overall pneumonia detection rate in the clinical trial (0.9%) is higher than, but in line with, what has been reported in the us for hospitalized cases. 24 however, our sample collection rate among pneumonia cases was only 62.5% compared with 80.0% for ili overall. in conclusion, our active surveillance of healthy children as part of a vaccine efficacy trial provided evidence of the burden of respiratory illness associated with a range of viruses. a substantial burden of illness occurs in older children. data on the epidemiology of respiratory viruses determined from active surveillance of healthy children are generally lacking, and are particularly sparse in the developing countries included in our study. a considerable amount of the burden would not be identified through hospital-based surveillance. these novel data fill an important gap in our knowledge of the epidemiology of viruses contributing to the substantial burden of respiratory disease in children, and may be useful in informing priorities for implementation of existing vaccine programs and development of new vaccines. this work was supported by glaxosmithkline biologicals s.a. who was the sponsor of the study and was involved in all stages of study conduct, including analysis of the data, and in addition paid the costs related to the development of the publication of this manuscript. epidemiology of viral respiratory infections viral infections of the lower respiratory tract: old viruses, new viruses, and the role of diagnosis estimates of world-wide distribution of child deaths from acute respiratory infections global burden of childhood pneumonia and diarrhoea comparison of multiplex pcr assays and conventional techniques for the diagnostic of respiratory virus infections in children admitted to hospital with an acute respiratory illness molecular diagnosis of respiratory virus infections the role of infections and coinfections with newly identified and emerging respiratory viruses in children relative efficacy of as03-adjuvanted pandemic h1n1 influenza vaccine in children: results of a controlled, randomized efficacy trial prevalence and incidence of respiratory syncytial virus and other respiratory viral infections in children 6 months to 10 years of age with influenza-like illness enrolled in a randomized trial comparison of the luminex xtag respiratory viral panel with xtag respiratory viral panel fast for diagnosis of respiratory virus infection the use of confidence or fiducial limits illustrated in the case of the binomial a simple method to calculate the confidence interval of a standardized mortality ratio (smr) community epidemiology of human metapneumovirus, human coronavirus nl63, and other respiratory viruses in healthy preschool-aged children using parentcollected specimens clinical epidemiology of bocavirus, rhinovirus, two polyomaviruses and four coronaviruses in hiv-infected and hiv-uninfected south african children epidemiology of respiratory viral infections in children enrolled in a study of influenza vaccine effectiveness incidence and etiology of acute lower respiratory tract infections in hospitalized children younger than 5 years in rural thailand severe lower respiratory tract infection in infants and toddlers from a non-affluent population: viral etiology and co-detection as risk factors viruses associated with acute respiratory infections and influenza-like illness among outpatients from the influenza incidence surveillance project picornavirus, the most common respiratory virus causing infection among patients of all ages hospitalized with acute respiratory illness respiratory viruses within homeless shelters in marseille, france human rhinoviruses and enteroviruses in influenza-like illness in latin america respiratory viral infections during the 2009e2010 winter season in central england, uk: incidence and patterns of multiple virus co-infections concurrent detection of other respiratory viruses in children shedding viable human respiratory syncytial virus community-acquired pneumonia requiring hospitalization among us children all authors participated in the design, or implementation, or analysis and interpretation of the study results; as well as in the development of this manuscript. all authors had full access to the data and gave final approval before submission.terry nolan, charissa borja-tabora, pio lopez, lily weckx, rolando ulloa-gutierrez, eduardo lazcano-ponce, angkool kerdpanich, miguel angel rodriguez weber, abiel mascareñas de los santos, marco aurelio p safadi, aurelio cruz-valdez and juan-carlos tinoco were coordinating investigators, and together with sumita roy-ghanta, david w vaughn and ping li were responsible for the conduct of the flu q-pan h1n1-035 pri (nct01051661) trial. fong seng lim, marcela hernandez-de mezerville and idis faingezicht also contributed to study material and data collection.sylvia taylor led the epidemiology team in collaboration with gerco haars.yang feng was responsible for the statistical input; statistical expertise was also provided by gerco haars, sumita roy-ghanta, ping li and terry nolan. serge durviaux led the laboratory analysis.terry nolan, charissa borja-tabora, yang feng, david w vaughn and sylvia taylor were members of the core writing team. terry nolan and sylvia taylor contributed equally to this manuscript and the corresponding author was responsible for the submission of the publication. the the authors moreover thank magali ribot (gsk vaccines) for multiplex pcr testing, jean-yves pirçon (gsk vaccines) for critical review of the statistical analysis and mich ele seil (keyrus biopharma on behalf of gsk vaccines) for writing the study report.mary greenacre (independent medical writer) was paid by gsk vaccines to prepare the manuscript draft and sophie vanwetswinkel (xpe pharma and science on behalf of gsk vaccines) provided editorial assistance and manuscript coordination. lily weckx declares research grants from gsk to federal university of são paulo for conduct of three clinical trials and received payment from gsk, novartis, pfizer, and sanofi for board membership or lectures.rolando ulloa-gutierrez discloses having received honoraria from gsk for the original influenza a h1n1 clinical trial discussed here, as well as from gsk, sanofi pasteur, pfizer/ wyeth and merck as a speaker in the past.marco aurelio p safadi has received grants to support research projects and consultancy fees from novartis, gsk, pfizer and sanofi pasteur.fong seng lim discloses having received travel grants from gsk as well as a grant from gsk to his institution to perform clinical trials.marcela hernandez-de mezerville declares having received honoraria from gsk for the original influenza a/ h1n1 clinical trial discussed here, as well as travel support from gsk, sanofi pasteur and pfizer outside the submitted work in the past.idis faingezicht received payment from gsk as principal investigator in a previous vaccine clinical trial and as co-investigator in the influenza a/h1n1 clinical trial.pio lopez, eduardo lazcano-ponce, angkool kerdpanich, miguel angel rodriguez weber, abiel mascareñas de los santos, juan-carlos tinoco, and aurelio cruz-valdez report having nothing to disclose. supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jinf.2016.09.003. key: cord-261282-r1nprlne authors: chughtai, a. a.; wang, q.; dung, t. c.; macintyre, c. r. title: the presence of fever in adults with influenza and other viral respiratory infections date: 2016-10-03 journal: epidemiol infect doi: 10.1017/s0950268816002181 sha: doc_id: 261282 cord_uid: r1nprlne we compared the rates of fever in adult subjects with laboratory-confirmed influenza and other respiratory viruses and examined the factors that predict fever in adults. symptom data on 158 healthcare workers (hcws) with a laboratory-confirmed respiratory virus infection were collected using standardized data collection forms from three separate studies. overall, the rate of fever in confirmed viral respiratory infections in adult hcws was 23·4% (37/158). rates varied by virus: human rhinovirus (25·3%, 19/75), influenza a virus (30%, 3/10), coronavirus (28·6%, 2/7), human metapneumovirus (28·6%, 2/7), respiratory syncytial virus (14·3%, 4/28) and parainfluenza virus (8·3%, 1/12). smoking [relative risk (rr) 4·65, 95% confidence interval (ci) 1·33–16·25] and co-infection with two or more viruses (rr 4·19, 95% ci 1·21–14·52) were significant predictors of fever. fever is less common in adults with confirmed viral respiratory infections, including influenza, than described in children. more than 75% of adults with a viral respiratory infection do not have fever, which is an important finding for clinical triage of adult patients with respiratory infections. the accepted definition of ‘influenza-like illness’ includes fever and may be insensitive for surveillance when high case-finding is required. a more sensitive case definition could be used to identify adult cases, particularly in event of an emerging viral infection. respiratory infections are common and one of the leading causes of morbidity and mortality, particularly in the extremes of age [1] [2] [3] . influenza a and b, human rhinoviruses (hrv), respiratory syncytial virus (rsv), adenoviruses (adv) and parainfluenza virus (piv) are common respiratory viruses in adults and children [1] [2] [3] [4] [5] . of respiratory infections, influenza is the most well studied viral infection, and is commonly reported (around 50%) as the cause of epidemics of respiratory infection, including nosocomial outbreaks [6] . influenza virus is commonly isolated from febrile paediatric and elderly patients presenting with influenza-like illness (ili) and acute respiratory illness (ari) symptoms [1] . the accepted clinical case definition of ili includes fever, which may be suitable for identifying paediatric cases, but less so for adults. fever is thought of as the most common presenting symptom of influenza in hospital emergency departments; however, the presence of fever depends on the age of person and the type of virus [7] [8] [9] [10] . it is known that fever is less common in adults than children with influenza, and that adults may have atypical presentations [5, 6, 11] . in a matched case-control study in finland, 317 laboratoryconfirmed influenza cases and 353 controls with respiratory symptoms were recruited in children aged 413 years. fever was present in 89·8% (317/353) and 35·7% (126/353) of cases and controls, respectively [12] . in contrast to this, fever is not a common presentation in adults with laboratory-confirmed influenza. monto et al. [13] examined clinical trial data of 3744 adult ili cases (defined as body temperature 537·8°c or patients subjective feeling of feverishness) and of those 2470 (66%) had laboratory-confirmed influenza. fever (537·8°c) was reported in 68% of laboratory-confirmed influenza cases, compared to 40% other ili cases [13] . during a randomized clinical trial (rct) around the efficacy of facemask and hand hygiene in the household setting, 44% (15/34) of secondary cases with influenza a and 32% (8/ 25) of cases with influenza b had fever history [14] . the rate of fever was 66% (137/207) in hospitalized influenza cases in a us study [15] . another us study showed that less than half (42·4%) of healthcare workers (hcws) with laboratory-confirmed influenza presented with fever [5] . fever is a less common presenting symptom in elderly people which may lead to diagnostic and treatment delays [16] . in patients admitted with myocardial infarction, 9% had unrecognized and undiagnosed influenza on testing at admission, highlighting the low level of clinical suspicion of influenza [17] . the rate of fever also varies between influenza strains, being more common in influenza a strains than b, and higher in h3n2 [7] [8] [9] [10] 18] . fever is a commonly reported symptom during influenza outbreaks and pandemics due to novel and more virulent nature of strains. in china 67·4% of the patients infected by influenza a(h1n1)pdm09 had fever [19] . in another study in beijing 465 suspected ili cases were tested and of those 318 (68%) were positive for influenza virus (pandemic h1n1-165 and seasonal influenza h3n2-153) and all had history of fever [20] . the aim of this study was to compare the rates of fever in adult subjects with confirmed influenza and other respiratory virus infections and examine predictors of fever. we analysed a dataset of laboratory-confirmed viral respiratory infections collected from three clinical trials of hcws where active surveillance for respiratory viral illness was conducted in prospective follow up [21] [22] [23] . the same methods, data collection forms and outcome measures, were used across the three studies, allowing the data to be pooled [21] [22] [23] . two studies were conducted in beijing china: trial 1 (2008/2009) and trial 2 (2009/2010) and another study (trial 3) was conducted in hanoi, vietnam in 2010/2011 [21] [22] [23] . in all clinical trials, participants were asked to complete diary cards on a daily basis to collect information on number of working hours, patients seen, mask use hours, high-risk procedures performed and appearance of respiratory symptoms. thermometers were given the participants to measure their temperature daily and at symptom onset. symptomatic cases were asked to complete sick patient follow-up forms and detailed information was collected on the following symptoms: chill or fever, cough, congestion, runny nose, sore throat, sneezes, lethargy, loss of appetite, abdominal pain, muscle or joint aches. swabs of both tonsils and the posterior pharyngeal wall were collected on the day of reporting. in all rcts, fever was defined as having body temperature 538°c. clinical respiratory illness (cri) and ili were in the primary outcomes in three clinical trials. cri was defined as two or more respiratory symptoms or one respiratory symptom and a systemic symptom and ili was defined as fever 538°c plus one respiratory symptom [21] [22] [23] . we analysed data from all subjects with a positive isolation of a respiratory virus by multiplex polymerase chain reaction (pcr). descriptive analysis was conducted for rates of fever by virus type. a logistic regression analysis was used to determine the predictors of fever. a multivariable log binomial model was fitted, using a generalized linear model to estimate relative risk (rr). all variables were included in initial model. in the final model, we included only those variables that were significant (p < 0·25) in initial analysis. a backward elimination method was used to remove the variables that did not have any confounding effect, i.e. could not make meaningful change (±10%) in the rr of the comparison arm. finally we estimated the rates of cri and ili in the laboratory-confirmed viral respiratory infections and laboratory-confirmed influenza infections. the data was analysed using sas v. 9.4 (sas institute inc., usa). the demographic characteristics of 158 cases with laboratory-confirmed viral infections are presented in table 1 . ninety (57%) cases were from china and 68 (43%) were from vietnam. the mean age of hcws was 32·8 years and most participants were nurses (65%) and female (87%). most cases were non-smokers (92%) and had not received influenza vaccine (86%). viruses isolated included rhinovirus (n = 75, 47%), rsv (n = 28, 18%), influenza (n = 13, 8%), piv (n = 12, 8%), human metapneumovirus (hmpv; n = 7, 4%), coronavirus (n = 7, 4%) and adv (n = 1, 1%). more than one virus was isolated in 15 cases (9·5%), including nine cases with influenza co-infection. fever was documented in 23·4% cases (37/158) with a positive laboratory viral diagnosis. table 2 details rates of fever (538°c) associated with individual viral respiratory infections. hrv was the most common infection and 25·3% (19/75) of these had a fever. in 28 cases of rsv, four (14·3%) had fever; 8·3% (1/12) of piv and 30% (3/10) of influenza a cases had fever. seven cases of coronavirus and hmpv each were confirmed and of those two (28·6%) had fever. when cases with influenza and a co-infection were included, 36·4% (8/22) had fever. in univariate analysis, country, gender and smoking were significant predictors of fever. country and smoking remained significant predictors in multivariate analysis while gender became non-significant. fever rate was significantly higher in hcws in vietnam compared to hcws in china [rr 2·99, 95% confidence interval (ci) 1·24-7·20]. smokers were around five times more likely to have fever compared to non-smokers (rr 4·65, 95% ci 1·33-16·25). virus type was not associated with fever in univariate analysis; however, after adjusting for other variables, rates of fever were significantly higher in hcws co-infected with more than one virus compared to all other viruses excluding influenza (rr 4·19, 95% ci 1·21-14·52) ( table 3) . cri symptoms were present in 84·8% (137/158) of hcws with laboratory-confirmed viral infections and 90·9% (20/22) laboratory-confirmed influenza infections. the corresponding rates of ili in the two groups were 9·5% (15/158) and 13·6% (3/22), respectively. we have shown, using prospectively collected data, that the rate of fever in adults with confirmed viral respiratory infections is much lower than described in children [1, 9] . the standard clinical case definition of ili requires fever to be presentthe majority of influenza cases in this series would have been missed using the ili definition. this has implications for effective triage, early antiviral treatment and preventive measures for adults with influenza, particularly during outbreaks and pandemic situations. for other respiratory infections, clinical case definitions need to be more sensitive, or >75% of cases will be missed. the main implication for future surveillance, measurements and research studies is that the ili case definition in adults may be highly insensitive. for some types of surveillance systems, this may not be an issue, but for diagnostic screening in event of an emerging viral infection (such as for triage and implementation of infection control protocols) [24, 25] , a more sensitive case definition is needed. rates of fever in influenza and other viral respiratory infections in this study were lower compared to other studies which report fever in around 50-70% adult cases [1, 5, 13, 15] . however, this variation may be due to different study base, case definition and viral strains, as well as the prospective measurement of incident infections. many research studies use fever as an inclusion criterion for laboratory testing [13, 26] . while this may be suitable for studies in children, it is not adequately sensitive for studies of adults, as we have shown the majority of confirmed cases will be missed. the cut-off point for fever could be another factor in sensitivity. some studies have set lower cut-off points for fever, and report higher rates of fever in laboratory-confirmed influenza cases [13, 27] . carrat et al. collected data of cases presented in 35 general practices in france and collected nasal swabs from suspected influenza cases and defined fever as 537·8°c. they found fever in influenza a(h3n2), influenza a(h1n1) and influenza negative cases in 95·2%, 77·5% and 72·7%, respectively. applying a cut-off of 538·2°c, the corresponding rates are 82·2%, 59·3% and 43·9% [27] . symptoms of feverishness (subjective feeling of fever) are included in ili definitions in some cases [13] . previous studies report high rates of fever in children compared to the adults [11] . low rates of fever in adults may also be due to protection via cross-reactive antibodies due to age-dependent differences in the immunity [7, 28] . continued exposure to influenza throughout life may result in a broader protection with age. infection may provoke a stronger immune response in children with minimal to no exposure history compared to adults. therefore influenza infection history might help explain potential differences in clinical symptom severity (and presence of fever) between children and adults. a recent study reported high rates of influenza and other respiratory virus in afebrile hcws with only respiratory symptoms [5] . of 22 laboratory-confirmed influenza cases in this study, only three (13·6%) had ili symptoms, which is very low compared to other studies. in a prospective influenza surveillance study, ili symptoms were present in 48% of adults and 61% of children with laboratory-confirmed influenza virus [1] . cri symptoms were present in 90·9% (20/22) of laboratory-confirmed influenza cases in this study. a highly sensitive definition of influenza may be required to diagnose most of adult influenza cases in the clinical setting to ensure rapid treatment and isolation, and prevention of nosocomial transmission. inclusion of ili cases may overestimate the proportion of febrile cases in influenza surveillance given fever is included in the definition. pre-symptomatic and asymptomatic influenza cases will also be missed, although infectivity and transmissibility of these cases is yet to be proven [29] . longitudinal studies, where all participants are tested, provide similar estimates around rates of fever as in our study [14] . we propose a more sensitive clinical case definition without fever as a requisite criterion. clinical signs and symptoms are less studied for other viral respiratory infections, but available evidence suggests that other respiratory viruses are associated with a lower rate of fever compared to influenza [5, [30] [31] [32] [33] . putto and colleagues [30] examined the clinical records of 258 children (>3 months) in a large hospital in finland, including adv (25 cases), influenza a and b (74 cases), piv (99 cases) and rsv (60 cases). fever (539·0°c) was recorded in 68% cases with adv, 84% influenza a virus, 65% influenza b, 41% piv-1, 50% piv-2, 47% piv-3, and 52% rsv. van den hoogen and colleagues estimated the prevalence and clinical symptoms of hmpv infection, in the netherlands and fever was reported in 61% of the hmpv-positive cases [31] . in hong kong, hmpv was found in 5·5% (32/587) of children admitted in hospitals and all had fever [32] . manoha et al. examined nasal wash specimens from 931 hospitalized children and found hmpv (6%), rsv (28·5%), rhinoviruses (18·3%), influenza a (6%), piv-1 (0·2%) and piv-3 (0·3%). fever was reported in 39·2% cases with hmpv, 37·8% cases with rsv and 30·2% with rhinovirus [33] . of the 210 elderly patients with influenza and 145 with rsv, fever was reported in 65% and 50%, respectively [3] . a us study also reported low rates of fever in hcws infected with coronavirus 229e (13·5%), coronavirus hku (11·4%), coronavirus nl63 (31·3%) and rsv (12·9%) and all cases of hmpv were without fever [5] . rate of fever for all other viruses (excluding influenza) was 21·5% (28/130) in this study. co-infection with more than one virus was the strongest predictor of fever for adults with confirmed viral respiratory infections in the present study. previous studies also show high rates of fever in cases with dual respiratory viral infections compared to single viral infection [34, 35] . rates of hospitalization and icu admission are also reported to be higher in cases with dual respiratory viral infections [36] [37] [38] . increased severity of symptoms in co-infection cases might be due to an altered immune response [34] . around 10% (15/158) of cases in our dataset were infected with more than one virus. drews et al. reviewed the data of eight prospective epidemiological studies and reported the rate of co-infection was 5% [37] . studies in children generally report higher rates of co-infection cases (17-20%) [34, 35, 38, 39] . clinicians should consider the possibly of co-infection if a patient presents with fever; however, further epidemiological and clinical studies are required. smoking was also a significant predictor of fever in this study. smoking increases the risk of viral and bacterial infections through changes in respiratory epithelial and altered immune response [40] [41] [42] [43] .the risk of influenza also increases several times in smokers, compared to non-smokers [40] . atypical clinical presentation of influenza and other respiratory infections in adults could be due to altered structural and immune response associated with active/passive smoking and other environmental hazards. the mechanism by which smoking increases the risk of fever is not clear. high rates of fever in smokers may also be due changes in immunoglobulin levels which could increase viral load. the severity of symptoms generally increases when high viral load is detected in the blood [44] . the difference in fever rates between china and vietnam may be due to prevalence of viruses and co-infection. rsv was the most commonly isolated pathogen from china (31%), followed by rhinovirus (20%) and influenza virus (13%). in contrast to this hrv was the most commonly isolated pathogen from vietnam (85·3%). the number of cases with co-infection were also different in the two countries -13 (14%) in china and two (3%) in vietnam. in multivariate analysis, we adjusted for country and type of virus. limited data are available regarding the prevalent viruses circulating in china during the study period. for the trial 1 period, all influenza was influenza a(h1n1)pdm. for the trial 2 period, 21·3% were h1n1pdm, 2·9% were h3n2, 3·0% were influenza b victoria, 2·6% were influenza b yamagata, 71·2% were influenza a unsubtyped (y. zhang, beijing centre for disease prevention and control, personal communication). we could not obtain data on the viruses circulating in vietnam during the study period. there are some limitations to this study. we did not subtype the influenza strains, and studies show that the rate of fever also varies between influenza strains [7] [8] [9] [10] 18] . fever data was self-reported but self-measured in three trials using a traditional glass and mercury thermometer. lower fever rates in chinese hcws in this study might be due to due to differences in circulating viruses (and their pyrogenicity) between the two countries when the studies were conducted. a japanese study of children with influenza reported a tendency towards shorter duration of fever with increasing age in children [18] ; however, age and other demographic characteristics were not significant in that study. compared to children, this study shows that adults are less likely to have fever with a respiratory viral infection, even influenza. the implication of this finding is that for rapid treatment and reducing the risk of transmission of infection, clinicians should be aware that a diagnosis of viral respiratory infection, even influenza, is possible in the absence of fever. many of these infections are transmissible even when infected persons are asymptomatic or presymptomatic, and greater vigilance for respiratory symptoms in hcws could reduce nosocomial transmission of respiratory viral infections. the absence of fever should not preclude a differential diagnosis of influenza or other respiratory viruses in adults. influenza surveillance in communitydwelling elderly compared with children viral respiratory infections in the institutionalized elderly: clinical and epidemiologic findings respiratory syncytial virus and influenza a infections in the hospitalized elderly respiratory viruses transmission from children to adults within a household influenza among afebrile and vaccinated healthcare workers clinical manifestations and consequences of influenza tecumseh study of illness. xiii. influenza infection and disease differing virulence of h1n1 and h3n2 influenza strains influenza a and b virus infections in children differences in clinical features between influenza a h1n1, a h3n2, and b in adult patients seasonal influenza in adults and children -diagnosis, treatment, chemoprophylaxis, and institutional outbreak management: clinical practice guidelines of the infectious diseases society of america signs and symptoms predicting influenza in children: a matched case-control analysis of prospectively collected clinical data clinical signs and symptoms predicting influenza infection viral shedding and clinical illness in naturally acquired influenza virus infections is influenza an influenza-like illness? clinical presentation of influenza in hospitalized patients fever in the elderly ischaemic heart disease, influenza and influenza vaccination: a prospective case control study natural course of fever during influenza virus infection in children clinical features of the initial cases of 2009 pandemic influenza a (h1n1) virus infection in china clinical predictors for diagnosing pandemic (h1n1) 2009 and seasonal influenza (h3n2) in fever clinics in beijing a cluster randomised trial of cloth masks compared to medical masks in healthcare workers a cluster randomized clinical trial comparing fit-tested and non-fit-tested n95 respirators to medical masks to prevent respiratory virus infection in health care workers a randomized clinical trial of three options for n95 respirators and medical masks in health workers hospital triage system for adult patients using an influenza-like illness scoring system during the 2009 pandemic -mexico pandemic influenza triage tools: user guide predicting influenza infections during epidemics with use of a clinical case definition evaluation of clinical case definitions of influenza: detailed investigation of patients during the 1995-1996 epidemic in france influenzavirus infections in seattle families, 1975-1979. i. study design, methods and the occurrence of infections by time and age does influenza transmission occur from asymptomatic infection or prior to symptom onset? fever in respiratory virus infections prevalence and clinical symptoms of human metapneumovirus infection in hospitalized patients children with respiratory disease associated with metapneumovirus in hong kong epidemiological and clinical features of hmpv, rsv and rvs infections in young children single versus dual respiratory virus infections in hospitalized infants: impact on clinical course of disease and interferon-gamma response correlation of viral load of respiratory pathogens and co-infections with disease severity in children hospitalized for lower respiratory tract infection single, dual and multiple respiratory virus infections and risk of hospitalization and mortality dual respiratory virus infections multiple simultaneous viral infections in infants with acute respiratory tract infections in spain multiple versus single virus respiratory infections: viral load and clinical disease severity in hospitalized children cigarette smoking and infection smoking and the outcome of infection cigarette smoke extract suppresses human dendritic cell function leading to preferential induction of th-2 priming the associations of race, cigarette smoking, and smoking cessation to measures of the immune system in middle-aged men correlation of rhinovirus load in the respiratory tract and clinical symptoms in hospitalized immunocompetent and immunocompromised patients the authors thank the staff at the beijing centre for disease control and national institute of hygiene and epidemiology. thanks are also due to the staff from the hospitals in china and vietnam which participated. 3m helped in fit testing during the first rct in china, no financial support was provided. the second rct in china was funded through the australian national health & medical research council of australia (grant no. 630787). funding to conduct the vietnam trial study was received from the australian research council (arc) (grant no. lp0990749). dr abrar chughtai had testing of filtration of masks by 3m for his phd. professor c. raina macintyre has held an australian research council linkage grant with 3m as the industry partner, for investigator-driven research. 3m have also contributed supplies of masks and respirators for investigator-driven clinical trials. she has received research grants and laboratory testing as in-kind support from pfizer, gsk and bio-csl for investigator-driven research. the remaining authors have no competing interests to declare. key: cord-286837-j2sqs20q authors: koetsier, antonie; van asten, liselotte; dijkstra, frederika; van der hoek, wim; snijders, bianca e.; van den wijngaard, cees c.; boshuizen, hendriek c.; donker, gé a.; de lange, dylan w.; de keizer, nicolette f.; peek, niels title: do intensive care data on respiratory infections reflect influenza epidemics? date: 2013-12-31 journal: plos one doi: 10.1371/journal.pone.0083854 sha: doc_id: 286837 cord_uid: j2sqs20q objectives: severe influenza can lead to intensive care unit (icu) admission. we explored whether icu data reflect influenza like illness (ili) activity in the general population, and whether icu respiratory infections can predict influenza epidemics. methods: we calculated the time lag and correlation between ili incidence (from ili sentinel surveillance, based on general practitioners (gp) consultations) and percentages of icu admissions with a respiratory infection (from the dutch national intensive care registry) over the years 2003–2011. in addition, icu data of the first three years was used to build three regression models to predict the start and end of influenza epidemics in the years thereafter, one to three weeks ahead. the predicted start and end of influenza epidemics were compared with observed start and end of such epidemics according to the incidence of ili. results: peaks in respiratory icu admissions lasted longer than peaks in ili incidence rates. increases in icu admissions occurred on average two days earlier compared to ili. predicting influenza epidemics one, two, or three weeks ahead yielded positive predictive values ranging from 0.52 to 0.78, and sensitivities from 0.34 to 0.51. conclusions: icu data was associated with ili activity, with increases in icu data often occurring earlier and for a longer time period. however, in the netherlands, predicting influenza epidemics in the general population using icu data was imprecise, with low positive predictive values and sensitivities. a limitation of current influenza surveillance systems is that timely information on severe influenza illness requiring hospital admission is not available. influenza surveillance in most countries is based upon sentinel general practitioners (gp) networks and the collected information on influenza like illness (ili) is dependent on the health care seeking behavior of the general population, which can fluctuate with for example media attention. the implementation of a hospital based surveillance system for severe acute respiratory infection (sari) is now promoted by the world health organization (who) and european centre for disease prevention and control (ecdc) as a public health priority worldwide, both for routine surveillance and for preparedness [1, 2] , such as in the case of the middle east respiratory syndrome coronavirus (mers-cov). sari surveillance can focus on admissions for respiratory infections in general hospital wards or in intensive care units (icu) . during the pandemic period, hospitalization for laboratory confirmed influenza a(h1n1)pdm09 infection was notifiable in the netherlands. in the season 2009/2010 as well as in the season 2010/2011, ili incidence as measured by gp sentinel practices, reached the epidemic threshold of 5.1 consultations per 10.000 enlisted patients at a time when already more than 100 patients had been hospitalized, with several icu admissions and deaths from laboratory confirmed influenza (national institute for public health and the environment, unpublished surveillance data). hospital admission for influenza is not notifiable anymore and in the netherlands sari cases are not routinely collected. an alternative source of information could be the dutch national intensive care evaluation registry (nice) [3], wherein diagnostic, and physiologic information from the first 24 hours of adult icu admissions, as well as length of stay and in-hospital mortality of all icu patients are registered. patients are admitted to the icu if they have a high severity of illness, and require constant monitoring of their vital functions, regardless of their expected outcome. respiratory infections, such as pneumonia, are among the most common conditions for which patients are admitted to the icu. in the early onset of an influenza epidemic, patients with multiple comorbidities are more likely to develop more severe influenza related diseases like pneumonia. they possibly get submitted to the icu before there is an epidemic in the general population [4] . therefore, increases in the number of admissions at the icu with a respiratory infection can possibly occur before a detectable increase in ili incidence in the gp sentinel network. in this study we explore whether icu data on respiratory infections reflect ili activity in the general population, and which relevant time lag exists between both data sources. additionally we assessed whether icu data can predict ili defined influenza epidemics (further referred to as influenza epidemics). in the netherlands, patients that consult the gps with symptoms of ili are reported on a weekly basis to the continuous morbidity registration sentinel general practice network (further referred to as sentinel gp registry), covering 0.8% of the dutch population being nationally representative by age, gender, regional distribution, and population density [4] . the sentinel gp data consists of the weekly number of patients presenting with ili at the gp. in 2009, the registry had 42 participating gp practices with 59 gps covering a population of approximately 130,000 patients [4] . ili was defined according to the criteria of pel [5] , by (1) acute onset, with prodromal stadium of three to four days, (2) a temperature increase to at least 38 degrees celsius, and (3) at least one of the following symptoms: cough, nasal congestion, raw throat, frontal headache, retrosternal pain, and body aches. incidence of ili is calculated on a weekly basis, using number of patients registered at the reporting gp practices as the denominator. this is acceptable as almost every person in the netherlands is registered with a gp. in the netherlands, an influenza epidemic is defined as more than 5.1 patients with ili per 10,000 inhabitants per week consulting the gp for at least two consecutive weeks [6] , combined with influenza a virus isolation from laboratory samples. data quality is assured by training of gps for data entry, and a pop-up appearing in the system when an ili diagnosis is entered reminding the gp to register the ili case in the sentinel gp registry. the nice registry was founded in 1996, with initially six participating icus. in 2003 this number had grown to 33 and in 2011 85 icus participated covering approximately 90% of the dutch adult icus. for each admission, among other items, the acute physiology and chronic health evaluation ii (apache ii) [7] reason for icu admission diagnosis is registered and sent to the nice registry on a monthly basis. upon receipt, the data is usually entered into the nice registry database and available for participants within one day. there is currently no distinct variable describing whether an icu patient has influenza like illness or was diagnosed with an influenza virus infection, or receives antiviral drugs. therefore, we defined an icu admission with a possible sari (further referred to as icu admission with respiratory infection) when the following four criteria were met: (1) the patient was admitted to the hospital less than two days before icu admission, (2) there was a medical (non-surgical) reason for admittance, (3) the icu admission was not a readmission to the icu within the hospitalized period, and (4) the apache ii reason for admission was 'respiratory infection. we used the percentage of icu admissions with a respiratory infection (relative to the total number of medical icu admissions), instead of the absolute number of icu admissions to adjust for the growing number of nice registry participants throughout the study period. thus our study dataset included study year, week number, number of patients with ili, population size of reporting gps, number of icu admissions with respiratory infection, and number of medical icu admissions. data quality is assured by regular on site visits of the icus [8] . in this study, we used weekly time series of patients presenting with ili from the sentinel gp registry and icu admissions of respiratory nature from the nice registry. in the nice registry, few icus were participating in the years 2000 until 2002, leading to large variations in the percentage of icu admissions with respiratory infections. therefore, from both registries, we used data from 2003 through 2011. as influenza generally occurs between week 40 and week 20 of the subsequent year [9] , we defined an influenza year (i.e. season) from july 1st until june 30th the next year, thereby having ten influenza years in our dataset. to explore the association between the weekly incidence of ili patients and the percentage of icu admissions with a respiratory infection, we plotted per week the percentage of icu admissions with a respiratory infection and the incidence of ili over the period january 1, 2003, through december 31, 2011. in addition, we performed a generalized estimating equations (gee) [10] additive poisson regression analysis [11] using an autoregressive working correlation matrix over this time period. the dependent variable was weekly incidence of patients with ili, and the independent variables were chronological week number, percentage of icu admissions with respiratory infection in the current week, and one to five weeks before the current week and one to five weeks after the current week, and sine and cosine terms to adjust for seasonality [12] . the sine term was sin(k2pt/t) and the cosine was cos(k2pt/t), where k is a constant with values 1 (yearly seasonality) or 2 (half year seasonality), t is current week number, and t is total number of weeks in the specific influenza year, e.g. 52 weeks (years 2004 and 2009 had 53 weeks, and the cases were added to week 52). we adjusted for autocorrelation in the residuals, where the unit of clustering was influenza year. we calculated the average time lag between the sentinel gp and icu data by computing the weighted average of the time lag in weeks (25, …, 0, …, +5), using the corresponding regression coefficients as weighting factors. the r-squared (r 2 ) value based on the deviance residuals [13] was also calculated. to assess the possibility of using icu data for predicting influenza epidemics, we used a subset of three years of training data to develop three gee models with the same characteristics as the aforementioned model, to predict the incidence of ili patients one to three weeks ahead. in each of these models, independent variables were removed by pseudo stepwise selection [14] , using a fixed scheme for removal. we first considered the time trend (chronological week number) for removal, then seasonal terms, and finally time lagged percentage of icu admissions with respiratory infection. in order for the final models to be useful in surveillance, the following restrictions also applied: (1) the percentages of icu admissions with respiratory infection one to five weeks after the current week were not included as they are unavailable and useless for prospective surveillance, (2) the percentage of icu admissions with respiratory infection in the current week cannot be removed from the model, (3) additional variables of lagged icu admissions should correspond to a range of subsequent weeks (e.g., one and two weeks before the current week, not one and three weeks before the current week), and (4) a seasonal term is always a combination of a sine and cosine function. we used data of the first three influenza years (january 1, 2003 1, , through june 30, 2005 to generate the final model for predicting the incidence of ili one week ahead. to accomplish this, we varied the decay factor l of the full model, giving weeks further back in time exponentially lower weights, from 0.97 to 1.00 with increments of 0.005, resulting in seven candidate models. variable selection for these seven models was performed with pseudo stepwise selection, with the quasi-likelihood information criterion (qic) [15] as performance measure. to determine the optimal value of l, 10-fold cross validation [16] was performed for the seven candidate models using the same three influenza years that they were built with. the model with the best r 2 value was selected as final model. the above steps were repeated for variable selection of the models predicting ili two and three weeks ahead. using the final models based on the 3 training years of data we started predicting the incidence of ili patients week by week starting from the fourth influenza year in our dataset (july 1, 2005) onward. before predicting each successive week, the model parameters were recalculated with an updated dataset that included the data of the previous week to make the model dynamic. we continued updating the model parameters week by week, until all remaining seven influenza years were predicted. from the fourth influenza year onward, the predicted incidence of ili patients was plotted together with the observed incidence of ili patients. we used the same threshold as the ili sentinel surveillance to define an influenza epidemic in the predicted ili numbers (incidence . = 5.1 ili patients per 10,000 inhabitants for at least two consecutive weeks). we compared the predicted epidemic weeks using icu data with observed epidemic weeks based on ili data. accordingly, we calculated the positive predictive value (ppv), and sensitivity of the predicted epidemic weeks on a weekly basis. for comparison, we also predicted the weekly incidence of ili with models that used only seasonal terms and auto-regressive ili variables, but excluded icu data. these models were also created with pseudo stepwise selection. the resulting ppv and sensitivity were also calculated. the statistical analyses were performed using the statistical package r, version 2.15.1 (http://www.r-project.org/; vienna, austria). in the period january 1, 2003, through december 31, 2011 there were a total of 477,422 icu admissions, of which 133,615 (28.0%) had a medical (non-surgical) reason for admittance. of the medical icu admissions, 17,786 (13.3%) were for a respiratory infection ( table 1 ). the incidence of ili was on average 3.2 per 10,000, with a standard deviation of 2.8, and a minimum of 0.1 per 10,000 and a maximum of 17.5 per 10,000. there were nine epidemics, consisting of 72 weeks in total, and an average length of eight weeks. on average, 43 gps supplied data on ili. both the incidence in ili and percentage of icu admissions with a respiratory infection show a similar timing of seasonal peaks (figure 1 ). the amplitude of the yearly peaks in the icu data were relatively lower than ili peaks, and often lasted for a longer time period. increases in the incidence of ili showed a yearly pattern and increased in a smoother pattern compared to icu data. while trends were roughly comparable, they differed in some instances since peaks in icu data occasionally occurred when ili increases were absent in the general population. the association between the different lags of icu admissions and ili incidence is shown in table 2 (assessed using gee additive poisson regression analysis). the r 2 value of the full model was 0.58. of each variable, the contribution to the r 2 of the full model is also shown, which is the r 2 value of the full model minus the r 2 of the model with the corresponding variable omitted. figure 1 also shows the predicted ili incidence (black line) according to the full model. statistically significant time lags were percentage of icu admissions with respiratory infection were one week before, current week, one week after, two weeks after, and four weeks after current ili incidence. the time lags mostly associated with increases in ili incidence one week before and in the current week, with coefficients of 0.12 and 0.11. for example, if the percentage of icu admissions with a respiratory infection in the current week increased by one percent, then the incidence of ili in the general population increased by 0.11 per 10,000 population. according to the contribution to r 2 , also icu admissions one week later was strongly associated with current ili (coefficient of 0.08). there is no linear time trend present, but seasonality exists in the data reflected by a half year sine function (sine term with k = 2) with a p-value of ,0.01 and a large contribution to r 2 . looking at figure 1 , a yearly time trend would be expected but is now partly reflected in the different icu time lagged variables. using the coefficients of the time lags in table 2 , the average of the weighted relative week numbers was 20.24 weeks implying that the increase in percentage of icu admissions with a respiratory infection was on average 1.68 days earlier than the increase in ili incidence. for our second research question, whether icu data can predict ili incidence ahead in time, we generated three gee models predicting the incidence of ili patients one to three weeks ahead using icu data. table 3 shows these three different gee models, and their ppv and sensitivity. predicting two weeks ahead yields the largest sensitivity of 0.51 and predicting one week ahead has the largest ppv of 0.78. for comparison, models using only auto-regressive ili variables and seasonal terms, showed the sensitivity to range between 0.21-0.22 and the ppv between 0.31-0.37. figure 2 shows three figures plotting the predicted incidence of ili patients one to three weeks ahead versus the actually observed ili incidence. the epidemic threshold of 5.1 patients (or more) with ili per 10,000 population is plotted and the weeks in which an influenza epidemic occurred according to the predicted versus the actual data is shown. predicting one week ahead detected three of the six epidemics, of which two were longer and one shorter according to the icubased predictions. one epidemic was predicted earlier, one at the same time and one later. using icu data to predict two weeks ahead resulted in five of the six epidemics detected, one is missed, and one false epidemic is predicted in the autumn of 2010 (which was not present in the ili data). according to the icu data, the predicted epidemics were longer in time except one. besides, most predictions were shifted in time compared to the actual occurrence: two epidemics were predicted earlier and three later. when predicting three weeks ahead all six epidemics were detected, however four were shorter in length, one longer, and one had the same length, again shifted in time: two epidemics were predicted earlier, three later, and one at the same time with icu data. the study showed that the percentage of medical icu admissions for respiratory infection was associated with weekly incidence of ili in the current week, and with one week positive and negative time lag. an increase in the percentage of icu admissions for respiratory infections on average preceded the increase in the incidence of ili (gp data) by 1.68 days, implying that before an epidemic the severely ill influenza cases get admitted to the icu. despite this precedence, our analyses showed that with the current models icu data do not accurately predict influenza epidemics in the general population, but including icu data showed an improvement in sensitivity and ppv compared to only including auto-regressive ili variables and seasonal terms. in our study we built three additive poisson gee regression models with icu data to predict the incidence of ili patients, thereby detecting influenza epidemics and aimed at detecting opportunities for enhancing the current national surveillance method. previous studies also aimed at enhancing their current surveillance of influenza epidemics, using laboratory or hospital data. steiner et al. [17] used an exponentially weighted moving average control chart to enhance and automate influenza epidemic detection. weekly laboratory notifications data of seven years were used instead of the ili data that we studied. the predicted influenza epidemics were compared to retrospective inspection of the same notification data by epidemiologists. the predictions were, just like our study, not the same as their reference data. however in their study there was a maximum of one week difference only, except for one year where there was a difference of eight weeks. a study by closas et al. [18] used a kolmogorov-smirnov test with virologic laboratory data of five years to detect influenza epidemics. the test provides a binary signal indicating epidemic activity and a quantitative measure of its confidence. they sequentially updated the test as new data became available. the results differed one to nine weeks with the retrospective data of the sentinel network, which is comparable to our results. google flu trends also aimed to detect influenza epidemics, but overestimated peak influenza levels [19] whilst our study underestimates peak influenza levels. these methods complement the current surveillance networks, but cannot replace them. a study by van den wijngaard et al. [20] did not aim to predict influenza epidemics, but instead explored whether excesses in influenza severity per season can be detected by combining gp, hospital, laboratory, and mortality data (7 years of data). their finding was that combining these data sources is of added value, allowing for better understanding of increases in severe morbidity and mortality due to influenza infections. also from our data we see that trends in icu related sari differ from the trends of ili in the general population and may thus be of value in offering additional information on severity of influenza seasons which need to be explored further. however, both respiratory icu admissions and ili in the general population are not necessarily caused by influenza alone. microbiological laboratory results would provide better insight but to date, these data are not available at the icu patient level. the major strength of our study is that we had access to two large historical datasets from the nice registry and the sentinel gp registry. this allowed us to retrospectively analyze ten influenza years of data, which, to our knowledge, is a longer time period than in comparable studies. a second strength of our study is that we used gee in our additive poisson model, thereby correcting for correlations between weeks. the last strength of our method is that for each additional week we sequentially updated the coefficients of the covariates in the model used for prediction of ili, adding a decay factor giving historic data less weight, and adjusted our models for seasonal changes. with these adjustments, our models always incorporated the most recent information on icu and ili trends. a limitation of the nice registry data is that there is no distinct variable describing whether a patient has an influenza like illness or whether an patients has been diagnosed with an influenza virus infection. furthermore it only contains adult patients thus representing an older population compared to the ili surveillance which also includes children. we extracted admissions with a medical respiratory infection, admitted to the icu within two days after hospital admission, and excluding readmissions. these admissions represent community-acquired respiratory infections and, therefore, included influenza virus infections. additionally, the data of the sentinel gp registry is weekly updated, whereas the nice registry is updated on a monthly basis. this frequency is developed because outcome data, e.g. mortality, is measured at hospital discharge. for sentinel purposes this delay is too long and more frequent updates are needed. however our results can give incentives to set up an additional registry of near real-time surveillance of sari cases at the icu. our statistical analysis also has some limitations. due to the weekly scope of the ili data, we aggregated the icu data on a weekly basis, losing detail as they are available on a daily basis. with regard to our chosen models to predict ili, in the ideal situation stepwise variable selection is combined with 10-fold cross validation. since automating this process is not possible in gee, we first performed stepwise selection and then 10-fold cross validation on the seven remaining candidate models. additionally, we used three years of data as training set to determine the best models, whereas a longer period would also have been an option but not necessarily better, since we continuously added data to the baseline data. another limitation is that during the 2009-2010 pandemic, the ili peaks were not detected or later. this means that our models were not sensitive to large or unexpected changes. apparently the association between icu admissions and ili in the general population can change greatly from season to season. icu related sari might occur at a very different rate (compared to symptoms in the general population) during a pandemic or unexpected seasons [9] . a probable explanation is that the influenza pandemic caused by the a(h1n1)pmd09 virus targeted another patient population than the previous epidemics with severe illness in younger patients, and fewer elderly with a severe infection. this could explain why increases in icu admissions during the pandemic were later than usual. additionally, the icu data reflects only sari cases. therefore, we do not know if the icu data reflect an influenza epidemic in the general population or possibly very different influenza dynamics in the icu population alone. icu data on respiratory infections was associated with ili incidence, with highest association in the same week and in the week before and the week after. increases in icu data on average occur two days sooner and for a longer time period than increases in ili. icu data thus contains additional information on icu related sari cases during a specific influenza epidemic. predicting influenza epidemics one, two or three weeks ahead in the general population using icu data was imprecise, reflected by the low ppvs and sensitivities. thus, icu data cannot improve the current surveillance method to detect influenza epidemics. due to the association between both data sources, a next step is to investigate the possibility of using icu data in combination with microbiological laboratory results for surveillance of severity of illness, and icu capacity prediction when an (severe) influenza epidemic is present. the performance of the gee models in predicting the start, end and length of an influenza epidemic is expressed by the positive predictive value (ppv), and sensitivity (n = 338 weeks) based on comparing the signals for an epidemic predicted with intensive care unit (icu) data with the reference standard from the observed influenza like illness data. doi:10.1371/journal.pone.0083854.t003 surveillance trends of the 2009 influenza a(h1n1) pandemic in europe responding to new severe diseases-the case for routine hospital surveillance and clinical networks in europe continuous morbidity registration sentinels: netherlands proefonderzoek naar de frequentie en de aetiologie van griepachtige ziekten in de winter 1963-1964 modelling influenza epidemics: can we detect the beginning and predict the intensity and duration? apache ii: a severity of disease classification system defining and improving data quality in medical registries: a literature review, case study, and generic framework comparing pandemic to seasonal influenza mortality: moderate impact overall but high mortality in young children longitudinal data analysis using generalized linear models fitting additive poisson models studying seasonality by using sine and cosine functions in regression analysis r-squared measures for count data regression models with applications to health-care utilization a biometrics invited paper. the analysis and selection of variables in linear regression akaike's information criterion in generalized estimating equations how biased is the apparent error rate of a prediction rule detecting the start of an influenza outbreak using exponentially weighted moving average charts sequential detection of influenza epidemics by the kolmogorov-smirnov test detection of excess influenza severity: associating respiratory hospitalization and mortality data with reports of influenza-like illness by primary care physicians key: cord-282668-bs634hti authors: niang, mbayame ndiaye; diop, ndeye sokhna; fall, amary; kiori, davy e.; sarr, fatoumata diene; sy, sara; goudiaby, déborah; barry, mamadou aliou; fall, malick; dia, ndongo title: respiratory viruses in patients with influenza-like illness in senegal: focus on human respiratory adenoviruses date: 2017-03-22 journal: plos one doi: 10.1371/journal.pone.0174287 sha: doc_id: 282668 cord_uid: bs634hti background: human adenoviruses (hadvs) are highly contagious pathogens that are associated with a wide spectrum of human illnesses involving the respiratory tract. in the present study, we investigate the epidemiologic and viral molecular features of hadvs circulating in senegal after 4 consecutive years of sentinel surveillance of influenza-like illness cases. methodology and results: from january 2012 to december 2015 swabs were collected from consenting ili outpatients. adenoviral detection is performed by rrt-pcr with the anyplex(™) ii rv16 detection kit (seegene) and molecular characterization was performed using a partial hexon gene sequence. 6381 samples were collected. more than half of patients (51.7%; 3297/6381) were children of ≤ 5 years. 1967 (30.8%) were positive for hadv with 1561 (79.4%) found in co-infection with at least one another respiratory virus. the most common co-detections were with influenza viruses (53.1%; 1045/1967), rhinoviruses (30%; 591/1967), enteroviruses (18.5%; 364/1967) and rsv (13.5%; 266/1967). children under 5 were the most infected group (62.2%; 1224/1967; p <0.05). we noted that hadv was detected throughout the year at a high level with detection peaks of different amplitudes without any clear seasonality. phylogenetic analysis revealed species hadv-c in majority, species hadv-b and one hadv4 genome type. the 9 hadv-b species like strains from senegal grouped with genome types hadv-7, hadv-55 and hadv-11 as shown by a phylogenetic branch with a high bootstrap value of (88%). conclusion: in conclusion, the results of the present study suggest strong year-round hadv activity in senegal, especially in children up to 5 years of age. molecular studies revealed that the dominant species in circulation in patients with ili appears to be hadv-c and hadv-b species. the circulation of though hadv-7 and hadv-55 genome types is of note as these serotypes are recognized causes of more severe and even fatal acute respiratory infections. from january 2012 to december 2015 swabs were collected from consenting ili outpatients. adenoviral detection is performed by rrt-pcr with the anyplex™ ii rv16 detection kit (seegene) and molecular characterization was performed using a partial hexon gene sequence. 6381 samples were collected. more than half of patients (51.7%; 3297/6381) were children of 5 years. 1967 (30.8%) were positive for hadv with 1561 (79.4%) found in co-infection with at least one another respiratory virus. the most common co-detections were with influenza viruses (53.1%; 1045/1967), rhinoviruses (30%; 591/1967), enteroviruses (18.5%; 364/1967) and rsv (13.5%; 266/1967). children under 5 were the most infected group (62.2%; 1224/1967; p <0.05). we noted that hadv was detected throughout the year at a high level with detection peaks of different amplitudes without any clear seasonality. phylogenetic analysis revealed species hadv-c in majority, species hadv-b and one hadv-4 genome type. the 9 hadv-b species like strains from senegal grouped with genome types hadv-7, hadv-55 and hadv-11 as shown by a phylogenetic branch with a high bootstrap value of (88%). in conclusion, the results of the present study suggest strong year-round hadv activity in senegal, especially in children up to 5 years of age. molecular studies revealed that the dominant species in circulation in patients with ili appears to be hadv-c and hadv-b a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 species. the circulation of though hadv-7 and hadv-55 genome types is of note as these serotypes are recognized causes of more severe and even fatal acute respiratory infections. human adenoviruses (hadvs) are highly contagious pathogens that are associated with a wide spectrum of human illnesses involving the respiratory, ocular, gastrointestinal, and genitourinary systems [1] . they belong to the family adenoviridae, genus mastadenovirus with seven species (a-g), including each various types [2] . ubiquitous in the environment, hadvs are non-enveloped, double stranded dna viruses that vary in size from 70 to 100 nm [3] . hadvs are recognized as a common cause of respiratory infection in persons of all ages. the illnesses range from influenza-like fever and discomfort to pneumonia and death [4] . indeed, hadvs infections are usely mild but some groups such as very young children, elderly, immunocompromised persons, or persons with underlying pulmonary or cardiac disease, might be at higher risk degree for severe disease [5, 6, 7, 8] . the most common hadvs species that cause respiratory tract infections in children are b (hadv-b3 and b7) and c (hadv-c1, c2, and c5). serotypes b3, b7, and b21 are the most frequent strains responsible for epidemics of acute febrile respiratory disease [9] . circulating hadvs can vary temporally and geographically with possibility of emergent genomic variants which can be associated with more severe illness [10, 11] . in the present study, we investigate the epidemiologic and viral molecular features of hadvs circulating in senegal after 4 consecutive years of sentinel surveillance of influenzalike illness cases. from january 2012 to december 2015 we collected specimens (nasal-pharyngeal and oral-pharyngeal swabs) and surveillance data for influenza and other viral respiratory pathogens from outpatients presenting with influenza-like-illness (ili) at different sentinel sites in senegal. once collected, swabs are placed in 2-ml cryovials with viral transport medium (universal transport medium; copan diagnostics inc., murrieta, ca), and transported at a controlled temperature of 2˚c-8˚c to the laboratory. an ili patient was defined as a person presenting with sudden onset of fever (>38˚c) or history of sudden onset of fever in the recent past ( 3 days) and either cough or sore throat and/or rhinorrhea in the absence of other diagnosis, according to the cdc case definition. each sample is accompanied by a case report form collecting demographic and clinical data. the questions included information on date of enrollment and symptom onset, sex, age, clinical symptoms, previous treatments, travelling history, vaccination status for influenza, and whether or not the patient was hospitalized. upon arrival at the laboratory, the specimens were processed immediately for virus diagnosis. aliquots of samples were also stored at −80˚c for additional analysis (isolation and/or molecular characterization). the data obtained daily were entered into an epi info database (centers for disease control and prevention, atlanta, ga) and analyzed using epi info. total viral nucleic acid (dna and rna) was extracted from 140 μl of each clinical specimen using the purelink™ viral rna/dna mini kit (invitrogen, carlsbad ca, usa) according to the manufacturer's recommendation. dna/rna are eluted with 60 μl nuclease-free water and stored at −80˚c until use. a two-step multiplex real-time rt-pcr was performed with a bio-rad cfx-96 thermocycler (bio-rad laboratories) and the anyplex™ ii rv16 detection kit (seegene) for a simultaneous testing of influenza viruses (flua and flub), human respiratory syncytial virus (rsva and rsvb), human adenoviruses (hadv), human metapneumovirus (hmpv), human coronavirus (229e, nl63, oc43), human parainfluenza virus (piv1, -2, -3 and -4), human rhinovirus (hrv), human enterovirus (hev) and human bocavirus (hbov), as previously described [12] . in consideration with low ct-values, 80 hadv positives samples (20 per year) were selected using a random number generator on ms excel for further molecular characterization using classical pcr and sequencing. viral dna was extracted as previously described and eluted with 50 μl water nuclease-free. dnas were stored at −20˚c until pcr reactions. for hadv molecular characterization the last 300 base pairs (bp) of the hexon gene were amplified with the following specific primers: adeno3 (5'-cctttggcgcatcccattct-3') and adeno4 (5'-tgggcacctatgacaagcgc-3') previously used by garcia et al, [13] . the phusion high-fidelity pcr master mix with hf buffer (new england biolabs, ipswich ma, usa) was used for amplifications. for each sample, pcr was carried out in a total reaction volume of 50 μl consisting of 15 μl h2o rnase free, 2.5 μl of each primer (diluted at 10μm), 25 μl of 2x phusion master mix and 5 μl of dna template. cycling conditions were as follows: denaturation step of 15 min at 95˚c, 40 pcr cycles including 30 s at 95˚c, 60 s at 55˚c, 60 s at 72˚c followed by an extension step of 10 min of 72˚c. five microliters of the pcr product was then mixed with 1 μl of 10x 5prime loading dye and loaded on to a 1% agarose gel along with an appropriated molecular weight markers (100 bp ladder, new england biolabs), and gels were stained with ethidium bromide (0.5 μg/ml) before visualization under uv. for positive samples (380 bp size band), amplicons were cut and purified using the gene-jet gel extraction kit (thermo scientific). purified products are then sent for sequencing to beckman coulter services. sequencing was performed in both directions with the same pcr primers (adeno3 and adeno4) on an abi prism bigdye terminator v3.1 ready reaction cycle sequencing kit (applied biosystems) on a 96-capillary abi prism 3730-xl (applied biosystems). data in fasta format were then sent to the laboratory for analysis. sequences successfully obtained were aligned with representative genbank sequences of previously published genotypes using the bioedit sequence alignment editor [14] . the search for sequence similarities were carried out using the basic local alignment search tool (blastn) from ncbi blast web portal. phylogenetic trees were performed in mega 6 software [15] using the neighbor-joining method, and the statistical significance of the tree topology tested by bootstrapping (1,000 replicates). the evolutionary distances were derived using the tamura-nei method. bootstrap replicates with values !70 are shown on the trees. statistical analysis. regarding hadv infection comparisons between age groups were performed using the fisher's exact test. p value < 0.05 was considered statistically significant and the 0-5 year age group was used as reference group. hadv mono-infections were also compared to hadv co-infections. the r.3.0.1 tool was used to perform the analyses. this study is a component of the 4s network syndromic surveillance [12] . the principles of the 4s network were approved by the ministry of health in its guidelines for influenza surveillance policy, finalized with the support of pasteur institute in dakar and the strengthening influenza sentinel surveillance in africa (sisa) project funded by the who. the protocol and oral consent were determined as routine surveillance activity, and therefore non-research by the senegalese national ethics committee and the steering committee for 4s network, an entity representing moh, ipd, who and clinicians in compliance with all applicable national regulations governing the protection of human subjects. data were collected in an objective of surveillance and are anonymous. the information provided to participants was an informal description of the study. respiratory specimens were collected, only after informed consent was granted, verbally, to local health care workers by the patients or parents in the case of minors. oral consent was documented in the patient form with two questions about received information and about oral consent. patients could refuse to participate, no specimen will be taken. for the surveillance activities, written consent is judged not necessary by the senegalese national ethics committee, which has also previously approved the work of the national influenza center. collections of non-sensitive data or an observation from normal care in which participants remain anonymous do not require ethics committee review. the patients included in this study were of all ages and consulted the sentinel sites due to influenza-like symptoms; the patients, or parents in the case of minors, accept the tests for respiratory viruses largely because they are free and safe. of 6381 specimens tested, 1967 (30.8%) were positive for hadv (table 2) . detection rates over the study period are almost similar in the first 3 years (2012, 2013 and 2014) while in 2015 there is a marked decrease in adenoviral infections. the mean age of infected patients was 8 years 7 months and median age was 3 years. regarding the viral detection per age group, most of hadv infected cases (62.2%; 1224/1967) were under 5 years patients, a statistically significant finding (p <0.05). however, the detection rates in the other groups including the elderly (above 50 years old) remain high. no significantly gender distribution of adenoviral infection was observed. the comparison of symptoms prevalence between ili patients with adenoviral infection and patients without adenoviral infection showed that cases of myalgia (p = 0.0014), cough (p = 0.0028), diarrhea (p < 0.001), rhinitis (p < 0.001) and headache (p = 0.01) are significantly higher in patients infected by adenoviruses ( table 3 ). the fig 1 shows the temporal distribution of hadv positivity rate per month in senegal from 2012 to 2015. we noted that hadv was detected throughout the year at a high level with detection peaks of different amplitude. the highest peak, with 62% of detection rate, was recorded on december 2013. hadv circulation pattern shows no seasonality even if results suggest a higher activity of these viruses during cold periods. it should be pointed out that the cold periods (between december and february) experience some instability in senegal with possibilities of shifting. for phylogenetic analysis, we were able to obtain the partial hexon gene sequence from 54 hadv-positive samples: 8 were from samples in 2012, 13 from 2013, 11 from 2014, 16 from 2015 and 6 from 2016. unfortunately, some samples showed no amplification or poor-quality sequences. the low sensitivity of conventional pcr compared with real time pcr on samples with low viral load, and certainly non-specific amplifications could be the cause of these failures. the nucleotide sequence alignment clustered the majority of senegalese isolates into hadv-c species (44/54). 9 isolates grouped with hadv-b species and the remaining isolate, from 2012, seems close to the hadv-4 genome type belonging to the hadv-e species. in all cases bootstrap values are high (more than 85%). within the hadv-c species, 16 senegalese isolates are grouped with the type hadv-6 (36.4%); 2 isolates with hadv-2 type (4.5%), 4 with hadv-5 type (9%), and 22 isolates formed a subcluster with hadv-1 and 57 types (fig 2) . the 9 hadv-b like species from senegal grouped with genome types hadv-7, hadv-55 and hadv-11 as shown by a phylogenetic branch with a high bootstrap value of (88%). we also noted that this dominance of species c and b is confirmed over the years. in this four-year retrospective study, we characterized hadv isolates derived from an ili surveillance program conducted as collaboration between pasteur institute of dakar and the senegalese ministry of health between 2012 and 2015. it is the first nationally molecular epidemiology investigation of hadvs and even in west africa. our study suggests that hadv are strongly associated with ili syndrome in senegal with an overall detection rate of 30.8% among 6381 patients. this rate seems very higher in comparison with available data. indeed, much lower rates are reported in similar other studies conducted in other countries. a study conducted in kenya [16] on refugees from different countries (somalia, sudan, ethiopia and kenya) yielded a detection rate of 21.7%, in gabon douki et al [17] detected hadv in 16.3% of outpatients with ili. these detection rates are still lower in other geographical regions: south korea with 10.1% or 0.6% [18, 19] , china with 2.7% or 6.3% [20, 21] , philippines with 0.9% [22] , malaysia less than 2% [23], usa with 5.7% or 2.8% reported [24, 25] , canada in the ontario provence with 0.9% [26] , peru with 6.2% [27] , venezuela with 1.6% [28] , england with 6.6% [29] . the analysis of these data tends to confirm a higher prevalence of adenoviruses in the respiratory sphere in african populations. this trend was largely confirmed when we investigated the importance of hadv in children with acute respiratory infections. indeed we observed that proportions in cameroon (27.3%) [30] and senegal (29.2%) (fall et al, on submission) were considerably higher than those found in other geographical areas: nascimento-carvalho et al., [31] in brazil with 3%, moe et al., [32] in norway (1.7%), wansaula et al., [33] in usa (1%) or lu et al., [34] in china. however, these discrepancies in hadv detection rates can be also due to differences in technical approaches, virus burden geographical differences, the number of patients tested, the periods during which samples were collected and even the duration of the study. it should be also noted that adenoviral detection does not necessarily prove disease causation as coincidental upper airway infection, asymptomatic viral carrier state [35] , or prolonged shedding [36] in a previous infection could explain adenoviral detection. regarding the group age, as expected, results showed that most patients with hadv infection were younger than 5 years (62.2%), a statistically significant finding. these results are in concordance with those of other studies which findings concluded that most children are infected by adenovirus at an early age [25, 37, [38] [39] [40] [41] . indeed, it is well established that by 5 years of age, 70% to 80% of children demonstrate antibodies to at least one serotype [42] . additionally more than 80% of diagnosed hadv infections occur in children < 4 years old (due to lack of humoral immunity) [43] . although most cases exhibit low to mild symptoms are and indistinguishable from other viral causes, acute respiratory infections caused by hadv can be severe [44] , or even fatal [7, 45] , and are associated with the highest risk of long term respiratory sequelae [46] . consistent with the report from many other studies, results here showed that 79.4% of hadv infected participants were co-infected with one or more other respiratory tract viruses. the most frequently co-detected viruses were influenza viruses (53.1%), rhinoviruses (30%), enteroviruses (18.5%) and rsv (13.5%). however, we noted no significant differences in clinical characteristics and laboratory findings between patients with single hadv infection and those co-infected. the same observation was reported in studies conducted in diverse geographical contexts [27, 41] . a previous study conducted in chilean children stated that the clinical severity in patients with single hadv infection and those with mixed infections was the same [47] . the overall finding is that the clinical value of such co-infections is not clear and still requires independent investigations in order to assess the association between co-infection and severe illness or symptoms. regarding the four years of surveillance, hadv circulation pattern shows no clear seasonality even if results suggest a higher activity of these viruses during cold periods. this lack of seasonality of hadv infection has been largely reported elsewhere [27, 48, 49] . however, seasonal peaks for hadv infection were noted in summer in some china areas [50] or in spring in northern china [51] , mexico [52] and taiwan [53] . in our study, the last 300 bp region of the hexon gene were used for molecular studies of the different hadv isolates. phylogenetic analysis showed that among the 54 sequenced strains hadv-c species were the most common hadv detected (81.5%) in patients with ili in senegal from 2012 to 2016. despite some divergences, the strains from senegal were close to types 1, 2, 5, 6 and 57. this hadv-c species predominance was reported in malaysia [23], in italy [54] , in many latina america countries [27, 52, 55] in contrast with studies done in the united states of america [56] , united kingdom [37] , korea [57] , in argentina [58] and china [59] , where hadv-b species were the most commonly isolated hadv. hadv-b species were the second most common in senegal with 9 strains, and only one type belonging to hadv-e species was sequenced. hadv-b species from senegal clustered with genome types hadv-7, hadv-7d, hadv-55 and hadv-11 (88% bootstrap value). hadv-7d serotype, firstly identified in 1980 in beijing [59] , is of particular concern as it was often associated with illnesses presenting with more severe and higher levels of morbidity than other respiratory hadv pathogens, and also may result in higher levels of fatalities [60] [61] [62] . the hadv-55 genome type, formerly known as hadv-11a, is a genotype resulting from recombination between hadv-11 and hadv-14 [63] . the serotype has recently reemerged as a highly virulent pathogen, causing severe [64] and sometimes fatal pneumonia among immunocompetent adults, particularly in asia [65] [66] [67] . so the circulation of such hadv genome types in senegal emphasizes the need to reinforce hadv surveillance, especially in hospitalized patients, by including hadv genome detection and genotyping in the documentation of severe respiratory infections. the single hadv-e species strain was typed as hadv-4, the unique human type in this species, which is more commonly associated with high rates of febrile respiratory illness in us military recruits [68] though associated with viral conjunctivitis outbreak in australia [69] for example. we observed some limitations in our study. first, considering the vast number of hadv positive samples, only a small number of hadv were typed. so the sequencing results do not reflect the full spectrum of hadv strains that may be circulating in ili patients in senegal, and even for selected samples it may have a bias toward samples with a high viral load. another limitation concerned the molecular methods used for typing hadvs in this study, a method which targeted a short hexon hypervariable region that has been shown to correlate closely with serotype. this method does not provide genomic detail and might miss recombination events located in other regions of the genome. therefore, full-genome sequencing would be more informative on senegalese strains, especially for hadv-b7 and hadv-b55 types. the results of this study should also be interpreted with caution especially for hadv ili causality (carriage in healthy or asymptomatic individuals). in conclusion, the results of the present study suggest strong year-round hadv activity in senegal, especially in children up to 5 years of age. molecular studies revealed that the dominant species in circulation in patients with ili appears to be hadv-c, hadv-b species. the circulation of though hadv-7d and hadv-55 genome types is of note as these serotypes are recognized causes of more severe and even fatal acute respiratory infections. so in the interest of global public health we strongly suggest molecular surveillance and genotyping of newly detected hadv strains in senegal and even by whole genome sequencing for some especial strains. our study offers also an important perspective on the burden of adenovirus-associated respiratory illness in senegal. such a perspective, especially among children, should include asymptomatic controls, sari cases, information on disease outcome, atypical clinical signs, duration of symptoms, and treatment. data regarding viral load, shedding, and other possible etiologies (e.g., bacterial and other viruses) would also enable a more thorough assessment of the viral effective disease (or symptom) causality. adenovirus infections in immunocompetent and immunocompromised patients family adenoviridae structural studies on adenoviruses worldwide epidemiology of human adenovirus infections adenovirus infections in transplant recipients clinical severity of respiratory adenoviral infection by serotypes in korean children over 17 consecutive years (1991-2007) molecular epidemiology of human adenovirus isolated from children hospitalized with acute respiratory infection in são paulo, brazil outbreak of adenovirus type 4 infection in a long-term care facility for the elderly rapid subgenus identification of human adenovirus isolates by a general pcr adenovirus type 7 genomic-type variant emergent us adenovirus 3 strains associated with an epidemic and serious disease influenza-like illnesses in senegal: not only focus on influenza viruses molecular characterization of adenovirus circulating in central and south america during the 2006-2008 period bioedit: a user-friendly biological sequence alignment editor and analysis program for windows 95/98/nt mega6: molecular evolutionary genetics analysis version 6.0 epidemiology of respiratory viral infections in two long-term refugee camps in kenya viral etiology and seasonality of influenza-like illness in gabon epidemiology of respiratory viral infection using multiplex rt-pcr in laboratory surveillance of influenza-like illness in seven teaching hospitals the clinical and etiological characteristics of influenza-like illness (ili) in outpatients in temporal changes in respiratory adenovirus serotypes circulating in the greater toronto area adenovirus respiratory tract infections in peru sentinel surveillance of influenza-like illness in two hospitals in respiratory viral infections during the 2009-2010 winter season in central england, uk: incidence and patterns of multiple virus co-infections viral etiology of severe acute respiratory infections in hospitalized children in cameroon the role of respiratory viral infections among children hospitalized for community-acquired pneumonia in a developing country respiratory virus detection and clinical diagnosis in children attending day care surveillance for severe acute respiratory infections in southern arizona epidemiology of human respiratory viruses in children with acute respiratory tract infections in jinan, china respiratory viral infection in lower airways of asymptomatic children viral and other infections of the human respiratory tract the epidemiology of adenovirus infections in greater manchester, uk 1982-96 estimates of world-wide distribution of child deaths from acute respiratory infections viral infections of the lower respiratory tract: old viruses, new viruses, and the role of diagnosis molecular identification of adenoviruses associated with respiratory infection in egypt from adenovirus infection in children with acute lower respiratory tract infections in beijing, china feigin rd, cherry jd. philadelphia: wb saunders adenovirus: epidemiology, global spread of novel serotypes, and advances in treatment and prevention. semin respir crit care med a community-derived outbreak of adenovirus type 3 in children in taiwan between fatal pneumonia associated with adenovirus type 7 in three military trainees long term sequelae from childhood pneumonia; systematic review and meta-analysis adenovirus and respiratory syncytial virus-adenovirus mixed acute lower respiratory infections in chilean infants comparative dynamics, morbidity and mortality burden of pediatric viral respiratory infections in an equatorial city community-acquired pneumonia requiring hospitalization among u.s. children molecular identification and epidemiological features of human adenoviruses associated with acute respiratory infections in hospitalized children in southern china molecular typing and epidemiology profiles of human adenovirus infection among paediatric patients with severe acute respiratory infection in china adenoviruses c in nonhospitalized mexican children older than five years of age with acute respiratory infection molecular and clinical characteristics of adenoviral infections in taiwanese children in 2004-2005 epidemiology and clinical characteristics of respiratory infections due to adenovirus in children molecular characterization of adenoviruses from children presenting with acute respiratory disease in the seattle virus watch. iv. observations of adenovirus infections lower respiratory tract infections due to adenovirus in hospitalized korean children: epidemiology, clinical features, and prognosis molecular typing of adenoviruses in pediatric respiratory infections in molecular epidemiology of adenovirus types 3 and 7 isolated from children with pneumonia in beijing molecular epidemiology of adenovirus type 7 in the united states re-emergent human adenovirus genome type 7d caused an acute respiratory disease outbreak in southern china after a twenty-one year absence fatal community-acquired pneumonia in children caused by re-emergent human adenovirus 7d associated with higher severity of illness and fatality rate. sci rep characterizing, typing, and naming human adenovirus type 55 in the era of whole genome data severe pneumonia associated with adenovirus type 55 infection outbreak of febrile respiratory illness associated with adenovirus 11a infection in a singapore military training camp epidemiology of human adenovirus and molecular characterization of human adenovirus 55 in china fatal pneumonia cases caused by human adenovirus 55 in immunocompetent adults respiratory diseases among u.s. military personnel: countering emerging threats outbreak of adenovirus type 4 conjunctivitis in south australia this study would not have been possible without the excellent support from all the health-care workers of the 4s network who contributes, every day, to the surveillance network. we convey our special thanks to kathleen victoir from the international network of pasteur institutes for her unwavering support to the 4s network. we acknowledge the senegalese ministry of health for their help in implementing the 4s network. key: cord-341260-vxsbv8t9 authors: loubet, p.; lenzi, n.; valette, m.; foulongne, v.; krivine, a.; houhou, n.; lagathu, g.; rogez, s.; alain, s.; duval, x.; galtier, f.; postil, d.; tattevin, p.; vanhems, p.; carrat, f.; lina, b.; launay, o.; seddik, k.; lesieur, z.; bonmarin, i.; loulergue, p.; bodilis, h.; servera-miyalou, m.; sadler, i.; momcilovic, s.; kanaan, r.; coolent, n.; tan boun, k.; blanche, p.; charpentier, j.; daviaud, f.; mongardon, n.; bretagnol, a.; claessens, y. e.; rozenberg, f.; yazdanpanah, y.; burdet, c.; harent, s.; lachatre, m.; rioux, c.; bleibtreu, a.; casalino, e.; choquet, c.; leleu, a.; belghalem, k.; colosi, l.; ranaivoson, m.; verry, v.; pereira, l.; dupeyrat, e.; bernard, j.; emeyrat, n.; chavance, p.; debit, a.; aubier, m.; pradere, p.; justet, a.; mal, h.; brugiere, o.; papo, t.; goulenok, t.; boisseau, m.; jouenne, r.; alexandra, j. f.; raynaud-simon, a.; lilamand, m.; cloppet-fontaine, a.; becheur, k.; pelletier, a. l.; fidouh, n.; ralaimazava, p.; beaumale, f.; costa, y.; munier, e.; betend, f.; amour, s.; loeffert, s.; francourt, k.; merle, c.; letois, f.; géraud, p.; driss, v.; noslier, s.; ray, m.; sebbane, m.; konaté, a.; bourdin, a.; klouche, k.; léglise, m. s.; couve-deacon, e.; fruit, d.; fenerol, c.; vallejo, c.; jouneau, s.; lainé, f.; thébault, e.; fillatre, p.; le pape, c.; beuzit, l.; chau, f.; goderel, i. title: clinical characteristics and outcome of respiratory syncytial virus infection among adults hospitalized with influenza-like illness in france date: 2017-04-30 journal: clinical microbiology and infection doi: 10.1016/j.cmi.2016.11.014 sha: doc_id: 341260 cord_uid: vxsbv8t9 abstract objectives the aim of this study was to analyse characteristics and outcome of respiratory syncytial virus (rsv) infection in adults hospitalized with influenza-like illness (ili). methods patients hospitalized with ili were included in this prospective, multicentre study carried out in six french hospitals during three consecutive influenza seasons (2012–2015). rsv and other respiratory viruses were detected by multiplex pcr in nasopharyngeal swabs. risk factors for rsv infection were identified by backward stepwise logistic regression analysis. results a total of 1452 patients hospitalized with ili were included, of whom 59% (861/1452) were >65 years and 83% (1211/1452) had underlying chronic illnesses. rsv was detected in 4% (59/1452), and influenza virus in 39% (566/1452). risk factors for rsv infection were cancer (adjusted or 2.1, 95% ci 1.1–4.1, p 0.04), and immunosuppressive treatment (adjusted or 2.0, 95% ci 1.1–3.8, p 0.03). patients with rsv had a median length of stay of 9 days (6–25), and 57% of them (30/53) had complications, including pneumonia (23/53, 44%) and respiratory failure (15/53, 28%). fifteen per cent (8/53) were admitted to an intensive care unit, and the in-hospital mortality rate was 8% (4/53). pneumonia was more likely to occur in patients with rsv than in patients with rsv-negative ili (44% (23/53) versus 26% (362/1393), p 0.006) or with influenza virus infection (44% versus 28% (157/560), p 0.02). conclusion rsv is an infrequent cause of ili during periods of influenza virus circulation but can cause severe complications in hospitalized adults. risk factors for rsv detection in adults hospitalized with ili include cancer and immunosuppressive treatment. specific immunization and antiviral therapy might benefit patients at risk. respiratory syncytial virus pneumonia adults influenza-like illness influenza elderly a b s t r a c t objectives: the aim of this study was to analyse characteristics and outcome of respiratory syncytial virus (rsv) infection in adults hospitalized with influenza-like illness (ili). methods: patients hospitalized with ili were included in this prospective, multicentre study carried out in six french hospitals during three consecutive influenza seasons (2012e2015). rsv and other respiratory viruses were detected by multiplex pcr in nasopharyngeal swabs. risk factors for rsv infection were identified by backward stepwise logistic regression analysis. results: a total of 1452 patients hospitalized with ili were included, of whom 59% (861/1452) were >65 years and 83% (1211/1452) had underlying chronic illnesses. rsv was detected in 4% (59/1452), and influenza virus in 39% (566/1452). risk factors for rsv infection were cancer (adjusted or 2.1, 95% ci 1.1 e4.1, p 0.04), and immunosuppressive treatment (adjusted or 2.0, 95% ci 1.1e3.8, p 0.03). patients with rsv had a median length of stay of 9 days (6e25), and 57% of them (30/53) had complications, including pneumonia (23/53, 44%) and respiratory failure (15/53, 28%) . fifteen per cent (8/53) were admitted to an intensive care unit, and the in-hospital mortality rate was 8% (4/53). pneumonia was more likely to occur in patients with rsv than in patients with rsv-negative ili (44% (23/53) versus 26% (362/1393), p 0.006) or with influenza virus infection (44% versus 28% (157/560), p 0.02). over the past decade, respiratory syncytial virus (rsv) has been increasingly recognized as an important pathogen in adults [1] , and especially the elderly [2e4], immunocompromised patients [5, 6] and individuals with underlying chronic respiratory diseases [7, 8] . in the elderly population, rsv is one of the three most common causes of respiratory disease, along with influenza virus and rhinovirus [3, 7] . respiratory virus detection has improved with the development of highly sensitive and specific molecular methods, which are providing interesting new epidemiological information. however, for reasons of cost and the lack of rsv-specific antiviral drugs, adult inpatients with influenza-like illness (ili) are frequently not tested for rsv but rather only tested for influenza virus. influenza surveillance systems based on ili and acute respiratory illness definitions are unsuitable for capturing cases of rsv in children and adults [9e11] . more studies are needed to determine the burden of rsv infection, and to characterize at-risk populations that could be targeted by vaccination and antiviral treatment. the aims of this study were to describe (a) the prevalence, (b) the clinical features, and (c) the outcome of rsv infection in adults hospitalized with ili. we analysed cases of laboratory-confirmed rsv infection during three consecutive influenza seasons (2012/13, 2013/14 and 2014/ 15), in a post hoc analysis of patients hospitalized with ili in the fluvac study. fluvac is a french prospective observational study of influenza vaccine efficacy conducted in six university hospitals (cochin hospital, paris; bichat hospital, paris; pontchaillou hospital, rennes; limoges hospital; university hospital, montpellier; edouard herriot hospital, lyon). the fluvac study design is further described in rondy et al. and loubet et al. [12, 13] . each season, enrolments take place during periods of influenza circulation (from november to march). data on adults hospitalized for at least 24 h for ili, with symptom onset <7 days before sampling, were collected. ili was defined as a combination of the following: (a) at least one of the following systemic symptoms: fever (38 c), headache, myalgia or malaise, and (b) at least one of the following respiratory symptoms: cough, sore throat or dyspnoea [14] . each participant was interviewed, and nasopharyngeal samples were obtained at enrolment. we collected demographic characteristics, chronic underlying diseases and their treatments, the characteristics of the current ili episode: clinical presentation, hospitalization ward, the length of hospital stay and outcome (occurrence of complication, intensive care unit admission and death). data were retrieved from the medical charts, interviews with the patients and families, and laboratory databases. all variables collected are detailed in the supplementary material (appendix s1). respiratory viruses were detected in nasopharyngeal swabs from all the patients by means of multiplex rt-pcr. any bronchoalveolar lavage fluid samples or tracheal aspirates ordered by the physician in charge were also tested. samples were first tested in the virology laboratory of each participating hospital by means of real-time influenza a and b pcr after manual nucleic acid extraction. all samples were then sent to the french national influenza reference centre (cnr-lyon) for influenza confirmation and screening for other respiratory viruses. rna and dna were extracted with the automated easymag system from biom erieux (marcy l'etoile, france), and influenza viruses were detected with an in-house real-time rt-pcr protocol [15] . the samples were also screened for a panel of other respiratory viruses (adenoviruses, bocaviruses, coronaviruses, human metapneumovirus, parainfluenza viruses 1e4, picornavirus and rsv) by real-time pcr using the respiratory multiwell system rgene ® on an abi 7300 analyser. we first described the characteristics of all the patients hospitalized with ili. then, univariate analysis was used to compare rsvpositive patients with (a) rsv-negative patients (including patients infected by other respiratory viruses and patients free of viral infection) and (b) influenza virus-positive patients. quantitative variables were expressed as mean and standard deviation (sd) or median and interquartile range (iqr), and qualitative variables as number and percentage. we used the wilcoxon rank sum test or fisher's exact test for univariable comparisons. missing data for each variable were excluded from the denominator. factors associated with rsv infection were identified by using rsv-negative individuals as the comparison group. individuals with influenza virus and rsv co-infection were excluded from the analysis. we used a backward stepwise logistic regression model, with rsv test results (positive/negative) as the dependent variable. all covariates with a p value <0.2 in univariate analysis were tested in the multivariate model, namely age (considered as a binary variable (<65 and 65 years)), dyspnoea, fever, myalgia, chronic lung disease, cancer (solid and haematological malignancies), diabetes, chronic renal failure, and immunosuppressive treatment. factors associated with pneumonia onset were analysed among all ili patients. individuals with influenza virus and rsv co-infection were excluded from the analysis. we used a backward stepwise logistic regression model in which pneumonia (positive/negative) was the dependent variable. covariates with a p value <0.2 in univariate analysis were tested in the multivariate model, namely age (considered as a continuous variable), chronic heart disease, chronic respiratory disease, cancer (solid and haematological malignancies), diabetes, chronic renal failure, immunosuppressive treatment, rsv infection, influenza virus infection, and influenza vaccination. the final model was adjusted for chronic respiratory disease and age because of their known role in the onset of pneumonia. results from both regression models were expressed as odds ratios (or) and adjusted ors (aor) with their 95% ci. a p value of 0.05 was considered statistically significant. all analyses used stata software (v12, © copyright 1996e2014 statacorp lpt, college station, tx, usa). the fluvac study (clinicaltrials.gov nct02027233) respected good epidemiological and clinical practices in clinical research, and the declaration of helsinki, and was approved by regional ethics committees. all the study participants gave their informed consent for respiratory virus testing. overall, 1452 patients hospitalized with ili were included during the three periods of influenza circulation (fig. 1 fig. 2 ). respiratory syncytial virus was the third most frequent virus, after influenza virus (39% (566/1452) of patients with ili, 73% (566/777) of patients with at least one virus) and picornavirus (5% (68/1452) of patients with ili, 9% (68/777) of patients with at least one virus). the other detected viruses were coronavirus (3.5% (51/1452) and 7% (51/777) respectively), human metapneumovirus (3% (41/1452) and 5% (41/777)), adenovirus (1% (18/ 1452) and 2% (18/777)) and bocavirus (0.5% (8/1452) and 1% (8/ 777)). six patients (6/1452, 0.4%) were diagnosed with both influenza virus and rsv infection and so were removed from further analyses. the median age of the 53 patients with rsv infection alone was 74 years (iqr, 61e84) ( table 2) . chronic underlying diseases were present in 45 cases (45/53, 85%), and consisted mainly of chronic respiratory diseases (29/53, 55%), chronic heart disease (24/53, 45%) and cancer (18/53, 34%; 12 solid tumours, 6 haematological malignancies). fifteen patients (15/53, 28%) were on immunosuppressive therapy. twenty-six patients (26/53, 49%) had been hospitalized in the previous year, an average of 1.4 times (sd 2.3). the median time from symptom onset to admission was 2 days (iqr, 1e3). cough (43/53, 81%), fever (44/53, 83%) and dyspnoea (45/53, 85%) were the main symptoms in patients with rsv infection. the median length of hospital stay was 9 days (iqr 6e25). a total of 55 medical complications occurred in 30 patients (30/52, 58%) during their hospital stay, including pneumonia (23 episodes, 42% (23/55) of complications), respiratory failure (15/55, 27%), heart failure (10/55, 18%), and acute respiratory distress syndrome (7/55, 13%). intensive care unit admission was necessary for eight patients (8/53, 15%). four patients (4/53, 8%) died during the hospital stay; all of them were men >65 years old with chronic respiratory diseases. furthermore, three patients had chronic heart disease. the mean time from admission to death was 13 days (sd 2.9). patients with rsv were older than patients with influenza virus (74 years (iqr 1e84) versus 68 years (iqr 52e81), p 0.05) ( table 2) . patients with rsv were more likely to have cancer or immunosuppressive treatment than patients without rsv (respectively 34% (18/53) versus 16% (219/1393), p 0.002; 28% (15/53) versus 15% (205/1393), p 0.003) and influenza virus-positive patients (34% (18/ 53) versus 13% (74/560), p 0.004; 28% (15/53) versus 14% (78/560), p 0.01). multivariate analysis of all patients with ili showed that cancer (or 2.1; 95% ci 1.1e4.1, p 0.04) and immunosuppressive treatment (or 2.0; 95% ci 1.1e3.8, p 0.03) were significantly associated with rsv detection ( table 3 ). the patients with rsv and influenza virus infection did not differ in terms of the length of stay, icu admission or mortality. after adjustment for chronic respiratory disease and age, we found that rsv infection (or 2.1; 95% ci 1.2e3.8, p 0.008), chronic renal failure (or 1.8; 95% ci 1.3e2.5, p 0.001) and active smoking (or 1.3; 95% ci 1.0e1.7, p 0.02) were significantly associated with the onset of pneumonia (table 4 ). in this study conducted during three consecutive influenza seasons among 1452 adults hospitalized for ili in france, rsv was the third most common respiratory virus, being detected in 4% of patients, compared with 39% for influenza virus. this rate is lower than that found in the usa by sundaram [7] . in the latter study, the rate was even higher (8/64, 12.5%) among hospitalized patients. in another study, falsey et al. reported a 10% (142/1388) prevalence of rsv among adults aged 65 years or with underlying cardiopulmonary diseases who were admitted with acute respiratory symptoms during four consecutive winters in rochester, ny [1] . several reasons may explain the broad range of reported rsv detection rates. for example, our study was restricted to periods of influenza virus circulation, which may not have included the peak of rsv circulation, although enrolments in the two studies by falsey et al. started in mid-november. another factor is age: we included all adults >18 years, whereas sundaram et al. and falsey et al. included only older subjects, in whom rsv infection may be more frequent. furthermore, the laboratory methods used in the rochester study included viral culture and serological tests, and rt-pcr was the only positive test in only two-thirds of cases. finally, the clinical definition used to trigger swab collection also differed among the studies. the median age of the rsv-infected patients in our study was 74 years but, contrary to other studies, we found no association between rsv infection and older age [3, 16, 17] . most of the patients with rsv infection had chronic underlying conditions (45/53, 85%), consisting mainly of chronic respiratory disease (29/53, 55%) or chronic heart disease (24/53, 45%). this is consistent with the report by walsh et al. that underlying pulmonary disease was a significant risk factor for severe rsv illness requiring hospitalization [18] . we found that the two underlying conditions independently associated with rsv infection were cancer and immunosuppressive treatment. this association was statistically significant whether rsv-positive patients were compared with all rsv-negative patients or only with influenza virus-positive patients. although immunocompromised patients (especially patients with haematological malignancies) are known to be more susceptible to rsv infection [19] , solid cancers and immunosuppressive therapy are not classical risk factors. respiratory syncytial virus was associated with significant morbidity: the median length of hospital stay was 9 days; 15% (8/ 53) of rsv-infected patients were admitted to the intensive care unit, and 8% (4/53) died. these findings are consistent with the literature [1, 20] . similar percentages were noted in the influenza group of our study. patients with rsv were significantly more likely than patients with influenza or without rsv to develop pneumonia (44% versus 28% and 26%, respectively). falsey et al. and lee et al. also found a high rate of pneumonia among 159 rsv-infected patients >65 years admitted for ili in the usa (70/159, 44%) and among 607 rsv-infected adults admitted to three hospitals in hong kong (261/603, 43%) [20, 21] . jain et al. [22] found at least one respiratory virus in 23% of 2259 american adults with communityacquired pneumonia requiring admission to one of five participating hospitals between 2010 and 2012. rsv was found in 3% (67/ 2259) of all patients. the strengths of this study include the large number of adults hospitalized with ili, the prospective multicentre design, standardized patient screening in the participating centres, centralized confirmation of respiratory viruses in an influenza reference centre and the lengthy study period spanning three consecutive influenza seasons. several limitations must, however, be acknowledged. first, as all the participating centres were teaching hospitals, the proportion of patients with underlying diseases may have been higher than in the general ili population, owing to a referral bias. second, although the sample was large, the study was probably underpowered to identify a possible impact of rsv on morbidity and mortality in multivariate analysis. third, the ili definition used here captures only a subset of rsv infections [10] . our study does not reflect the real burden of severe rsv infection, which may include other clinical manifestations. fourth, we report post hoc results of the fluvac study, which was not designed to answer this research question. indeed, as the fluvac study was designed to assess influenza vaccine efficacy, patients were enrolled during periods of influenza virus circulation, which differ slightly from periods of rsv circulation. this means that our data reflect the prevalence of rsv among patients hospitalized for ili during periods of influenza virus circulation and not during peak rsv circulation, which usually occurs earlier, owing to epidemiological interference [23] . however, french surveillance data (renal system) show that the peak of the rsv epidemic overlapped with the beginning of influenza virus circulation during the three seasons of interest, and that the two viruses co-circulated for at least 2e3 weeks (fig. 3) . in conclusion, this prospective observational study conducted in france during three influenza seasons reveals that 4% of adults hospitalized for ili had rsv infection, of whom 58% developed cardiopulmonary complications and 8% died. it also shows that elderly individuals and patients with cancer and/or immunosuppressive treatment are more likely to have rsv isolated when hospitalized for ili. potential benefits of enhanced rsv testing, antiviral treatment, and vaccine development in these groups should be considered. the outcome analysed was rsv infection. the authors declare no competing interest related to the study. o launay is an investigator for clinical trials sponsored by janssen and other companies and received travel support to attend scientific meetings from pharmaceutical companies. the current work received no funding. however, the study sites received funding from sanofi pasteur and sanofi pasteur msd for the fluvac study. vaccine producers had no role in the study design, data analysis, decision to publish or preparation of the manuscript. respiratory syncytial virus infection in elderly and high-risk adults prevalence and clinical characteristics of human respiratory syncytial virus in chinese adults with acute respiratory tract infection medically attended respiratory syncytial virus infections in adults aged 50 years: clinical characteristics and outcomes seasonal incidence of medically attended respiratory syncytial virus infection in a community cohort of adults 50 years old morbidity and mortality among patients with respiratory syncytial virus infection: a 2-year retrospective review community respiratory virus infections following lung transplantation respiratory syncytial virus and other respiratory viral infections in older adults with moderate to severe influenza-like illness risk factors for respiratory syncytial virus illness among patients with chronic obstructive pulmonary disease evaluation of case definitions for estimation of respiratory syncytial virus associated hospitalizations among children in a rural community of northern india the diagnosis of viral respiratory disease in older adults progress in the surveillance of respiratory syncytial virus (rsv) in europe seasonal influenza vaccines effectiveness against confirmed a(h3n2) influenza hospitalisation: pooled analysis from a european network of hospitals. a pilot study factors associated with poor outcomes among adults hospitalized for influenza in france: a three-year prospective multicenter study available at pandemic a(h1n1)2009 influenza virus detection by real time rt-pcr : is viral quantification useful? rates of hospitalizations for respiratory syncytial virus, human metapneumovirus, and influenza virus in older adults hospitalizations associated with influenza and respiratory syncytial virus in the united states risk factors for severe respiratory syncytial virus infection in elderly persons respiratory syncytial virus infection in adults high morbidity and mortality in adults hospitalized for respiratory syncytial virus infections respiratory syncytial virus and influenza a infections in the hospitalized elderly community-acquired pneumonia requiring hospitalization among u impact of the 2009 influenza a (h1n1) pandemic wave on the pattern of hibernal respiratory virus epidemics we thank all the renal network laboratories and the cnr laboratory for agreeing to share their data on influenza and rsv epidemics. we specifically thank vincent enouf (cnr influenza, paris) for creation of fig. 3 .we are very grateful to the following persons and institutions who made significant contributions to the fluvac study: r eseau the authors thank h el ene bricout, laurence pagnon and christine sadorge from sanofi pasteur msd for their inputs during the study design phase and for critical review of the results. additional supporting information may be found in the online version of this article at http://dx.doi.org/10.1016/j.cmi.2016.11.014. key: cord-263464-fdosch11 authors: nuvey, francis sena; edu-quansah, elijah paa; kuma, george khumalo; eleeza, john; kenu, ernest; sackey, samuel; ameme, donne; abakar, mahamat fayiz; kreppel, katharina; ngandolo, richard bongo; afari, edwin; bonfoh, bassirou title: evaluation of the sentinel surveillance system for influenza-like illnesses in the greater accra region, ghana, 2018 date: 2019-03-14 journal: plos one doi: 10.1371/journal.pone.0213627 sha: doc_id: 263464 cord_uid: fdosch11 background: influenza-like illness (ili) is a medical diagnosis of possible influenza or another respiratory illness with a common set of symptoms. the deaths of four schoolchildren, during a pandemic influenza outbreak in december 2017 in ghana, raised doubts about the ili surveillance system’s performance. we evaluated the ili surveillance system in the greater accra region, ghana, to assess the system’s attributes and its performance on set objectives. methods: cdc guidelines were used to evaluate the data of the ili surveillance system between 2013 and 2017. we interviewed the surveillance personnel on the system’s description and operation. additionally, routinely entered ili data from the national influenza center provided by the six sentinel sites in accra was extracted. we sampled and reviewed 120 ili case-investigation forms from these sites. surveillance activities were examined on system’s performance indicators, each being scored on a scale of 1 to 3 (poorest to best performance). results: all population and age groups were under ili surveillance over the period evaluated. overall, 2948 suspected case-patients, including 392 (13.3%) children under-five were reported, with 219 being positive for influenza virus (predictive value positive = 7.4%). the predominant influenza subtype was h3n2, recorded in 90 (41.1%) of positive case-patients. the system only met two out of its four objectives. none of the six sentinel sites consistently met their annual 260 suspected case-detection quota. samples reached the laboratory on average 48 hours after collection and results were disseminated within 7 days. of 120 case-investigation forms sampled, 91 (76.3%) were completely filled in. conclusions: the ili surveillance system in the greater accra region is only partially meeting its objectives. while it is found to be sensitive, representative and timely, the data quality was sub-optimal. we recommend the determination of thresholds for alert and outbreak detection and ensuring that sentinel sites meet their weekly case-detection targets. introduction influenza-like illnesses (ili), often also called acute respiratory infection or flu-like syndrome, are acute viral infections of the respiratory tract with similar signs and symptoms to influenza. ili is a syndrome and affected persons may become infectious before, during or after the onset of symptoms. the pathogen can be transmitted both directly (by droplets) and indirectly through contact with contaminated fomites. children, the elderly and pregnant women, as well as persons with chronic illnesses or immunosuppression are at the highest risk for morbidity and mortality from ilis [1, 2] . according to the world health organization (who), most influenzas in the global circulation are of zoonotic origin. sub-types implicated in epidemics include h1n1, h5n1, h7n9, h7n7 and h3n2 [3] . influenza has a global annual attack rate of 5-10% in adults and 20-30% in children, causing between 3-5 million cases of severe illness and about 500,000 deaths yearly [4] . pandemics of influenza have had high fatality rates in the past and robust surveillance systems are key to global efforts to prevent similar outbreaks [5] . due to the epidemic-prone nature of influenza pathogens and their high propensity for mutations, the world health organization (who) recommends strict adherence to infection control and prevention measures including increased handwashing during peak flu seasons [2] . in ghana, laboratory-based surveillance for respiratory tract infections (rtis) only tests for influenza in suspected cases. rtis remain a major cause of morbidity and mortality in ghana ranking second among the top 10 diseases seen at outpatient departments (opds) in healthcare facilities across the country, in 2016 [6] . respiratory diseases may occur because of an invasion of a susceptible host by microbes including bacteria and viruses. three main types of influenza viruses exist, namely; influenza a, b, and c but epidemics are often linked to the influenza a strain [7] . influenza surveillance data between 2012 and 2014 in ghana indicated 1041 positive influenza cases out of a total 8601 respiratory samples tested, with 6 different subtypes [influenza a (h3, h1n1, h1, h5) and influenza b (victoria, yamagata)] identified [8] . ghana began influenza surveillance in 2007 to obtain data on strains in circulation. the national influenza center (nic) was formally recognized by the who in 2010 after the 2009 pandemic influenza [9] . the nic is a member of the who global influenza surveillance and response system (gisrs) and is located in the department of virology of the noguchi memorial institute of medical research (nmimr). in collaboration with the ghana health service (ghs) and the ministry of defense (mod), it currently operates sentinel surveillance for influenza in 27 sites across all regions in ghana with support from the u.s. naval medical research unit no. 3 (namru-3), centers for disease control and prevention (cdc) and who [8] . ili surveillance is conducted all year round across the sentinel sites. the ili surveillance system aims to detect early unusual events indicating a change in circulating influenza sub-types, identify and monitor vulnerable groups for influenza, determine influenza thresholds and detect antiviral resistance. in december 2017, an outbreak of influenza in a school in the ashanti region of ghana was followed by the death of four children (case fatality rate = 5.2%) [10] . this raised concerns about the effectiveness of the influenza surveillance system, particularly that in the greater accra region of ghana, which recorded no alerts over the past five years. we evaluated the effectiveness of the ili sentinel surveillance system to determine if its objectives are being met and to assess its attributes and usefulness. the findings of this study are important in order to give useful recommendations to improve the current system. we described the attributes and effectiveness of the ili sentinel surveillance system in the greater accra region (gar) of ghana using a descriptive cross-sectional survey. we extracted and evaluated routinely recorded ili data between january 2013 and december 2017. the gar is one of ten regional demarcations in ghana with a population of about 5 million people [11] . as the most densely populated region in ghana, it is mainly an urban settlement and lies in the southeastern part of the country along the coast of the gulf of guinea. it is the administrative capital of ghana with over 500 public and private health care facilities. six of these facilities in the region conduct ili surveillance including four ghs facilities, one military and one quasi-government facility (see fig 1) . they fall under the three main levels of healthcare delivery in ghana; primary, secondary and tertiary. the manhean health center provides primary health services, while secondary healthcare providers involved in ili surveillance are tema polyclinic, achimota hospital and university of ghana hospital, legon. lastly, the greater accra regional hospital (garh) and 37 military hospital provide tertiary care. data on ili from all six sentinel facilities in gar was extracted and abstracted from the nic database into microsoft office excel format. additionally, we selected three sentinel facilities, based on the different levels of care they provided, and obtained permission for site visits. these sentinel sites; garh, achimota hospital and manhean health center, were visited for at least one week each and their records were collected and reviewed. the researchers interviewed all personnel directly involved in ili surveillance and also partook in surveillance activities while observing practices. the staff members were interviewed using a structured questionnaire and guided interviews on their knowledge on ili surveillance activities. forty case investigation forms were randomly sampled from each of the three facilities, using microsoft excel. the unique serial numbers on case investigation forms for each sentinel site were used to retrieve the randomized forms for examination. in addition, we collected and reviewed ili registers at the sites. the evaluation was conducted using the cdcs guidelines for evaluating public health surveillance systems [12] . nine attributes: simplicity, flexibility, data quality, acceptability, sensitivity, predictive value positive, representativeness, timeliness and stability, usefulness and the utility of the system to achieve its objectives were evaluated. the indicators for each attribute or characteristic measured were scored one point each. based on the evaluator's assessment, an indicator is scored zero (0) if the key finding from evaluation do not support the indicator assessed in relation to the attribute. a score of one (1) is given to each indicator that supports the attribute assessed. the assessment scores were then summed and divided by the total number of indicators used in evaluating each attribute. attributes with relative scores of more than twothirds of the total score, were considered as major strengths of the system and scored 3 overall. those with less than one-third were major weaknesses (overall score = 1). scores between one-third and two-thirds were relative strengths (overall score = 2). scores were on a scale of 1 to 3: poorest to best performance, respectively. the evaluation was done within the framework of integrated disease surveillance and response matrix implemented by the ghana health service and therefore did not have to receive formal review by ethical review committees. the field epidemiology and laboratory training programme and national influenza center in ghana approved the study. permission was sought and obtained from the public health directorate of ghana health service and authorities in the sentinel facilities before commencement of the evaluation. all respondents provided informed, written consent and were assured of confidentiality. the surveillance system was found to be utilizing the syndromic approach by screening suspected cases and thereafter conducting further laboratory confirmatory tests on collected nasopharyngeal or oropharyngeal specimen. a reverse transcriptase-polymerase chain reaction (rrt-pcr) is used to confirm real influenza cases from among those suspected using the influenza case definition and determine the influenza virus sub-type. we found that data on patients meeting the ili case definition (s2 table) from the sentinel sites are collected together with nasopharyngeal or oropharyngeal specimen. specimen are stored in sample bottles with virus transport media (vtm) in a cold chain system and submitted to nmimr for laboratory confirmation. each site is required to suspect five cases weekly using the case definition. thus, the expected annual total for each site is 260 suspected cases. firstly, clinicians at the opds suspect cases and trained personnel (i.e. nurses, laboratory scientists and surveillance officers) collect the specimen for storage in designated refrigerators at the disease control units. case detection is, in most cases, directly done by ili surveillance team members at the opds. socio-demographic, epidemiologic and clinical data are collected on each case using case investigation forms. the nic supplies the sentinel sites with the tools including case investigation forms, vtm, and specimen bottles, for the system's operation. they contact the sentinel sites via phone calls and visit the sentinel sites in the gar at least twice every week to pick up specimen. the nic reports results from the polymerase chain reaction laboratory test to each sentinel site, the national surveillance department (nsd), and who via e-mail. data is entered into the flu-net system, a who (gisrs) global database specifically designed for influenza surveillance, weekly. sentinel sites receive printed laboratory results from the nic averagely every 7 days. at the facility level, health information units (hiu) enter the data into the district health information management system 2 (dhims 2). ili data entry into dhims 2 started in 2017. the ghs, nic and other stakeholders periodically hold review meetings to discuss influenza activities and publish the data generated by the system in reports and journals. the information and data flow in the ili surveillance system is further illustrated in fig 2. of the 2948 suspected cases tested, 219 (7.4%) cases were tested positive for influenza viruses. s1 table) . utility of the ili sentinel surveillance system in attaining set objectives. the ili surveillance system has four objectives: early detect events that show change in severity or patterns of influenza infection or emergence of new strains, determine influenza thresholds, identify and monitor at risk groups and detect antiviral resistance in circulating strains. the system met two out of its four objectives over the period evaluated. it was able to detect and characterize influenza viruses in circulation, as from influenza a and b lineages, and established the groups of persons most at risk of influenza infection. it however did not set thresholds, which is the minimum number of suspected cases above which the system is alerted, and was not performing antiviral resistance testing. surveillance system attributes. usefulness: data generated by the system informed the choice of vaccines used in controlling influenza outbreaks in ghana. in addition, who and cdc use the system's information to monitor influenza activity globally, through sharing of confirmed influenza samples. tables 1 and 2 show the key findings on the indicators and scores for all the qualitative and quantitative attributes evaluated respectively. sensitivity: the ili system in gar has been able to confirm influenza cases in each of the years evaluated (s1 table) . evidence of aggregate data from ghana can also be found in the who flunet system, available free at http://apps.who.int/flumart/default?reportno=1. predictive value positive (pvp): the predictive power of the ili case definition is tested using rrt-pcr. overall pvp for the period was 7.4% (range 4.7%-14.8%). see s3 table for yearly pvps over the evaluated period. simplicity: surveillance personnel interviewed identified case definition to be simple. only opds partake in ili surveillance and follow-up of confirmed cases done within 5 days. however, specialized training is required for specimen collection and laboratory confirmation, data collected on each suspected case is comprehensive and laboratory testing is not done at sentinel sites. flexibility: modification of the ili system in 2016, enabled detection of severe acute respiratory infections (sari) with no difficulty. furthermore, addition of other common respiratory symptoms to case definition at the sites did not disrupt the system. data quality: key surveillance information was completely provided in 76% (91/120) of case investigation forms sampled and data extracts were comparable at sentinel facilities and the nic. in spite of this, only two sites: garh and achimota hospital, had 2017 influenza data table 1 entered in the dhims-2 platform and laboratory results were not entered in ili registers at the three sites. acceptability: on average, five out of the six sentinel sites detect influenza cases each year (83%) and about 80% of case investigation forms sampled were completely filled out by ili personnel. all sentinel staff interviewed reported satisfaction with feedback from nic and test results from collected specimen are on average tested within 2 days. most of the sites (5/6) however, do not meet the case detection quota for each year with some recording as low as three suspected cases per year only (see fig 3) . representativeness: age and sex distribution of total cases detected reflect the general distribution of ghana's population. the median age of cases was 30 years (range: 3 weeks to 90 years). females constituted 60% (1775/2948) of cases. timeliness: it takes on average 10 days between symptom onset and detection at facilities. the majority of specimens are tested within 48 hours after collection and results are disseminated within 7 days. this conforms with the who set standard timelines for influenza surveillance. stability: data flow in the system conforms to set standards; testing is routinely done within 48 hours after case detection and results are released within 7 days to the sites on average. the system is sustainable with donors mainly funding its operation. there was not a time influenza surveillance in the greater accra region, ghana during the period evaluated when no site recorded cases, but all sites detected cases only for 60% of the period. influenza causes considerable morbidity and about 500,000 deaths per year globally. the disease is highly infectious with high pandemic potential. therefore, worldwide surveillance systems to record influenza-like illnesses (ili) in real time and detect possible influenza outbreaks are essential to prevent and control epidemics. ghana's ili surveillance system since its inception in 2007, aims to early detect changes in circulating influenza, identify vulnerable groups to influenza infection, determine influenza thresholds and detect antiviral resistance to influenza viruses. the late detection of a pandemic influenza outbreak in ghana raised doubts about the ili surveillance system's performance. our study provides evidence, that the ili sentinel surveillance system in the greater accra region (gar), ghana, is only partially meeting its objectives because it did not have thresholds for alerting the health system and does not perform antiviral resistance testing. it is sensitive and timely in detecting influenza cases but of low predictive value positive (pvp). it is representative of the population under surveillance and flexible to modifications. the system is fairly stable and acceptable to key stakeholders; the quality of data is relatively high. predominant influenza subtype in circulation is influenza a (h3n2) virus. the sentinel sites consistently failed to meet their case detection quotas annually over the period evaluated. despite these apparent weaknesses in the ili system, the good performance of the laboratory component is commendable and key to detection of novel influenza viruses for prompt response. even though the who's standards for influenza surveillance alluded to the possibility of resource limitation hindering achievement of all influenza surveillance system objectives, it advocates for influenza surveillance systems capable of collecting the minimum amount of data needed for decision making [5] . for the past two decades of influenza surveillance in ghana, the system proved its utility by its ability to detect and classify circulating strains as well as providing key information for public health action. similar findings were made in the south african [13] and madagascar [14] influenza surveillance systems. even though the pvp is low, it generally conforms to other syndromic surveillance systems with broad case definitions but specific for respiratory diseases in this case, aimed at maximizing influenza case detection [15] . nevertheless, the unmet objectives, lack of thresholds and antiviral resistance testing, of the system require attention. the lack of antiviral resistance testing to detect the emergence of treatment resistant strains and the absence of thresholds preventing the issuing of alerts, are a major drawback. this situation is not specific to the gar alone but is found in all sites in ghana. in addition, the mainly conservative management of clinical signs and symptoms, of confirmed influenza cases in the health system, without the use of recommended antiviral agents, may partly explain the absence of antiviral susceptibility testing. these shortfalls may, however, be caused by a lack of capacity or resources (antiviral drugs and laboratory reagents) at the sentinel facilities to determine thresholds for influenza alerts and test for antiviral resistance, as was observed in evaluations done in other settings [16, 17] . threshold establishment for disease surveillance is paramount to be able to alert the health system early when outbreaks occur for prompt public health action [1] . a study to test the moving epidemic method in determining thresholds for ili and sari surveillance systems in europe, was able to detect epidemic periods with few or no false alarms in different countries [18] . other methods employed in cambodia [19] and australia [20] had similar findings. the nic must take the lead using the who manual [21] , to choose an epidemic threshold determination method and train focal persons in the facilities to use it, to ensure this key information is available. even though implementing an integrated (human and animal) surveillance could be costeffective and improve case detection and response [22] , discussions with ili surveillance personnel at sentinel sites revealed the absence of a link between influenza surveillance in humans and animals. human infection with zoonotic influenza strains is possible and may cause mild to severe form of the disease. there have been recorded episodes of pandemics in humans over a century due to cross-species transmission of zoonotic respiratory viruses including influenza viruses, notably spanish flu (1918), severe acute respiratory syndrome (2003) and swine flu (2009), resulting in high morbidity and mortality globally [23, 24] . thus, the importance of a one-health approach in the surveillance and response with control of pandemics in an increasingly globalized world is evident. research findings in africa, europe and the americas, have shown incremental gains derived when interventions are integrated between human and animal health systems [25] . the nic and veterinary service department in ghana should collaborate to formalize protocols for engaging each other to integrate influenza surveillance systems in humans and animals, using a one health approach, as this can provide additional key information on possible cross-species transmission in the country and enhance savings. the usage of one laboratory to test for both human and animal pathogens in canada was shown to cost about 30% less than the combined original operational costs of testing individually in both laboratories [26] . in ghana, the timeliness of the system in case detection, laboratory confirmation and result dissemination is commendable. this strength may be as result of the weekly visits nic makes to sentinel sites to collect specimen, supply virus transport media in specimen bottles and transfer results. the nic should further take advantage of these visits and collaborative meetings to ensure each site meets weekly case detection quotas. focal persons at the various sites must take the lead. in spite of the usefulness and fair performance of ili surveillance in gar on indicators evaluated, it is only partially meeting its set objectives. it is sensitive in detecting circulating influenza types, representative of the population under surveillance and timely. however, sentinel sites do not consistently meet annual case detection quotas. there is the need to address shortfalls in the system's objectives as well as improving case detection at sentinel facilities. this would ensure that the successes chocked are not undermined, thereby preventing increasing morbidity and mortality related to influenza infections. failure of the system to address these shortfalls would also affect ghana's contribution to the who global influenza surveillance and response system. considering the zoonotic character of most influenza viruses, it is important that a one health approach is adopted with influenza surveillance in ghana. we propose strict adherence to case detection targets by individual sentinel sites and determination of alert thresholds for the system to allow for a more effective monitoring of influenza activity in the region for prompt public health actions. supporting information s1 table. ili case definitions used for screening and enrolment by the ili sentinel surveillance technical guidelines for integrated disease surveillance and response in the african region world health organization. community case management during an influenza outbreak: participant's handbook influenza at the human-animal interface; sumary and assessment influenza (seasonal) fact sheet [internet]. who. world health organization world health organization. global epidemiological surveillance standards for influenza virological surveillance of influenza-like illness among children in ghana ghana health service/ministry of health. interim report: surveillance response to influenza type a (h1n1) outbreak in kumasi academy, asokore -mampong municipality population projections by districts updated guidelines for evaluating public health surveillance systems: recommendations from the guidelines working group evaluation of two influenza surveillance systems in south africa evaluation of the influenza sentinel surveillance system in madagascar overview of syndromic surveillance: what is syndromic surveillance? morb mortal wkly rep strategy to enhance influenza surveillance worldwide a summary of influenza surveillance systems in australia influenza surveillance in europe: establishing epidemic thresholds by the moving epidemic method establishing seasonal and alert influenza thresholds in cambodia using the who method: implications for effective utilization of influenza surveillance in the tropics and subtropics exploring a proposed who method to determine thresholds for seasonal influenza surveillance world health organization. who interim global epidemiological surveillance standards for influenza creating a framework towards integrated health syndromic surveillance and response in africa global burden of influenza: contributions from resource limited and low-income settings sars, the first pandemic of the 21st century one health: the theory and practice of integrated health approaches. croydon: cab international the world bank. people, pathogens and our planet: towards a one health approach for controlling zoonotic diseases we acknowledge support received from the national influenza center and noguchi memorial institute of medical research, and afrique one alliance. in addition, we would like to thank the physicians, nurses, surveillance officers, laboratory scientists and health information officers who contribute to the influenza-like illnesses sentinel surveillance system in the greater accra region of ghana, as well as the staff and cohort 11 residents of the ghana field epidemiology and laboratory program. key: cord-334424-z7ygy25e authors: mccaw, james m; howard, peter f; richmond, peter c; nissen, michael; sloots, theo; lambert, stephen b; lai, michael; greenberg, michael; nolan, terry; mcvernon, jodie title: household transmission of respiratory viruses – assessment of viral, individual and household characteristics in a population study of healthy australian adults date: 2012-12-11 journal: bmc infect dis doi: 10.1186/1471-2334-12-345 sha: doc_id: 334424 cord_uid: z7ygy25e background: household transmission of influenza-like illness (ili) may vary with viral and demographic characteristics. we examined the effect of these factors in a population-based sample of adults with ili. methods: we conducted a prospective cohort study in community-dwelling australian adults nested within an influenza vaccine effectiveness trial. on presentation with ili, participants were swabbed for a range of respiratory viruses and asked to return a questionnaire collecting details of household members with or without similar symptoms. we used logistic and poisson regression to assess the key characteristics of household transmission. results: 258 participants from multi-occupancy households experienced 279 ili episodes and returned a questionnaire. of these, 183 were the primary case in the household allowing assessment of factors associated with transmission. transmission was significantly associated in univariate analyses with female sex (27% vs. 13%, risk ratio (rr) = 2.13 (1.08, 4.21)) and the presence of a child in the house (33% vs. 17%, rr = 1.90 (1.11, 3.26)). the secondary household attack proportion (shap) was 0.14, higher if influenza was isolated (rr = 2.1 (1.0, 4.5)). vaccinated participants who nonetheless became infected with influenza had a higher shap (incidence rr = 5.24 (2.17, 12.6)). conclusions: the increased shap in households of vaccinated participants who nonetheless had confirmed influenza infection supports the hypothesis that in years of vaccine mismatch, not only is influenza vaccine less protective for the vaccine recipient, but that the population’s immunity is also lower. improved characterisation of the determinants of household transmission of influenza-like illness (ili) remains an important public health priority, particularly in light of the past decade's events in which we have witnessed the emergence of severe-acute-respiratory-syndrome (sars) and the 2009 h1n1 influenza pandemic. the evidence base for pandemic influenza public health interventions such as home-quarantine, provision of antiviral agents for post-exposure prophylaxis, school-closure and vaccination builds upon an appropriate understanding of the patterns and timing of infection within the household unit [1] [2] [3] [4] [5] [6] . while influenza viruses, rhinoviruses (hrvs), adenoviruses, respiratory syncytial virus (rsv) and parainfluenza viruses (pivs) are the most common aetiological agents in acute-respiratory-infection (ari) episodes [7, 8] , in 30 -40% of all ari episodes no known respiratory virus can be identified [9, 10] . this is despite discovery of a number of previously undescribed viruses since 2001 from clinical specimens from the human respiratory tract (human metapneumovirus [11] , sars coronavirus [12] , coronavirus nl63 [13] , coronavirus hku1 [14] , novel rhinoviruses [8] , human bocaviruses [15] and k1 and wu polyomaviruses [16, 17] ). reflecting the need to improve our understanding of household transmission of ari, the literature examining factors associated with household transmission of influenza [6, [18] [19] [20] [21] has expanded significantly since the 2009 h1n1 influenza pandemic [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] , including a systematic review and meta-analysis [32] . donnelly et al. estimated the serial interval for all ili (without laboratory confirmation) from case reports during the 2009 pandemic [25] . only two studies of which we are aware explicitly consider the impact of virus type on infectiousness. principi et al. found less onwards transmission to household members from influenza-negative than influenza-positive children presenting to a hospital emergency department [33] . similarly, in a cohort study of ari in young children, lambert et. al. observed significant heterogeneity in the proportion of participants' households in which one or more illness events were observed (ranging from 13% for isolation of hmpv from the child to 61% for isolation of influenza) [34] . here we report on the household transmission of a range of viruses in a cohort study of healthy, community-dwelling adults reporting symptoms of influenza-like illness (ili). the study population was sourced from a large, industry sponsored placebocontrolled phase iv efficacy trial of a licensed seasonal trivalent influenza vaccine (fluvax w , csl ltd), conducted prior to the 2009 pandemic between march and november 2008. in a previous article [35] we have described the viral aetiology of ili in the cohort, and examined the influence of virus type, host and spatiotemporal factors on disease symptomatology. full details of subject recruitment and selection for the primary phase iv vaccine efficacy trial have been described previously [35] . briefly, across 23 study sites in australia and new zealand, 7544 healthy adults aged ≥18 to <65 years were recruited for a placebocontrolled trial of a licensed trivalent influenza vaccine (fluvax w , csl ltd) in 2008 (clinicaltrials.gov #nct00562484). study participants were randomized to receive either placebo or vaccine in a 1:2 ratio prior to the southern hemisphere 2008 influenza season. from an available pool of 5624 participants from the primary study at 12 study sites, we consider the 581 persons (adults) who experienced at least one ili episodemeeting the case definition of at least one respiratory symptom (cough, sore throat, runny nose or nasal congestion) and at least one systemic symptom (fever (oral temperature ≥ 37.8°c), feverishness, chills or myalgia) [35] , and who provided written informed consent for participation in the nested cohort-study which required contribution of a valid biological sample (copan tm dry flocked swab). samples were tested for a range of respiratory viruses using a combination of multiplexed and uniplexed conventional and real-time polymerase chain reaction (pcr) assays [35] . of those, 258 were members of multi-occupancy households, allowing investigation of transmission within the household. using a non-specific and sensitive ili definition, they reported an episode of ili on 279 occasions. for each episode they returned a study questionnaire (additional file 1) detailing respiratory symptoms (see [35] for details), health seeking behaviour (health care provider consultations, hospital admission, time off work), household characteristics (number of adults (≥18 years) and children (<18 years)) and temporally associated symptoms of ili (if any) in other household members. from herein, we consider the illness episode as the primary unit of analysis. the virology results were classified into 5 virus groups [35] : influenza (influenza a, influenza b), coronaviruses (oc43, 229e, nl63, hku1), picornaviruses, other viruses (parainfluenza viruses (1, 2, 3), adenoviruses, human metapneumovirus (hmpv), bocaviruses, rsv and ki and wu polyomaviruses) or none, where none indicates that no 'tested-for' virus was detected in the participant's sample, as opposed to a missing sample or inconclusive result. study participants' vaccination status (as determined by the primary phase iv trial intervention), physical location (i.e. study site), and socio-demographic characteristics were also recorded. the relevant outcome measure for this sub-analysis was evidence of transmission within households based on experience of symptomatic illness in at least one other member of the study participant's household. study participants were asked to complete the diary on the day following cessation of their own symptoms. they recorded the date of onset of symptomatic ilis in household members from between 14 days prior to the study participant's illness through to the day of diary completion. note that the primary case in the household may or may not be the study participant. for household ili events in which the participant was the primary case (183 of 279), transmission may or may not have occurred in the household and so it is statistically valid to develop univariate and multivariate explanatory models. for participants who reported recurrent ili episodes during the study in which the same virus was isolated, we exclude all but the first episode. events with co-introduction, defined as onset of symptoms in the participant and one or more household members on the same day, were also excluded. following these exclusions, 177 episodes remained for analysis. for household ili events in which the participant was not the primary case (95 of 279), while we do know who the introducer was for these events (via the questionnaire data), other household ili events initiated by that introducer that did not involve the participant are unobserved. that is, any events in which a child (or for that matter, any other adult member of the household) introduced an infection that did not infect the study participant are not captured by the study protocol. this observation necessitates the exclusion of all household ili events in which the participant was not the primary case from the analyses. in the one remaining household ili event, the status of the participant (primary or not) was unknown, so the episode was excluded from the analysis. for the 177 episodes in which the participant was the primary case, logistic regression models were used to explore associations between host, demographic or virus variables with any observation of within-household transmission (outcome variable = transmission in household for each recorded ili episode in a study participant). the secondary household attack proportion (shap) was calculated as the proportion of potentially exposed household members (assumed susceptible) experiencing illness, averaged over all recorded episodes. we present descriptive statistics for the shap and its variation by virus, participant and demographic variables. poisson regression models were used to assess the influence of virus, participant and demographic variables on the number of secondary cases within a given household, offset against the number of potentially exposed household members (outcome variable = number of secondary cases in household for each recorded ili episode in a study participant). vaccination status of participants was not included in the primary logistic and poisson statistical analyses due to its known mitigating effect on the likelihood of influenza acquisition [35] . investigation of the influence of prior immunisation on influenza transmission in 'breakthrough cases' was explored in a secondary analysis by inclusion of a statistical interaction term between vaccination and influenza-identification status. we make an empirical calculation of the mean time between the onset of symptoms in the primary case and the onset of symptoms in the household contacts (the serial interval), for all household ili events, events in which the participant was the primary case, and events in which influenza was isolated from the participant's virological sample. all statistical analyses were conducted in stata/ic 11.1. figure 1 reports characteristics of the 258 multioccupancy households in which transmission did and did not occur. 28 study participants reported two or more ili episodes during the course of the study. for two participants, who both experienced two episodes, picornavirus was isolated on both occasions. we only retain the first episode for each participant. the distribution of household size is dramatically different based on the absence or presence of children within the household ( figure 2 ). in households without children, the distribution is left-skewed (mean household size = 3.02, standard deviation (sd) = 1.18, skewness = 1.45), while in households with children there is minimal skew (mean household size = 4.18, sd = 1.11, skewness = 0.103). table 1 summarizes the descriptive statistics (and logistic model results) associated with presence or absence of transmission in the household for the 177 household ili events in which the participant was the primary case. there is marked co-linearity between the variables 'presence of child in household' , 'age-category' and 'household size'. for example, respondents aged 35 -44 years had significantly greater odds of having a child in the household than those aged 18 -24 years (or 53.2 (13.0, 217)), while no participant aged more than 55 years lived with a child. the relationship between the household size distribution and presence or absence of children is depicted in figure 2 . we retained 'presence of child in household' in the final multivariate model for transmission due to its strong predictive role, intuitive appeal, presumed causal role in our observed (univariate) association with age-category, and previous research indicating an association between transmission and children [18, 22, 23, 29, 36] . in the multivariate model, the observed increased risk of transmission with female sex remains (or = 2.45 (1.01, 5.93), p = 0.047). presence of children in the household is both the strongest and most statistically significant factor associated with transmission (or = 2.63 (1.18, 5.88), p = 0.018). within 258 multi-occupancy households, 177 primaryparticipant introductions gave rise to 54 secondary cases among 391 potentially exposed individuals, a secondary household attack rate (shap) of 0.138. of 102 exposed children, 22 in households in which the participant was female, 41 secondary infections were reported among 238 exposed household members (shap = 0.172), compared with 13 secondary cases among 153 contacts in households in which the participant was male (shap = 0.085), a riskratio of 2.03 (1.12, 3.66), p = 0.016 (2-sided fisher's exact). in households with children, 27 secondary infections were reported among 162 exposed household members (shap = 0.167), compared with 27 secondary cases among 229 contacts in households without children (shap = 0.118), a risk-ratio of 1.41 (0.863, 2.32), p = 0.182. a multivariate poisson regression model ( table 2 ) was used to consider the influence of virus group and demographic characteristics on the number of reported secondary cases within a given household, offset against the number of potentially exposed household members. in correspondence with the logistic regression model for transmission, we include presence of children in the in a secondary analysis, we considered the influence of prior vaccination on the reported number of secondary household cases among participants testing positive for influenza compared with all other participants. in a poisson model for secondary attacks including a statistical interaction between influenza detection (true/false) and vaccination status (placebo/vaccine), the irr for influenza positive cases in those receiving placebo was 1.69 (0.421, 6.80), p = 0.459. the factor increase (interaction term) for the irr for vaccinated participants was 3.10 (0.608, 15.8), p = 0.174, yielding a net irr for vaccinated influenza-positive participants relative to vaccinated influenza-negative participants of 5.24 (2.17, 12.6), p < 0.001). under the simplifying assumption that the introducer of infection into the household is responsible for all subsequent infections, we may calculate an empiric serial interval, the time from symptom onset in one individual until symptom onset in another. we first consider infections to be related if symptoms are reported within 14 days following onset in the primary case. across all virus-type isolations, we calculate a mean serial interval of 6.0 days (sd = 3.6) for all household ili events (where the study participant was the primary case or otherwise), and 5.1 days (sd = 3.2) for the events in which the participant was the primary case. for the five events in which the primary participant had virologically confirmed influenza and transmission occurred, the mean serial interval was 4.5 days (sd = 1.6). if we limited the maximum serial interval to seven days, the mean was reduced to 4.0 days (sd = 1.7) for all household ili events and 3.9 days (sd = 1.9) for events in which the in our main analysis, we made two assumptions that we now subject to a sensitivity analysis. of the 183 events in which our participant was the primary case, 6 were classified as co-introductions as (at least) one other household member recorded symptoms beginning on the same day. as the latent period for respiratory infections may vary from individual to individual, here we exclude a further 4 episodes in which there was a 1 day interval from onset of symptoms in the study participant to onset of symptoms in another household member. the resulting multivariate models (equivalent to tables 1 and 2 ) are materially unchanged, with the expected slight reduction in statistical power (data not shown). a second assumption made was that, for participants who reported multiple ili episodes during the study period, we only excluded the latter ili episode where the same respiratory pathogen was isolated on both occasions. however, if we conservatively exclude all ili episodes except for the first (10 episodes excluded (by virus type: 6 "none", 1 "picornavirus", 2 "influenza", 1 "coronavirus")), again we find no material change in either the logistic or poisson analyses (data not shown). this study, notable in its consideration of a broad range of respiratory pathogens in addition to influenza, demonstrates that household transmission of ili is most strongly associated with host and demographic factors: female sex and the presence of children within the household (tables 1 and 2) . the observation that female sex may be associated with increased transmission in the absence of children (rr = 2.33 (0.919, 5.90), p = 0.059) is novel, perhaps suggesting that females are fundamentally more infectious, and not simply more connected to children (in terms of both their susceptibility compared with males if a child introduces infection, and their infectiousness to children if they are the primary household case). behavioural differences whilst ill may drive such an observation. alternatively, mechanisms by which influenza pathogenesis is sex dependent have been investigated [37] ; whether or not differences extend to infectiousness and susceptibility is not clear. barbara et al. have recently identified that the reporting of respiratory symptoms may be linked with risk perception [38] and hence gender [39] . clearly, we cannot exclude the possibility of gender difference in the reporting of within household transmission. the association between transmission and the presence of children within the household is consistent with many other studies [18, 22, 23, 29, 36] . the logistic and poisson model findings (tables 1 and 2 ) are consistent with an increased susceptibility for children. this is further supported by the observed increased shap in children compared to adults (0.216 compared to 0.110, a risk-ratio of 1. 95 (1.19, 3.19 ). the shap in adults did not differ by whether or not their household contained children, suggesting that other 'indirect' effects of children are less likely. as our study design limited the analysis to household events with an adult introducer, we were unable to assess the hypothesis that children may be more infectious than adults. our poisson regression analysis on the number of secondary cases given that the participant was the primary case ( table 2) indicates that isolation of influenza in the introducer of infection to the household is associated with an increase in the number of secondary cases. we explored this finding more deeply using a statistical interaction model. while somewhat limited by sample size, we found that in placebo recipients identification of influenza was not significantly associated with an increase in the number of secondary cases (irr = 1.69 (0.421, 6.80), p = 0.459), while in vaccine recipients the irr (relative to identification of any other virus, including 'none') was 5.24 (2.17, 12.6) , p < 0.001. note that our previous analysis confirms that vaccination is associated with a reduced probability of influenza virus identification [35] . additionally, 'breakthrough' influenza cases have similar symptoms compared to unvaccinated individuals [35] . we therefore suggest that our finding of increased transmission may be explained by infection with an influenza virus mismatched to the vaccine-strain (known to be in circulation during the year of study [40] ), which furthermore may be relatively antigenically novel and to which household members may be expected to have heightened susceptibility. with no virological samples available from other household members and the small number of vaccinated participants who were infected with influenza we are unable to explore this hypothesis further. across all virus types isolated and all household ili events, and assuming that all secondary cases within the household are directly infected by the introducer, we calculate a serial interval of 6.0 days. restricting to events in which the participant was the primary case and in which influenza was isolated, we calculate a serial interval of 4.5 days. this simple approach, as taken by others [22, 29] , cannot account for two important factors: community importation and infection of household members by other non-introducing members (i.e. tertiary cases). while others have partially accounted for these effects [25, 29, 41] , a mechanistically-motivated statistical model is required to fully account for such possibilities, for example as introduced by cauchemez et al. [23] who determined a serial interval for influenza of 2.6 days (sd = 1.3) compared to 2.9 days if calculated directly from empirical observations. with just 5 events, application of these more advanced model-based techniques is not justified for our data. of primary interest for this sub-analysis focussed on transmission is the complication introduced by the monitoring and assessment of ili in an individual rather than a household. ideally, a protocol such as that suggested by klick et al. would have been employed [42] . the lack of virological assessment of household secondary cases and the broad nature of the question used to establish the secondary case count in each house also contributes to uncertainty with regards to our assignment of temporally associated ili to within-household transmission. both of these limitations were an unavoidable consequence of the nesting of the data-collection protocol within a randomized placebo-controlled trial. furthermore, due to the requested timing for completion of the questionnaire, we cannot exclude the possibility that late onset of secondary (or tertiary etc.) cases may have been missed, particularly if a participant's experience of symptoms was of short duration. similarly, because the study protocol and analyses effectively assume that individuals are infectious until the end of their symptoms, any systematic differences (by virus type) in this relationship may influence the results. however the prompt to return the diary upon symptom cessation was in an effort to ensure timely reporting of questionnaire information to minimise recall bias. conversely, our poisson model implicitly assumes independence among household members, attributing all household infections to the primary case. more advanced model based methods that account for tertiary (and subsequent) cases and community introduction would be warranted with more complete data sources. as with all protocols based purely on symptomatic presentation (as opposed to active surveillance for nonclinical signs of infection such as virological or immunological measures [24, 42] ), we are unable to account for potential sub-clinical infection routes, with potential impact for our assessment of whether or not transmission did occur, the primary case status of our participants and determination of the size of the susceptible pool within a given household. conversely, taking a nonsimulation approach to analysis, we are unable to discount our estimate for the shap due to the effects of community introduction into the household, or account for community introduction and tertiary cases in our estimate for the serial interval [23] . our study sample had an over-representation of females (166 of 258 (64.3%) individuals for the 279 captured episodes; 105 of 167 (62.9%) individuals for the 177 primary-participant introductions). furthermore, it should be noted that the study population were originally volunteers in a randomized controlled trial and as such more likely to represent a group who were more concerned with their health than the general population. eligibility was restricted to healthy adults without recognized risk factors for severe influenza infection. in the context of a literature focussed on the transmission characteristics of laboratory confirmed influenza, our study is the only one that we know of to systematically explore the relationship between transmission and virus aetiology. the analyses suggest that influenza is more transmissible than other causative agents of ili, at least when introduced to the household by an adult. host and demographic factors are also of importance. further studies combining active surveillance of all household members with specimen collection and testing for a range of respiratory pathogens are warranted to elucidate these relationships. additional file 1: illness visit questionnaire. competing interests pcr has previously served on a scientific advisory board regarding influenza vaccines for csl ltd and has received a grant for an investigator initiated epidemiological study of otitis media from glaxosmithkline australia. he has also received travel support for himself and staff employed by the vaccine trial group to attend and present data at scientific meetings from baxter, glaxosmithkline, sanofi and pfizer. mdn has received travel grants from wyeth australia to present independent research at international meetings, and currently and previously has been the principal investigator for clinical trials sponsored by abbott, baxter, csl, gsk, medimmune, merck, novartis, sanofi-pasteur, wyeth, and pfizer. ml is an employee of csl limited and has an equity interest in the company. authors' contributions jmc, ph and jmv conducted the statistical analyses, provided the primary interpretation of the results and wrote the manuscript. pr was principal investigator on the vaccine efficacy trial within which the sub-study was conducted. tn, jmv, ts, mn, sl and pr conceived the sub-study and secured funding for its conduct, in partnership with csl limited represented by ml and mg. jmv coordinated conduct of the study at multiple sites and oversaw collation of the questionnaire data. ts, mn and sl oversaw conduct of and reporting of the virological testing at the queensland paediatric infectious diseases laboratory. ml was medical monitor for the main vaccine study and a partner investigator on the sub-study, as was mg. all authors contributed to critical revision of the manuscript and have seen and approved the final version of the manuscript. reducing the impact of the next influenza pandemic using household-based public health interventions the transmissibility and control of pandemic influenza a (h1n1) virus effective, robust design of community mitigation for pandemic influenza: a systematic examination of proposed us guidance estimating antiviral effectiveness against pandemic influenza using household data modeling targeted layered containment of an influenza pandemic in the united states estimating household and community transmission parameters for influenza a review of vaccine research and development: human acute respiratory infections frequent detection of human rhinoviruses, paramyxoviruses, coronaviruses, and bocavirus during acute respiratory tract infections the common cold vii international symposium on respiratory viral infections osterhaus ad: a newly discovered human pneumovirus isolated from young children with respiratory tract disease a novel coronavirus associated with severe acute respiratory syndrome identification of a new human coronavirus characterization and complete genome sequence of a novel coronavirus, coronavirus hku1, from patients with pneumonia cloning of a human parvovirus by molecular screening of respiratory tract samples identification of a third human polyomavirus identification of a novel polyomavirus from patients with acute respiratory tract infections risk factors of influenza transmission in households influenza virus infections in seattle families, 1975-1979. ii. pattern of infection in invaded households and relation of age and prior antibody to occurrence of infection and related illness studies of the community and family: acute respiratory illness and infection acute respiratory illness in the community: effect of family composition, smoking, and chronic symptoms secondary attack rate of pandemic influenza a(h1n1) 2009 in western australian households household transmission of 2009 pandemic influenza a (h1n1) virus in the united states comparative epidemiology of pandemic and seasonal influenza a in households serial intervals and the temporal distribution of secondary infections within households of 2009 pandemic influenza a (h1n1): implications for influenza control recommendations risk factors for laboratory-confirmed household transmission of pandemic h1n1 2009 infection influenza a (h1n1) in victoria, australia: a community case series and analysis of household transmission household transmission of pandemic (h1n1) assessing secondary attack rates among household contacts at the beginning of the influenza a (h1n1) pandemic in ontario, canada shedding and transmission of novel influenza virus a/h1n1 infection in households-germany transmissibility of seasonal and pandemic influenza in a cohort of households in hong kong household transmission of 2009 pandemic influenza a(h1n1): a systematic review and meta-analysis burden of influenza in healthy children and their households community epidemiology of human metapneumovirus, human coronavirus nl63, and other respiratory viruses in healthy preschool-aged children using parent-collected specimens virus detection and its association with symptoms during influenza-like illness in a sample of healthy adults enrolled in a randomised controlled vaccine trial. influenza other respi viruses the effect of age on transmission of 2009 pandemic influenza a (h1n1) in a camp and associated households mechanisms of sex disparities in influenza pathogenesis agreement between self-report and medical records on signs and symptoms of respiratory illness gender differences in risk perception: theoretical and methodological perspectives annual report of the national influenza surveillance scheme estimation of the serial interval of influenza optimal design of studies of influenza transmission in households. i: case-ascertained studies submit your next manuscript to biomed central and take full advantage of: • convenient online submission • thorough peer review • no space constraints or color figure charges • immediate publication on acceptance • inclusion in pubmed, cas, scopus and google scholar • research which is freely available for redistribution we thank study staff at the 12 sites for the recruitment of participants and collection of and processing of samples for this sub-study. we thank dale key: cord-345315-y3bdjnhg authors: dai, yaoyao; wang, jianming title: identifying the outbreak signal of covid-19 before the response of the traditional disease monitoring system date: 2020-10-01 journal: plos negl trop dis doi: 10.1371/journal.pntd.0008758 sha: doc_id: 345315 cord_uid: y3bdjnhg new coronavirus cases and related deaths are continuing to occur worldwide. early identification of the emergence of novel outbreaks of infectious diseases is critical to the generation of timely responses. we performed a comparative study to determine the feasibility of the early detection of the covid-19 outbreak in china based on influenza surveillance data and the internet-based baidu search index to evaluate the timelines of the alert signals compared with the traditional case reporting and response systems. an abnormal increase in the number of influenza-like illnesses (ili) occurred at least one month earlier than the clinical reports of pneumonia with unknown causes and the conventional monitoring system. the peak of the search volume was 20 days earlier than the issuance of the massive official warning about the epidemic. the findings from this study suggest that monitoring abnormal surges of ili and identifying peaks of online searches of key terms can provide early signals of novel disease outbreaks. we emphasize the importance of broadening the potential of syndromic surveillance, internet searches, and social media data together with the traditional disease surveillance system to enhance early detection and understanding of emerging infectious diseases. synopsis: early identification of the emergence of an outbreak of a novel infectious disease is critical to generating a timely response. the traditional monitoring system is adequate for detecting the outbreak of common diseases; however, it is insufficient for the discovery of novel infectious diseases. in this study, we used covid-19 as an example to compare the delay time of different tools for identifying disease outbreaks. the results showed that both the abnormal spike in influenza-like illnesses and the peak of online searches of key terms could provide early signals. we emphasize the importance of testing these findings and discussing the broader potential to use syndromic surveillance, internet searches, and social media data together with traditional disease surveillance systems for early detection and understanding of novel emerging infectious diseases. introduction new coronavirus cases and related deaths are continuing occur worldwide. [1] the who, on march 11, 2020 , declared the coronavirus disease 2019 (covid-19) outbreak a global pandemic. this pandemic dates back to december 2019, when a cluster of unexplained pneumonia cases was identified, which were linked to a seafood market in wuhan, china. [2] subsequent investigations determined that a novel coronavirus, severe acute respiratory syndrome coronavirus 2 (sars-cov-2), was the causative agent now at the heart of the pandemic of an emerging infectious disease (eid). the virus jumped from the transportation hub to other areas during the peak seasonal travel periods of the winter holiday and the traditional spring festival. [3] to control the spread and mitigate the risk of the virus, a series of strong, unprecedented measures were taken by the chinese government. these measures included the mandatory wearing of face masks in public, canceling of mass events, closing of scenic attractions, suspending of long-distance buses, and asking hundreds of millions of chinese citizens to stay indoors to break the transmission chain. [4, 5] despite the rapid increase in the number of covid-19 cases in january, china has now passed the peak of the epidemic and has effectively controlled the disease. [4] no new infections of the novel coronavirus were reported on march 18 in wuhan, the epicenter of the epidemic in china, marking a notable first success in the months-long battle with the virus and showing hope of suppressing the pandemic. because this is an infectious disease caused by a new virus, it took approximately one month from the initial detection of unexplained pneumonia cases to the definite conclusion of "human-to-human transmission" and the inclusion of the disease in the management of statutory infectious diseases by the national health commission, china. the traditional disease monitoring system is useful for detecting the outbreak of common infectious diseases, but it is insufficient for the discovery of new diseases. [6] how to build a comprehensive early warning system of public health emergencies from multiple sources has become the focus of attention of all countries. to compensate for the shortcomings of the traditional disease monitoring system, some scholars have tried to use digital data streams, [7] network density, [8] and google trends (gt) [9] as early warning indicators; these attempts have achieved remarkable results; nevertheless, the roles of these indicators in covid-19 remain unclear. in this study, we performed a comparative study to discuss the early warning capability, timelines, and validity of alert signals for the first wave of the covid-19 outbreak in china based on the surveillance data of influenza-like illness (ili) and the baidu search index (bsi) compared with the traditional case reporting system. covid-19 data from china were obtained from the center for disease control and prevention of china and national health commission of china as well as the report of the who-china joint mission on coronavirus disease 2019 (https://www.who.int) and vital surveillances report on the epidemiological characteristics of an outbreak of covid-19-china, 2020. [10] the data source of influenza-like illnesses we extracted data regarding ili reported from january 2015 to may 2019 from the national health commission of china. after the 2003 sars epidemic, the chinese government built the world's most extensive internet-based disease reporting system, called the china information system for disease control and prevention (cisdcp). [11] cases of infectious diseases, categorized as class a, b, and c, are required to be reported through the cisdcp within a limited time. we compared the monthly morbidity of ili during the last five years and plotted a line chart to describe the long-term trend. we also compared the peak of ili with the onset of the covid-19 in the late 2019 in china. we used the baidu search engine (http://index.baidu.com/v2/#/) to analyze the bsi for searches of the keywords of "pneumonia" and "sars" from november 1, 2019, to february 1, 2020. baidu is the world's largest chinese search engine and china's largest internet integrated service company. the bsi reflects active searches by internet users. we compared the timeline of peak searches for these key terms with the time of official response to the epidemic. data were entered into excel and analyzed using spss 25 (ibm, ny, usa). the ili cases across several years were compared using the analysis of variance. the dunnet method was used for pairwise comparison. the test level for significance was set at 0.05. data of this study were extracted from a public database. no individual information was published in this paper. therefore, this study is exempt from ethical approval. on december 29, 2019, the department of health of hubei province and wuhan city received a report from a local hospital regarding patients with unexplained pneumonia, all of whom were employees of the south china seafood wholesale market. on december 31, the national health commission and cdc sent a team of experts to wuhan. the investigators excluded several suspected causes, including influenza, avian influenza, adenovirus, severe acute respiratory syndrome coronavirus (sars-cov), and the middle east respiratory syndrome coronavirus (mers-cov). on january 1, 2020, the local government closed this seafood market and disinfected the area. on january 3, 2020, the chinese government informed the who of the outbreak of unexplained pneumonia. on january 7, 2020, the pathogen was identified as a new type of coronavirus, and then, the full genome sequences of this new virus were shared. on january 10, an expert group and a who team were invited to visit wuhan for a field investigation. by january 19, 198 novel coronavirus cases have been reported in wuhan. as of january 19, the risk of human-to-human transmission of this new virus had not been determined, and officials have not realized the potential global epidemic risk. on january 20, the novel coronavirus pneumonia was incorporated as a notifiable disease under the infectious disease law and health and quarantine law in china. on january 23, the whole city of wuhan was locked down, and all the residents were required to stay at home. two days later, the chinese government made the highest-level commitment to mobilize all forces to stop the epidemic. [12] as of january 28, 2020, there were more than 5900 confirmed cases and more than 9000 suspected cases of covid-19 across 33 chinese provinces or municipalities. [13] human-to-human transmission of the pathogen was also confirmed. [14] huang et al. analyzed laboratory-confirmed covid-19 cases in wuhan and showed that the symptom onset date of the first patient was december 1, 2019. [14] it is estimated that the origin of covid-19 was most likely earlier than december 2019. as shown in fig 1a, it took more than one and a half months for the traditional surveillance system to trigger the alert of the outbreak of this eid. as shown in fig 1b, there was a search peak for the terms of "pneumonia" (39641 times) and "sars" (297864 times) on december 31, 2019, mainly in wuhan (pneumonia: 11304 times; sars: 53887 times), where the outbreak of covid-19 occurred. with the official announcement of the exclusion of sars and the absence of apparent human to human transmission, the number of searches decreased rapidly the following day. until around january 20, the bsi of these two terms began to rise again, resulting in a second search peak, which was consistent with the increase in confirmed covid-19 cases countrywide (fig 1b) . overall, there were differences in the number of ili in 2014-2019 (f = 8.03, p<0.001). as shown in fig 2a, the ili case numbers in 2019 were significantly higher than those reported in the previous years of 2014-2018 (p<0.05). we observed an early spike in ili in winter of 2019, with a fast-growing period from november to december (fig 2b) . this observation suggests that covid-19 cases may have occurred before december 2019. the signal of the abnormally rapid increase in ili cases was earlier than the report of clinical cases of pneumonia with unknown causes through the official routine disease monitoring system. early identification of the emergence of an outbreak of a novel infectious disease is critical to generating a timely response. the traditional monitoring system is adequate for detecting the outbreak of common diseases; however, it is insufficient for the discovery of novel eids. in this study, we used covid-19 as an example to compare the delay time of different tools for identifying disease outbreaks. the results showed that both the abnormal spike in ili and the peak of online searches of key terms could provide early signals of novel eids. for centuries, infectious diseases have been among the leading causes of death and have presented growing challenges to human health. the threat is further increased by the continued emergence of new and unrecognized infectious disease epidemics. [15] due to the lack of sensitive and specific diagnostic tools, infections are often undiagnosed and therefore untreated, or are diagnosed at late stages. early detection of infectious diseases plays a crucial role in all treatment and prevention strategies. a crucial goal of infectious disease surveillance is the early detection of epidemics, which is essential for disease control. in china, the current surveillance system is based on confirmed case reports. [16] it is not practical for health units to perform laboratory tests to confirm a novel infectious disease. most infectious disease outbreaks start with clinicians noticing unusual patterns. patients may present with patterns of symptoms that are similar to those of more common diseases but which, after repeated observation and diagnostic testing, may deviate in scale, seasonality, or severity. [17] the discovery of covid-19 is an example. in december 2019, clinicians from wuhan city reported several patients with unexplained pneumonia, all of whom were employees of south china seafood wholesale market. bronchoalveolar lavage samples were collected and sequenced for the whole genome. bioinformatic analyses indicated that the pathogen was a novel coronavirus, showing the closest relationship with the bat sarslike coronavirus strain batcov ratg13. on january 8, 2020, the novel coronavirus was confirmed as the cause of unexplained pneumonia. however, at that time, people did not realize the potential risk of an epidemic (even less a pandemic) caused by this new pathogen. it was not until mid-to late-january that the risk of widespread transmission was taken seriously. in other words, clinical symptom monitoring and case reporting can help identify new diseases; however, these practices do not provide timely signals of an epidemic. internet-derived information has recently been recognized as a valuable tool for epidemiological investigation. [18] timeliness and precision in the detection of infectious disease outbreaks from the information published on the web are crucial for prevention of their spread. arsevska et al. retrieved data from a corpus of relevant documents and compared them with african swine fever (asf) outbreaks from the google search engine and the pubmed database. [19] the results showed that relevant documents could serve as a source of terms to detect infectious animal disease emergence on the web. walker et al. used google trends (gt) to investigate whether there was a surge in searches for information related to the covid-19 epidemic. these authors observed a strong correlation between the frequency of searches for smellrelated information and the onset of covid-19 infection in italy, spain, uk, usa, germany, france, iran and netherlands. [20] li et al. demonstrated that the data obtained from gt, bsi and the sina weibo index on searches for the keywords 'coronavirus and 'pneumonia' correlated with the published daily incidence of covid-19, with the maximum r > 0.89. [21] however, few studies explored the role of web-based search index in detecting the first occurrence of the covid-19. in this study, we used the bsi to explore the correlation between the internet search index and the outbreak of covid-19. the bsi is a public sampling database of search queries users entered into the predominant search engine (baidu) in china. unlike gt, the bsi reflects the absolute baidu search volume and is not displayed as normalized values. [22] one important issue that emerges from web-based searches is that they tend to underestimate the real epidemiological burden when the general population has poor knowledge of the disease. [18] additionally, the bsi can be influenced by media clamor. therefore, the real scientific usefulness of the so-called "digital epidemiology" remains questionable, at least when using gt or bsi. although the source of information cannot be taken for granted or even replace the "real life" epidemiological data, mining the web is an intriguing perspective for eids. when and where sars-cov-2 originated remains unclear. the similarity between covid-19 and influenza symptoms makes it possible that the excess ili cases were due to covid-19 cases. the presence of sars-cov-2-positive swabs in the patients supports this possibility. [23] the predominant symptoms associated with covid-19 are fever, cough, and sore throat; that is, patients often present with an ili. at the early stage of the epidemic, covid-19 cases may have been misdiagnosed as influenza or other respiratory diseases. thus, we hypothesized that ili surveillance data could be used as a tool for early detection of covid-19. kong et al. analyzed 640 throat swabs collected from patients with ili in wuhan from october 6, 2019, to january 21, 2020, and found that nine samples were positive for sars-cov-2, suggesting community transmission of sars-cov-2 in wuhan in early january 2020. [24] the dramatic increase in ili in wuhan in early december further supported this hypothesis. [24] spellberg et al. observed a seasonal spike in ili in los angeles, usa. [25] among patients with mild ili, 5% were tested positive for sars-cov-2. such transmission is consistent with the countywide unusual third ili spike that occurred late in the season and with declining rates of influenza positivity. [25] however, seasonal influenza activity was lower in 2020 than in previous years in japan. [26] it may have been affected by temperature or virulence and by measures taken to constrain the sars-cov-2 outbreak. [26] the coinfection of covid-19 and influenza a reported in iran also highlighted the importance of considering sars-cov-2 pcr assay regardless of positive findings for other pathogens during the epidemic. [27] silverman et al. explored how ili outpatient surveillance data could be used to estimate the prevalence of covid-19, and they found a surge in noninfluenza ili above the seasonal average in march 2020 and showed that this surge correlated with covid-19 across states. [28] in our study, several potential limitations should not be neglected. first, the web-based search for key terms or ili surge counts in relation to the emergence of covid-19 may be attributed to potential confounders. second, the observed ili surge may represent more than just sars-cov-2-infected patients. whether ili surveillance data could be used for the signal of the eids without dominant features of covid-19, such as cough and fever, is unclear. third, the web-based search can be affected by media coverage, the population's knowledge or the degree of information disclosure. in conclusion, monitoring abnormal surges in ili and identifying online search peaks of key terms can provide early signals of novel disease outbreaks. we emphasize the importance of testing these findings and discussing the broader potential to use syndromic surveillance, internet searches, and social media data together with traditional disease surveillance systems for early detection and understanding of eids. the covid-19 epidemic. tropical medicine & international health what we know so far: covid-19 current clinical knowledge and research a novel coronavirus outbreak of global health concern successful containment of covid-19: the who-report on the covid-19 outbreak in china the positive impact of lockdown in wuhan on containing the covid-19 outbreak in china new technologies in predicting, preventing and controlling emerging infectious diseases an early warning approach to monitor covid-19 activity with multiple digital traces in near real-time detecting early signals of covid-19 global pandemic from network density applications of google search trends for risk communication in infectious disease management: a case study of the covid-19 outbreak in taiwan the novel coronavirus pneumonia emergency response epidemiology team zhonghua liu xing bing xue za zhi = zhonghua liuxingbingxue zazhi emergence and control of infectious diseases in china gaps remain in china's ability to detect emerging infectious diseases despite advances since the onset of sars and avian flu genomic characterisation and epidemiology of 2019 novel coronavirus: implications for virus origins and receptor binding clinical features of patients infected with 2019 novel coronavirus in wuhan emerging and neglected infectious diseases: insights, advances, and challenges establishing a web-based integrated surveillance system for early detection of infectious disease epidemic in rural china: a field experimental study tracking virus outbreaks in the twenty-first century is google trends a reliable tool for digital epidemiology? insights from different clinical settings identification of terms for detecting early signals of emerging infectious disease outbreaks on the web. computers and electronics in agriculture use of google trends to investigate loss-of-smell-related searches during the covid-19 outbreak retrospective analysis of the possibility of predicting the covid-19 outbreak from internet searches and social media data, china, 2020. euro surveillance: bulletin europeen sur les maladies transmissibles = european communicable disease bulletin correlations of online search engine trends with coronavirus disease (covid-19) incidence: infodemiology study pubmed central euro surveillance: bulletin europeen sur les maladies transmissibles = european communicable disease bulletin sars-cov-2 detection in patients with influenza-like illness community prevalence of sars-cov-2 among patients with influenzalike illnesses presenting to a los angeles medical center pubmed central seasonal influenza activity during the sars-cov-2 outbreak in japan co-infection of coronavirus disease 2019 and influenza a: a report from iran using influenza surveillance networks to estimate state-specific prevalence of sars-cov-2 in the united states key: cord-307674-4fb5xnil authors: weaver, anne m.; khatun‐e‐jannat, kaniz; cercone, emily; krytus, kimberly; sohel, badrul munir; ahmed, makhdum; rahman, mustafizur; azziz‐baumgartner, eduardo; yu, jihnhee; fry, alicia m.; luby, stephen p.; ram, pavani k. title: household‐level risk factors for secondary influenza‐like illness in a rural area of bangladesh date: 2017-01-05 journal: trop med int health doi: 10.1111/tmi.12820 sha: doc_id: 307674 cord_uid: 4fb5xnil objective: to describe household‐level risk factors for secondary influenza‐like illness (ili), an important public health concern in the low‐income population of bangladesh. methods: secondary analysis of control participants in a randomised controlled trial evaluating the effect of handwashing to prevent household ili transmission. we recruited index‐case patients with ili – fever (<5 years); fever, cough or sore throat (≥5 years) – from health facilities, collected information on household factors and conducted syndromic surveillance among household contacts for 10 days after resolution of index‐case patients’ symptoms. we evaluated the associations between household factors at baseline and secondary ili among household contacts using negative binomial regression, accounting for clustering by household. results: our sample was 1491 household contacts of 184 index‐case patients. seventy‐one percentage reported that smoking occurred in their home, 27% shared a latrine with one other household and 36% shared a latrine with >1 other household. a total of 114 household contacts (7.6%) had symptoms of ili during follow‐up. smoking in the home (rr (adj) 1.9, 95% ci: 1.2, 3.0) and sharing a latrine with one household (rr (adj) 2.1, 95% ci: 1.2, 3.6) or >1 household (rr (adj) 3.1, 95% ci: 1.8–5.2) were independently associated with increased risk of secondary ili. conclusion: tobacco use in homes could increase respiratory illness in bangladesh. the mechanism between use of shared latrines and household ili transmission is not clear. it is possible that respiratory pathogens could be transmitted through faecal contact or contaminated fomites in shared latrines. annual influenza epidemics occur worldwide with sporadic pandemics. influenza is an important aetiological agent for febrile illness and pneumonia among children in urban dhaka, bangladesh [1] [2] [3] , where influenza incidence is approximately 100 episodes per 1000 childyears, and an estimated 10% of childhood pneumonia episodes are influenza-associated [2] . influenza-like illness (ili) refers to a syndrome with symptoms typical of influenza virus infection: fever with sore throat and/or cough [4] . in community-based surveillance in bangladesh, 14% of all people who died during 2009, excluding those who died from injury, suicide or homicide, had symptoms of ili within 14 days before death [5] . although 2009 was a pandemic year, which may have influenced mortality from influenza, hospital-based surveillance indicates a similar incidence of influenza-associated ili in 2008 (10 cases per 100 person-years), 2009 (6.6 cases of seasonal influenza and 4.4 cases of pandemic influenza per 100 person-years) and 2010 (17 cases per 100 person-years) [3] . in bangladesh, influenza and ili result in a high economic burden for families of ill individuals. families of individuals with influenza identified during surveillance paid a median of 16% of monthly household income in out-of-pocket costs for treatment of influenza-associated illness [6] . many families reported reducing monthly food expenditures and/or borrowing money in order to pay for treatment [6, 7] . ill individuals may be unable to work and/or attend school for several days, further increasing the financial burden on families [6, 7] . annual vaccination is a key strategy for the prevention of influenza in high-and middle-income countries [8] . in bangladesh, as in many low-income countries, vaccination against influenza viruses has not been widely promoted, likely due to high costs and competing priorities of the healthcare system [9] . non-pharmaceutical interventions that modify influenza transmission risk factors would be particularly useful in such a setting. respiratory virus transmission has been demonstrated in hong kong and the united states to be common among household contacts [10, 11] . household contacts are in frequent contact with infected individuals and have similar risk factors to infected household members [10, 11] . crowding and poor hand hygiene, which are prevalent in low-income settings, facilitate transmission of influenza and other respiratory viruses [12] [13] [14] [15] . handwashing has been associated with a reduced risk of acute respiratory infections in children [13, 16] and influenza transmission [11, 17] in high-and low-income settings. exposure to indoor and ambient air pollution has been associated with an increased risk of all-cause acute respiratory infections [18] [19] [20] [21] . exposure to air pollution may damage lung tissue and compromise immunity, increasing susceptibility to respiratory infection [22, 23] . air pollution concentrations in a home can be affected by tobacco smoking, biomass fuel use for cooking and proximity to biomass cookstoves [24, 25] . influenza and ili carry a high disease burden and subsequent economic burden in bangladesh, a lower middle-income country where widespread pharmaceutical interventions may not be currently feasible or affordable for patients. however, most studies on non-pharmaceutical interventions for influenza have been conducted in high-income settings. it is, therefore, important to identify and address modifiable factors associated with secondary ili, defined as ili in another household compound member after the first patient has been identified, at the household level in bangladesh and other highburden, low-income settings in order to design interventions to reduce transmission. for this study, we aimed to identify household-level risk factors associated with secondary ili in rural bangladesh. we conducted this analysis using the control group of a randomised controlled trial, bangladesh interruption of secondary transmission of influenza study (bistis) [26] . during the 2009 and 2010 influenza seasons, patients who sought care for respiratory symptoms at jahurul islam medical college hospital, two district health complexes, and six local pharmacies in rural kishoreganj district, bangladesh, were recruited as index-case patients. study physicians screened patients for the presence of influenzalike illness (ili), which was defined as fever in those less than 5 years of age and fever with cough or sore throat in those 5 years or older. as this study was investigating transmission of influenza at the household level, patients who were admitted to the hospital were ineligible to participate. consenting index-case patients were accompanied to their home by study staff. typically, residents of this area live with extended family members in compounds of several households, sometimes with a shared cooking space and a latrine. if at least two people other than the index-case patient intended to reside in the compound for the subsequent 20 days, we sought to enumerate and enrol all members of the compound ( figure 1 ). eligibility requirements of index-case patients varied during the study period [26] . briefly, in 2009, we recruited index-case patients who experienced symptom onset in the prior 7 days, who lived within 30 min travel time to the health facility, and had no ili among household compound members in the prior 3 days (n = 18). due to a lack of recruitment, after one month, we expanded this criteria to include those living within two hours' travel time and those with ili in other household compound members (n = 65). in 2010, in response to literature indicating that handwashing was effective against influenza transmission within 36 h of symptom onset [11] , we limited enrolment to index-case patients with symptom onset within 48 h. we allowed recruitment of those compounds where individuals who did not live in the indexcase patient's home had ili (n = 103). full details of the eligibility requirements are described elsewhere [26] . household contacts who had fever at enrolment (n = 53) were excluded from these analyses. randomisation to an intensive handwashing intervention or control arm was carried out at the compound level. details of the handwashing intervention are described elsewhere [26] . the following analyses were conducted among participants randomised to the control group to reflect household-level risk factors for ili. at the initial healthcare facility visit of the index-case patient, a trained study physician procured specimens using a nasal swab and an oropharyngeal swab, which were batched and tested by rt-pcr for influenza viral rna using the world health organization protocol [27] . after index-case patients were recruited and tested for influenza virus infection, study staff accompanied index-case patients to their homes and recruited members of their compounds into the study. a staff member then administered a questionnaire to the male or female head of each household in the compound to assess demographics, socio-economic factors and individual-and household-level characteristics. the staff member observed each household for certain factors such as presence of a handwashing station with soap and water, location of cooking area, type of fuel used, water source and latrine facilities. study staff visited each household compound daily until the tenth day after resolution of the index-case patient's symptoms to conduct surveillance for ili symptoms. any member of the compound with new ili symptoms during the follow-up period was considered a secondary ili case. after consent was obtained, the secondary ili case patients were tested for influenza in the same manner as the index-case patient. written informed consent for specimen collection was obtained from all adult index-case patients and secondary ili cases. if the index-case patient or secondary ili case was a child, written informed consent for specimen collection was obtained from a parent or guardian. written informed consent was obtained from the head of the compound (typically the eldest male) for all household data collection activities. all study procedures were approved by the international centre for diarrhoeal disease research, bangladesh (icddr,b) research and ethics review committees. as few (n = 35) index-case patients had laboratory-confirmed influenza in the control arm, we included all index-case patients with ili and conducted analyses to determine household-level risk factors associated with secondary ili in household members. we examined the following household-level characteristics as potential risk factors for secondary ili: crowding, building materials of homes, exposure to indoor air pollution, presence of handwashing materials, water source, latrine quality and sharing, education of the household respondent and household wealth status. crowding was assessed as number of people per room (the number of people in the household divided by the number of rooms in the home, excluding latrine and kitchen). we assessed indicators of exposure to indoor air pollution, such as frequency of index-case patients with influenza-like illness (ili) identified at hospitals, health complexes, pharmacies, tested for influenza (n = 377) household compound members of index-case patients recruited, baseline questionnaire given (n = 3159) handwashing intervention given at repeated visits household compound members with ili tested for influenza (n = 115) all household compound members followed for ili for 10 days after resolution of index-case patient's symptoms (n = 1498 household compound members) all household compound members followed for ili for 10 days after resolution of index-case patient's symptoms (n = 1661 household compound members) household compound members with ili tested for influenza (n = 158) exclude those with missing questionnaire data from final analysis (n = 7) smoking in the home, cooking fuel use, building material of the home and the distance between the cooking and sleeping spaces. we observed handwashing materials, soap and/or water at a handwashing station [28] . we defined latrine quality as improved (flush/pour flush to piped sewer system, septic tank or pit latrine; or pit latrine with slab) or unimproved (flush/pour flush to elsewhere, open pit latrine, bucket, hanging latrine or no facility/bush/field), according to the who/unicef joint monitoring programme for water supply and sanitation. for socio-economic status, we examined education level of the household respondent and developed a wealth index using principal component analysis of household assets [29] . we used the first principal component as our wealth index and categorised it into quintiles. we also examined each household asset that weighed on the wealth index in principal components analysis as indicators of wealth. we reported household-level factors potentially associated with ili transmission at the household and individual levels. those factors with 10-90% variability among all households were considered for multivariable analysis. we adjusted multivariable models for age of the indexcase patient (<5 years, ≥5 years), as previous analyses showed age to be associated with ili transmission in bistis [26] . we examined age of the susceptible contact as a potential confounder, both as a continuous variable and defined in the following categories: very young child (less than 2 years), young child (2-4 years), older child (5-14 years), adult (15-49 years) and older adult (50 years and older). we examined sex and wealth status of the susceptible household contact, as well as any factors associated with risk of ili in the bivariate models (p < 0.05) as potential confounders. since case definition varied by age, we conducted a sensitivity analysis in which we stratified analyses by age of the index-case patient (<5 years, ≥5 years). we also examined bivariate associations between household factors associated with secondary ili and multiple daily interactions with the index-case patient (collected in 2010), as this was shown in our prior study to be associated with ili [26] . we conducted mixed-effects log-binomial regression to evaluate the relationship between household-level factors and identification of a secondary case of ili, adjusting for age of the index-case patient and the susceptible household contact, and we accounted for clustering at the household level. in order to evaluate independent associations, we adjusted models for all other householdlevel factors associated with secondary ili in bivariate analyses (p < 0.05). we estimated the adjusted risk ratios of developing a secondary ili case among those who lived in households with factors of interest compared with those who lived in households without the factors of interest. among 1498 susceptible household contacts of 184 index-case patients, seven individuals (0.5%) from two households were excluded due to missing data. a total of 114 (7.6%) susceptible contacts developed ili symptoms during follow-up. among 1491 household contacts included in this analysis, 722 household members were from 181 index-case patient households and 769 from 182 households in the compound other than the indexcase patient's household (table 1) . houses typically consisted of one (50%) or two (30%) rooms, were made of brick or concrete (77%) and had a separate cooking space outside of the main living area (86%). almost all households cooked with biomass fuels and used tube wells for drinking water. smoking occurred in approximately 69% of homes. of 1491 household contacts, 207 (14%) reported smoking; 197 (29%) of adult men were smokers vs. 10 (1.3%) of adult women (results not shown). most (83%) household respondents had eight or fewer years of education. our wealth index accounted for 31% of the variance in asset ownership. a total of 46 (40%) of the 114 secondary ili cases lived in the index-case patient's household (table 2 ). in our final negative binomial regression models, we evaluated the independent associations between ever smoking in the home or sharing a latrine with one other household or more than one other household, and secondary ili, adjusting for age category of the index-case patient (<5, ≥5 years). models examining smoking in the home were also adjusted for shared latrine use, and models examining shared latrine use were also adjusted for smoking in the home. all other models adjusted for both smoking in the home and shared latrine use. sex and age of secondary contacts were not included as model covariates, as sex was not associated with risk of secondary ili in bivariate analysis, and addition of age of the secondary contact did not substantially change model estimates. addition of further covariates resulted in unstable models. in our final models, the risk of developing secondary ili was 91% (95% ci 1.23-2.96) greater in those who lived in a household in which smoking ever occurred, compared with those who lived in a household with no smoking. additional adjustment for education, wealth quintile and each individual asset that weighed on the wealth measure (ownership of a chair, table, mobile phone, watch or clock, sewing machine and electricity in the home) did not substantially change the estimates of the relative risk for ili among those who lived in a household where smoking occurred compared with those who did not (rr adj between 1.85 and 1.94). those who lived in a household with water at a handwashing station had a 29% lower risk of developing secondary ili compared with those without water at a handwashing station, but this association was not statistically significant (95% ci 0.39-1.28). after adjustment, having soap and water at a handwashing station was not associated with risk of secondary ili. compared with those living in a household with a private latrine, those who lived in households sharing their latrine with one other household were at a 2.07-fold increased risk of developing secondary ili (95% ci: 1.18, 3.64) and those who shared their latrine with more than one other household had a 3.08-fold increased risk of developing secondary ili (95% ci: 1.81, 5.23). additional adjustment for education, wealth quintile and each individual asset that weighed on the wealth measure did not substantially change the estimates of the relative risk for ili among those sharing a latrine with one other household (rr adj between 1.98 and 2.10) or among those sharing a latrine with more than one other household (rr adj between 3.00 and 3.12). living in the same household as an index-case patient, crowding (number of people per room), building material of home, water source and improved latrine use were not associated with risk of secondary ili. in stratified analysis, associations between household-level risk factors and risk of secondary ili did not substantially differ by age of index-case patient. sex of the secondary contact and relationship of the secondary contact to the index-case patient were not associated with risk of developing secondary ili in this analysis or in prior bistis analyses (results not shown) [30] . multiple interactions with the index-case patient were not associated with shared latrine use or smoking in the home (results not shown). in this study of household-level risk factors for ili, we found that smoking in the home and sharing a latrine with other households were associated with increased risk of secondary ili among household contacts. these results suggest that exposure to environmental tobacco smoke increases the risk of secondary ili; there are several potential mechanisms for the increased risk of ili due to shared latrine use. both factors are potentially modifiable. our results support exposure to indoor air pollution from environmental tobacco smoke as a potential risk factor for ili. exposure to indoor air pollution is a wellestablished risk factor for all-cause acute respiratory infections, due to its detrimental effects on respiratory tissue and immune functioning in the respiratory tract [31] [32] [33] . exposure to environmental tobacco smoke is also a well-established risk factor for numerous other conditions, including low birthweight, various cancers and chronic respiratory and cardiovascular diseases [34] . the prevalence of smoking in the home was high in this study, highlighting the need for tobacco control measures in bangladesh. greater use of effective tobacco control measures, such as taxation, could help to reduce tobacco smoking prevalence in bangladesh [35] . the global adult tobacco survey estimated that 45% of adult men in bangladesh smoke tobacco products [36] . in contrast, only 1.5% of adult women in bangladesh smoke. our study showed a lower proportion of men who smoke (29%) compared with the global adult tobacco survey. in our study, the household head reported tobacco smoking for all members of the household; it is possible that respondents may underreport smoking habits of other household members. although biomass fuels are considered to be the major source of indoor air pollution in low-and middle-income countries [19, 32, 37] , we were unable to assess the effect of biomass fuel use on secondary ili, as nearly every participant (96.7%) reported using biomass fuels for cooking. sharing a latrine with at least one other household was the strongest exposure associated with secondary ili observed in this study. although shared latrines have not previously been shown to be associated with respiratory infections, there is some evidence of an association between shared latrines and diarrhoeal disease [38, 39] . shared latrines may not be cleaned as frequently as private latrines [38] , so it is possible that pathogens remain present longer on surfaces in shared latrines compared with private latrines. contact transmission, by either direct contact with infected fluids or indirect contact via fomites, may be an important route of transmission for influenza and other respiratory pathogens [40, 41] as well as diarrhoeal pathogens [38] . contaminated fomites in shared latrines, such as doors and traditional pots used for anal washing after defecation, may provide a route of transmission for pathogens in bangladesh. as ili may be caused by many different pathogens, it is possible that shared latrines may expose users to a number of different pathogens that may cause ili symptoms. specifically, influenza viruses [42, 43] and coronaviruses [44] have been recovered from faeces of patients, suggesting that some respiratory viruses may be transmitted through faecal contact. interactions with people with influenza have been shown to be associated with risk of secondary influenza [45] [46] [47] [48] [49] , and ili [26] ; it is plausible that those who use shared sanitation may have increased interactivity due to a commonly used resource (latrine). we did not observe an association between multiple daily interactions with the index-case patient and shared latrine use. however, we were unable to thoroughly investigate this possibility due to limited data. it is also possible that the association between sharing a latrine and ili may be spurious or that latrine sharing represents a proxy for an unknown factor that is associated with ili, but the effect estimates did not change substantially when adjusted for measures of wealth, age or smoking making this a less likely explanation. nearly 8% of household contacts reported ili in this study. this proportion is similar to previous investigations of the burden of ili in the general population of bangladesh [5] . although age of the index-case patient did not modify the effects of household-level risk factors on ili, in this analysis and our prior analysis, ili incidence was higher in susceptible contacts who were younger than 5 years compared with those who were 5 years or older [26] . residing in the index-case patient's household was not associated with ili risk, indicating that all members of a compound in a densely populated area are at risk of contracting infectious diseases from their compound members or the community at large. important limitations of this study include lack of detail regarding intravs. extra-household transmission pathways. we do not know whether pathogens were transmitted between members of the same household compound, whether they were acquired outside of the compound or whether the index-case patient we identified is in fact the primary ili case in each compound. it is possible that control households had contact with intervention households and subsequently modified handwashing behaviour. however, our main study results do not suggest an association between handwashing and secondary ili, so contact with the intervention arm is unlikely to have affected our results. as few participants had influenza, we did not test for other pathogens, and our definition of ili for those under 5 years was broad, our results may not be relevant to influenza transmission, but rather, transmission of respiratory pathogens more broadly. air pollution is a wellestablished household-level risk factor for respiratory illness [31] [32] [33] , but reliable data on concentrations of household air pollutants are not available from this study. however, we did observe associations between indoor smoking, one proxy indicator of air pollution and secondary ili incidence. as this study recruited participants from selected healthcare facilities, our sample may not be representative of people who sought care elsewhere [3, 5] . in addition, our sample may not be generalisable to urban bangladesh, where there may be more crowding and more accessible health care. smoking in the home and use of shared latrines are associated with an increased risk of secondary influenza-like illness in households in this study. our data highlight the possible benefit of efforts to reduce exposure to indoor air pollution from environmental tobacco smoke, including effective approaches to smoking cessation and clean air initiatives. interventions focused on improving access to private latrines may also be helpful in low-income countries. influenza a and b infection in children in urban slum influenza is a major contributor to childhood pneumonia in a tropical developing country incidence of influenza-like illness and severe acute respiratory infection during three influenza seasons in bangladesh household transmission of pandemic (h1n1) influenza-associated mortality in 2009 in four sentinel sites in bangladesh economic burden of influenza-associated hospitalizations and outpatient visits in bangladesh during 2010. influenza other respir viruses international centre for diarrhoeal disease research bangladesh. the economic burden of influenza-like illness in mirpur, dhaka, during the 2009 pandemic: a household cost of illness study transmission of influenza: implications for control in health care settings infectious diseases and vaccine sciences: strategic directions role of social networks in shaping disease transmission during a community outbreak of 2009 h1n1 pandemic influenza facemasks and hand hygiene to prevent influenza transmission in households: a cluster randomized trial [summary for patients in ann intern med impact of housing conditions on the health of the people at al-ama'ri refugee camp in the west bank of palestine effect of handwashing on child health: a randomised controlled trial factors involved in the aerosol transmission of infection and control of ventilation in healthcare premises environmental risk factors for acute respiratory infections among children of military personnel in uganda physical interventions to interrupt or reduce the spread of respiratory viruses: systematic review effects of hand hygiene campaigns on incidence of laboratory-confirmed influenza and absenteeism in schoolchildren the contributions of emissions and spatial microenvironments to exposure to indoor air pollution from biomass combustion in kenya indoor air pollution: a poverty-related cause of mortality among the children of the world indoor exposure to particulate matter and the incidence of acute lower respiratory infections among children: a birth cohort study in urban bangladesh indoor exposure to particulate matter and age at first acute lower respiratory infection in a low-income urban community in bangladesh air pollution and health air pollution-physiological effects the burden of disease from indoor air pollution in developing countries: comparison of estimates household factors associated with indoor air pollution in a low-income urban area in bangladesh impact of intensive handwashing promotion on secondary household influenza-like illness in rural bangladesh: findings from a randomized controlled trial world health organization. manual for the laboratory diagnosis and virological surveillance of influenza practical guidance for measuring handwashing behavior estimating wealth effects without expenditure data-or tears: an application to educational enrollments in states of india individual level risk factors for secondary transmission of influenza-like illness: secondary data analysis from the bangladesh interruption of secondary transmission of influenza study household air quality risk factors associated with childhood pneumonia in urban dhaka, bangladesh indoor air pollution in developing countries and acute lower respiratory infections in children effect of biomass smoke on pulmonary host defence mechanisms world health organization tobacco free initiative. international consultation on environmental tobacco smoke (ets) and child health. world health organization tobacco free initiative the price sensitivity of cigarette consumption in bangladesh: evidence from the international tobacco control (itc) bangladesh wave 1 (2009) and wave 2 (2010) surveys tobacco use in 3 billion individuals from 16 countries: an analysis of nationally representative cross-sectional household surveys indoor air pollution in developing countries: recommendations for research shared sanitation and the prevalence of diarrhea in young children: evidence from 51 countries shared sanitation versus individual household latrines: a systematic review of health outcomes hand hygiene and risk of influenza virus infections in the community: a systematic review and meta-analysis pathogen survival in the external environment and the evolution of virulence influenza virus infection among pediatric patients reporting diarrhea and influenza-like illness influenza a/h5n1 virus infection in humans in cambodia detection of human coronaviruses in simultaneously collected stool samples and nasopharyngeal swabs from hospitalized children with acute gastroenteritis measured dynamic social contact patterns explain the spread of h1n1v influenza what types of contacts are important for the spread of infections?: using contact survey data to explore european mixing patterns social contacts and mixing patterns relevant to the spread of infectious diseases social mixing patterns in rural and urban areas of southern china using data on social contacts to estimate age-specific transmission parameters for respiratory-spread infectious agents university at buffalo, school of public health and health professions first, we would like to thank the participants of bistis for their time and patience. we would like to thank our field staff for all of their hard work. this research study was funded by us centers for disease control and prevention. icddr,b acknowledges with gratitude the commitment of cdc to its research efforts. icddr,b also gratefully acknowledges the following donors who provide unrestricted support: government of the people's republic of bangladesh; global affairs canada; swedish international development cooperation agency and the department for international development. the findings and conclusions in his report are those of the authors and do not necessarily represent the official position of the centers for disease control and prevention. key: cord-002438-b8t4a57r authors: cheng, wei; yu, zhao; liu, shelan; zhang, xueying; wang, xiaoxiao; cai, jian; ling, feng; chen, enfu title: comparison of influenza epidemiological and virological characteristics between outpatients and inpatients in zhejiang province, china, march 2011–june 2015 date: 2017-02-22 journal: int j environ res public health doi: 10.3390/ijerph14020217 sha: doc_id: 2438 cord_uid: b8t4a57r given the rapid rate of global spread and consequently healthcare costs related to influenza, surveillance plays an important role in monitoring the emerging pandemics in china. however, the characteristics of influenza in southeast of china haven’t been fully studied. our study use the surveillance data collected from 16 sentinel hospitals across zhejiang province during march 2011 through june 2015, including the demographic information and respiratory specimens from influenza-like illness (ili) patients and severe acute respiratory illness (sari) patients. as analysis results, most sari and ili patients were in the age group of 0–4 years old (62.38% of ili and 71.54% of sari). the respiratory specimens have statistically significantly higher positive rate for influenza among ili patients than that among sari patients (p < 0.001). the comparison between ili patients and sari patients shows no statistically significantly difference in detecting influenza virus type and influenza a virus subtype. the sari and ili patients were found to be positively correlated for overall positive rate (r = 0.63, p < 0.001), the weekly percentage of a(h1n1)pdm09 (r = 0.51, p < 0.001), influenza b virus (r = 0.17, p = 0.013), and a/h3n2 (r = 0.43, p < 0.001) among all the positive numbers. our study demonstrated that the activities of influenza virus, including its subtypes, had a similar temporal pattern between ili and sari cases. influenza virus is estimated to cause 3 to 5 million cases of severe illness and 250,000 to 500,000 deaths each year, while 5%-10% of adults and 20%-30% of children are infected with the influenza virus worldwide [1] . in lower and middle-income countries, influenza could result in large economic burden encompassing direct costs to the health service and households, and indirect costs of productivity losses [2, 3] . vaccination is a cost-effective way to reduce the public health and economic impacts caused by influenza. because of the antigenic shift and drift of the virus, the influenza vaccine composition needs regular updates. currently, the selection of strains for the annual influenza vaccine are primarily 2 of 12 based on the predominant strain of influenza virus detected in influenza-like illness (ili) [4] . however, ili only represents a proportion of acute respiratory infections with mild clinical manifestations. severe acute respiratory illness (sari) surveillance is another type of surveillance for severe influenza-associated disease. it is recommended to be monitored with established surveillance systems with ili according to the guidelines of the world health organization (who) [5] . in china, patients meeting with the definition of sari are required to have inpatient observation and necessary treatment. due to the emerging high healthcare cost and severe consequences of sari, effective vaccination is needed to reduce the incidence of sari caused by influenza [6] [7] [8] . therefore, whether influenza epidemiological and virological characteristics among sari patients were consistent to those of ili patients is critical to the effectiveness of current vaccination plan. zhejiang province located in the southeastern china, featuring by its blooming economy and high density of population. it has over than 55 million permanent residents, which are distributed in 11 metropolitan areas. influenza surveillance in zhejiang province launched at the same time with national surveillance in the year of 2001, and expanded to 16 ili and four sari sentinel hospitals by the year of 2011. both ili and sari are monitored in the surveillance system. although studies describing influenza surveillance with both ili and sari are well documented in the northern and southern hemispheres [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] , few studies have fully compared the influenza epidemiological and virological characteristics between ili and sari cases. in this study, we used four-year continuous surveillance data to compare the epidemic and virological characteristics of influenza virus between ili cases and sari cases in zhejiang province. the influenza surveillance network in zhejiang province is a part of the national influenza surveillance system [20] . the ili surveillance in zhejiang province was initially launched in 2001. as to 2009, the ili surveillance has been expanded into 16 sentinel hospitals that cover all of the 11 metropolitan areas in zhejiang province. the types of sentinel hospitals include general hospital, specialized clinic as well as maternal and children hospital. the sari surveillance started in 2009. as to march 2011, the surveillance has expended to four sentinel hospitals including three general hospitals and one children's hospital. all four sari sentinel hospitals were selected from the existing ili surveillance network ( figure 1 ). ili is defined as any person with sudden onset of fever >38 • c and cough or sore throat in the absence of other diagnosis [21] . the definition of sari is varied by ages. a patient >5 years old is defined as having sari if, upon or during admission, presenting an acute onset of elevated temperature (axillary temperature ≥38 • c) and cough or sore throat, as well as tachypnea (respiratory rate ≥25/min) or dyspnea (difficulty breathing). a patient ≤5 years old is defined as having sari if, upon or during admission, presenting with acute onset of cough or dyspnea, and at least one of the following six signs or symptoms: (a) tachypnea(respiratory rate >60/min for ages <2 months, respiratory rate >50/min for ages 2 to <12 months, and respiratory rate >40/min for ages 1 to ≤5 years); (b) inability to drink or breastfeed; (c) vomiting; (d) convulsions; (e) lethargy or unconsciousness; (f) chest in-drawing or stridor in a calm child [4] . ili is defined as any person with sudden onset of fever >38 °c and cough or sore throat in the absence of other diagnosis [21] . the definition of sari is varied by ages. a patient >5 years old is defined as having sari if, upon or during admission, presenting an acute onset of elevated temperature (axillary temperature ≥38 °c ) and cough or sore throat, as well as tachypnea (respiratory rate ≥25/min) or dyspnea (difficulty breathing). a patient ≤5 years old is defined as having sari if, upon or during admission, presenting with acute onset of cough or dyspnea, and at least one of the following six signs or symptoms: (a) tachypnea(respiratory rate >60/min for ages <2 months, respiratory rate >50/min for ages 2 to <12 months, and respiratory rate >40/min for ages 1 to ≤5 years); (b) inability to drink or breastfeed; (c) vomiting; (d) convulsions; (e) lethargy or unconsciousness; (f) chest in-drawing or stridor in a calm child [4] . each week, physicians from departments of pediatrics and respiratory, as well as emergency rooms in all of the 16 ili sentinel hospitals were required to report the number of total visits to outpatient, as well as the number of outpatients who presented with non-specific symptoms that meet a case definition of ili. as for four sari sentinel hospitals, the number of sari and total admissions from pediatrics ward, the respiratory medicine ward and the intensive care unit were also reported. all of the reported numbers in the above-mentioned departments were required to record by age classified as 0-, 5-, 15-, 25-, and 60-years old. nasopharyngeal or throat swabs were required to collect from all of sari cases and 5-15 ili cases of each sentinel hospital on every week. the collection of swabs was conducted by trained nurses from who had not received antiviral each week, physicians from departments of pediatrics and respiratory, as well as emergency rooms in all of the 16 ili sentinel hospitals were required to report the number of total visits to outpatient, as well as the number of outpatients who presented with non-specific symptoms that meet a case definition of ili. as for four sari sentinel hospitals, the number of sari and total admissions from pediatrics ward, the respiratory medicine ward and the intensive care unit were also reported. all of the reported numbers in the above-mentioned departments were required to record by age classified as 0-, 5-, 15-, 25-, and 60-years old. nasopharyngeal or throat swabs were required to collect from all of sari cases and 5-15 ili cases of each sentinel hospital on every week. the collection of swabs was conducted by trained nurses from who had not received antiviral drugs. in addition, a standardized case report form containing demographic and sample information was also required to complete among whose biological samples had been collected. due to the outbreak of avian influenza a(h7n9) in april 2013, ili surveillance was strengthened by increasing the weekly number of the biological samples collected in each hospital from 5-15 to 20. the biological samples were collected and saved in cryovial tubes, stored at 4 • c at the sentinel site, and then sent to regional center for disease control and prevention (cdc) within 48 h after the sample collected. the regional cdc laboratory tested influenza using real-time reverse transcription polymerase chain reaction (rrt-pcr) assay following the standard protocols. specimens tested as positive for influenza a were further tested for subtypes (i.e., a(h1n1), a(h3n2), a(h1n1)pdm09, and a(h7n9)) using specific rrt-pcr. once the laboratory complete, regional cdc submitted the laboratory results to the online surveillance system. data obtained from the surveillance system were reported weekly by the staff of the sentinel hospitals and laboratories. the mean and standard deviation or median and interquartile range (iqr) were calculated for continuous variables, and percentages were calculated for categorical variables. chi-squared test and fisher's exact test were used to assess the differences of age, sex, season, influenza virus type/subtype, and influenza positive rate between all the sampled ili and sari cases. the weekly number of positive influenza by subtype and the percentage of specimens tested positively were plotted to describe seasonality and circulation of influenza types/subtypes among sari and ili cases. spring, summer, autumn, and winter were defined from week 11 to 21, 22 to 38, 39 to 48, and 49 to 10 of the next year, respectively [22] . cochran-armitage trend test was used to analyze the trend change of influenza virus positivity with the increase of age. spearman correlation was applied to analyze the linear relationship of the influenza virus positive rate, weekly percentage of influenza virus subtypes accounted for all the positive numbers between sari and ili patients. ili percentage was calculated as the percentage of total outpatient visits that were due to ili and sari percentage was calculated as the percentage of total admissions that were due to sari. to compare ili percentage and sari percentage of which can better reflect the activity of influenza virus among outpatients and inpatients, spearman correlation analysis was used to assess the linear relationship between weekly ili influenza-positive rate and ili percentage, as well as weekly sari percentage and sari influenza-positive rate. a two-sided p-values were considered as statistically significant if it was found less than 0.05. all the statistics were conducted using sas version 9.2 (sas institute, cary, nc, usa). verbal consent was obtained from all patients in prior to survey and specimen collection. for children aged under 15 years old, verbal consent was obtained from at least one parent or legal guardian. the influenza surveillance were a national-wide, governmental public health activity. therefore, institutional review board approval was not required in china. in this study, the personal identifiers (e.g., names, address, occupations and so on) were not disclosed in order to maintain patient confidentiality, all the patient information was analyzed anonymously. during the study period, 52,293 patients completed both the standardized case report and laboratory sample test, of which 46,868 (89.63%) were ili patients and 5425 (10.37%) sari patients. the median age of the tested patients was 15 years (iqr: 3-34), and the median age of ili patients was significantly older than it of sari patients (p < 0.001). in addition, the group of sari patients had higher proportion of children at 0-4 years old (71.54% for sari versus 27.82% for ili), male during the study period, 52,293 patients completed both the standardized case report and laboratory sample test, of which 46,868 (89.63%) were ili patients and 5425 (10.37%) sari patients. the median age of the tested patients was 15 years (iqr: 3-34), and the median age of ili patients was significantly older than it of sari patients (p < 0.001). in addition, the group of sari patients had higher proportion of children at 0-4 years old (71.54% for sari versus 27.82% for ili), male (61.25% for sari versus 50.67% for ili), and patients enrolled in the winter (34.14% for sari versus 28.32% for ili) than those of ili patients (table 1) . the most identified influenza virus were influenza a virus (61.97% among ili, and 63.05% among sari), followed by influenza b virus (37.92% among ili, and 36.95% among sari), and mixed type virus (0.11% among ili; 0.00% among sari), with no statistical significance between the two groups (p = 0.774). for influenza a virus in the ili group, a(h3n2) was the most identified subtype (75.00%), followed by a(h1n1)pdm09 (24.87%), a(h7n9) (0.10%), and a(untype) (0.04%). those proportions were correspondent to that in the sari group-a(h3n2) (73.49%), a(h1n1)pdm09 (25.58%), a(h7n9) (0.93%), and a(untype) (0.00%) with p-value 0.067 (table 1) . influenza viruses were found in the specimen of 8601 of 52,293 (16.44%) all patients, with 8260 of 46,868 (17.62%) ili patients and 341 of 5425 (6.29%) sari patients. table 2 shows the specific positive rate of influenza virus by age groups, genders, and seasons. for ili patients, the highest (24.74%) rate is in the 40-59 years age-group, followed by >60 years age-group (23.01%). meanwhile, the positive rate of influenza viruses of sari patients was highest in the >60 years age-group (11.07%), followed by 5-14 years age-group (10.06%). among both ili and sari cases, influenza virus was found in all age groups, and cochran-armitage trend test showed that the influenza virus positive rates tend to be higher at older ages (figure 3) . the positive rate was highest in the winter, and lowest in the autumn. overall, the influenza virus positive rate among ili cases was significantly higher than that among sari cases across different groups of age, sex and season ( table 2 ). mixed type virus (0.11% among ili; 0.00% among sari), with no statistical significance between the two groups (p = 0.774). for influenza a virus in the ili group, a(h3n2) was the most identified subtype (75.00%), followed by a(h1n1)pdm09 (24.87%), a(h7n9) (0.10%), and a(untype) (0.04%). those proportions were correspondent to that in the sari group-a(h3n2) (73.49%), a(h1n1)pdm09 (25.58%), a(h7n9) (0.93%), and a(untype) (0.00%) with p-value 0.067 (table 1) . influenza viruses were found in the specimen of 8601 of 52,293 (16.44%) all patients, with 8260 of 46,868 (17.62%) ili patients and 341 of 5425 (6.29%) sari patients. table 2 shows the specific positive rate of influenza virus by age groups, genders, and seasons. for ili patients, the highest (24.74%) rate is in the 40-59 years age-group, followed by >60 years age-group (23.01%). meanwhile, the positive rate of influenza viruses of sari patients was highest in the >60 years age-group (11.07%), followed by 5-14 years age-group (10.06%). among both ili and sari cases, influenza virus was found in all age groups, and cochran-armitage trend test showed that the influenza virus positive rates tend to be higher at older ages (figure 3) . the positive rate was highest in the winter, and lowest in the autumn. overall, the influenza virus positive rate among ili cases was significantly higher than that among sari cases across different groups of age, sex and season ( table 2 ). due to the outbreak of avian h7n9 virus in april 2013, ili surveillance was strengthened by increasing the number of samples for testing from 5-15 to 20. figure 4 shows that although the weekly number of the samples tested has increased since the week 14 of year 2013, the weekly number of all ili patients remained at similar level from year 2011 to 2014 (figure 4 ). due to the outbreak of avian h7n9 virus in april 2013, ili surveillance was strengthened by increasing the number of samples for testing from 5-15 to 20. figure 4 shows that although the weekly number of the samples tested has increased since the week 14 of year 2013, the weekly number of all ili patients remained at similar level from year 2011 to 2014 (figure 4) . figure 5a ). consistently, this influenza activity was also observed among sari patients ( figure 5b ). seven a(h7n9) viruses (five in the ili group and two in the sari group) were detected during the early year of 2014 ( figure 5 ). we found that the weekly percentage of influenza virus types/subtypes among all the identified influenza cases were significantly correlated between sari and ili patients, with a(h1n1)pdm09 (r = 0.51, p < 0.001), influenza b virus (r = 0.17, p = 0.013), and a(h3n2) (r = 0.43, p < 0.001) ( table 3) . table 3 . correlation analysis of weekly influenza virus type/subtype constitution among total positive numbers between influenza-like illness (ili) and severe acute respiratory illness (sari). figure 5a ). consistently, this influenza activity was also observed among sari patients ( figure 5b ). seven a(h7n9) viruses (five in the ili group and two in the sari group) were detected during the early year of 2014 ( figure 5 ). we found that the weekly percentage of influenza virus types/subtypes among all the identified influenza cases were significantly correlated between sari and ili patients, with a(h1n1)pdm09 (r = 0.51, p < 0.001), influenza b virus (r = 0.17, p = 0.013), and a(h3n2) (r = 0.43, p < 0.001) (table 3 ). 3.6. correlation analysis between weekly ili influenza-positive rate and sari influenza-positive rate, ili percentage and ili influenza-positive rate, sari percentage and sari influenza-positive rate table 4 shows the results of spearman correlation analysis among influenza virus positive rate and percentage of people with ili or sari. the positive rate for influenza virus of ili cases and those of sari cases was statistically significantly correlated (r = 0.63, p < 0.001). the percentage of ili cases and ili influenza-positive rate was statistically significantly correlated (r = 0.53, p < 0.001), which was higher than it of sari (r = 0.19), whose coefficient was statistical significant though (p < 0.001). to our knowledge, this is the first study to compare the epidemic characteristics of influenza between outpatients and hospitalized inpatients in zhejiang province. children less than 5 years of age were found to be the largest group of both ili and sari patients, which was consistent with those reported by other studies [10, 16, 17] . besides, we found a good agreement between sari and ili patients for the weekly proportion of samples tested positively for influenza virus and the distribution of the influenza virus types/subtypes among all the identified patients. this demonstrated that the seasonal pattern and predominant circulation types of influenza were similar between ili and sari patients. finally, we found that the correlation between the weekly influenza positive rates and percentages of patients meeting the definition of ili and sari is higher among ili patients than it among sari patients, which indicated that compared to ili, influenza may less important in causing sari. although the largest age groups were same (0-4 years old) among ili patients and sari patients, this age group had higher proportion within sari patients than it within ili patients. this difference might be caused by the different behaviors when a child or an adult is found to be sick. compared to adults, children are more likely to be taken to hospital, especially for sari cases. therefore, children have sari were more likely to be involved in the surveillance. similar results have been obtained in mongolia [10] , philippines [11] , jordan [23] . our findings further demonstrated that young children are vulnerable for both mild and severe respiratory infection, and the low influenza detection rate among 0-4 years age-group in both sari and ili patients foreshadow the need of expand the respiratory illness surveillance to more types of pathogens [12, 24] . being consistent with other studies, influenza virus was detected in all age groups among both ili and sari cases [10, 14] . and overall, the proportion of samples tested positively for influenza viruses in different age groups presented to be higher with the increase of age. therefore, all persons aged 6 months and older are recommended for vaccination and elders should be considered with priority [25] . to understand the temporal characteristics of influenza epidemics is essential for planning influenza vaccination programmes because vaccine effectiveness wanes over time, a boost of vaccine is essential to prevent the spread of diseases [26] . the simple and preferred measure to assess and compare seasonality patterns was the proportion of influenza positives [27] . the highly correlated influenza virus positive rate between sari and ili patients demonstrated the similar temporal pattern of influenza activity in the two groups. similar findings were recently reported in a description of influenza surveillance in egypt, which showed that the seasonality of influenza among ili cases and sari cases was consisted in november-february [19] . moreover, the comparison between ili patients and sari patients shows no statistically significantly difference in detecting influenza virus type and influenza a virus subtype, which was similar to the findings from nigeria [16] . finally, the correlation of weekly percentage of influenza virus type/subtypes accounted for all the positive numbers between sari and ili patients indicated that the predominant influenza types/subtypes among ili and sari cases was corresponded. these findings are essential for planning influenza vaccination programmes for those severe cases given recommendations for strain inclusion within the vaccine are based on the ili surveillance system. our results were consistent to the study conducted by peng et al. in the year of 2015 [4] . but in that study, the detailed analyses such as correlation analysis between weekly ili influenza-positive rate and sari influenza-positive rate, ili percentage and ili influenza-positive rate, sari percentage and sari influenza-positive rate were not performed. therefore, we believe the present study can provide more comprehensive information on the comparison of influenza epidemiological and virological characteristics between outpatients and inpatients. in line with the findings of other studies [11, 13, 14] , we also found higher influenza detection rate among ili patients compared to that among sari patients. moreover, we also found the correlation coefficient of ili percentage and ili influenza-positive rate was higher than that of sari percentage and sari influenza-positive rate. these phenomena may due to the higher specificity of the ili diagnosis compared to the sari diagnosis [28] . some studies have demonstrated that influenza played an important role in the viral aetiologies of ili cases [29] [30] [31] , while other respiratory virus such as respiratory syncytial virus, rhinovirus, human bocavirus were essential for the cause of sari, especially in children [32, 33] . this indicated that it is more necessary to conduct pathogen spectrum test for sari cases so as to accurately understand the cause of those severe cases. of note, we detected seven cases of a(h7n9) viruses (five in the ili group and two in the sari group) during early year of 2014, when this virus was outbreak in zhejiang province [34] . the detection rate of a(h7n9) virus in the sari patients was higher than that in the ili patients because this strain frequently cause severe syndromes. however, our study found that the surveillance network has low sensitivity on the capture of patients infected with a(h7n9) virus. although one of the critical functions of influenza surveillance is to detect novel strains of influenza, the rapid detection of emerging novel influenza strains or outbreaks of respiratory disease calls for other surveillance with standardized methodology [5] . this study has several limitations. first, although sari surveillance was required to catch all patients in accordance with the definition, sometimes the cases may not be fully recorded because of physicians' oversights and absenteeism. in the ili surveillance, to get all ili patients surveyed was impossible due to the limited amount of resources and personnel. therefore, the subject chosen for ili sampled may prone to sicker patients or younger patients. second, compared to ili surveillance, sari surveillance may be less representative due to its sparse surveillance sites. in the future, we should consider to expand and to enhance the coverage of the surveillance network, with priority to choose hospitals from ili surveillance. third, we did not test for pathogens other than influenza, which made us unable to exclude other viral, bacterial, and fungal pathogens that could be the causes of ili and sari. this study demonstrated circulating types/subtypes of influenza strains and seasonality pattern of ili cases were similar to that of sari cases in zhejiang providence. this reassured the effectiveness of influenza vaccine as strain selection based upon ili surveillance. our study results suggest that compared to ili patients, it is more necessary to conduct pathogen spectrum detection among sari patients. in the future, the expanded and enhanced ili and sari surveillance in the province may contribute to identify the novel virus, detect pandemics at early stage, and then improve the understanding of the etiology of ili and sari. a systematic review of the social and economic burden of influenza in low-and middle-income countries the economic burden of influenza-associated outpatient visits and hospitalizations in china: a retrospective survey characterizing the epidemiology, virology, and clinical features of influenza in china's first severe acute respiratory infection sentinel surveillance system global epidemiological surveillance standards for influenza the burden of influenza-associated hospitalizations in oman direct medical cost of influenza-related hospitalizations among severe acute respiratory infections cases in three provinces in china determining the provincial and national burden of influenza-associated severe acute respiratory illness in south africa using a rapid assessment methodology clinicand hospital-based sentinel influenza surveillance epidemiology and impact of influenza in mongolia estimating influenza outpatients' and inpatients' incidences from 2009 to 2011 in a tropical urban setting in the philippines incidence of influenza-like illness and severe acute respiratory infection during three influenza seasons in bangladesh influenza sentinel surveillance in rwanda influenza surveillance in 15 countries in africa influenza virus surveillance in pakistan during influenza viruses in nigeria, 2009-2010: results from the first 17 months of a national influenza sentinel surveillance system influenza sentinel surveillance among patients with influenza-like-illness and severe acute respiratory illness within the framework of the national reference laboratory sentinel surveillance for influenza-like illness, severe acute respiratory illness, and laboratory-confirmed influenza in kinshasa, democratic republic of congo influenza virus positivity and circulating subtypes among cases of influenza-like illness and severe acute respiratory infection influenza-associated mortality in temperate and subtropical chinese cities characterization of regional influenza seasonality patterns in china and implications for vaccination strategies: spatio-temporal modeling of surveillance data method for four seasons division in zhejiang province influenza hospitalization epidemiology from a severe acute respiratory infection surveillance system in jordan influenza surveillance among outpatients and inpatients in morocco cdc. summary recommendations: prevention and control of influenza with vaccines: recommendations of the advisory committee on immunization practices (acip)-united states temporal patterns of influenza a and b in tropical and temperate countries: what are the lessons for influenza vaccination? expert group meeting on seasonal influenza vaccine composition for tropics and subtropics what are the most sensitive and specific sign and symptom combinations for influenza in patients hospitalized with acute respiratory illness? results from western kenya viral etiology of influenza-like illnesses in cameroon the clinical and etiological characteristics of influenza-like illness (ili) in outpatients in laboratory surveillance of influenza-like illness in seven teaching hospitals viral etiology and clinical profiles of children with severe acute respiratory infections in china identification of viral and bacterial pathogens from hospitalized children with severe acute respiratory illness in lusaka, zambia epidemiology of human infections with avian influenza a(h7n9) virus in the two waves before and after october 2013 in zhejiang province acknowledgments: this work was financially supported by grants from zhejiang province (major science and technology programme, grant number 2014c03039), (medical research programme, grant number 2015rcb011, 2016rca008). we would like to thank physicians and staff in the 16 influenza surveillance hospitals across zhejiang province for their dedication in completing the countless numbers of data forms that made this work possible. we also want to thank physicians and staff of hangzhou, ningbo, huzhou, shaoxing, jiaxing, quzhou, taizhou, zhoushan, wenzhou, jinhua, lishui and yiwu centers for disease control and prevention for their invaluable assistance with samples collection and detection. the authors declare no conflict of interest. key: cord-309860-otx45b8x authors: conway, nicholas t.; wake, zoe v.; richmond, peter c.; smith, david w.; keil, anthony d.; williams, simon; kelly, heath; carcione, dale; effler, paul v.; blyth, christopher c. title: clinical predictors of influenza in young children: the limitations of “influenza-like illness” date: 2012-09-03 journal: j pediatric infect dis soc doi: 10.1093/jpids/pis081 sha: doc_id: 309860 cord_uid: otx45b8x background: influenza-like illness (ili) definitions have been infrequently studied in young children. despite this, clinical definitions of ili play an important role in influenza surveillance. this study aims to identify clinical predictors of influenza infection in children ≤5 years old from which age-specific ili definitions are then constructed. methods: children aged 6–59 months with a history of fever and acute respiratory symptoms were recruited in the western australia influenza vaccine effectiveness (waive) study. clinical data and per-nasal specimens were obtained from all children. logistic regression identified significant predictors of influenza infection. different ili definitions were compared for diagnostic accuracy. results: children were recruited from 2 winter influenza seasons (2008–2009; n = 944). of 919 eligible children, 179 (19.5%) had laboratory-confirmed influenza infection. predictors of infection included increasing age, lack of influenza vaccination, lower birth weight, fever, cough, and absence of wheeze. an ili definition comprising fever ≥38°c, cough, and no wheeze had 58% sensitivity (95% confidence interval [ci], 50–66), 60% specificity (95% ci, 56–64), 26% positive predictive value (95% ci, 21–31), and 86% negative predictive value (95% ci, 82–89). the addition of other symptoms or higher fever thresholds to ili definition had little impact. the centers for disease control and prevention definition of ili (presence of fever [≥37.8°c] and cough and/or sore throat) was sensitive (92%; 95% ci, 86–95), yet lacked specificity (10%; 95% ci, 8–13) in this population. conclusions: influenza-like illness is a poor predictor of laboratory-confirmed influenza infection in young children but can be improved using age-specific data. incorporating age-specific ili definitions and/or diagnostic testing into influenza surveillance systems will improve the accuracy of epidemiological data. population surveillance is used to guide preventative strategies for influenza such as choosing strains for seasonal influenza vaccine constitution and early identification of pandemic or epidemic antigenic drift variants [1] . early diagnosis of influenza disease also influences clinical decision making, especially when managing those at higher risk of severe disease [2] . influenza disease surveillance usually includes a combination of community-and hospital-based syndromic surveillance and routinely collected data concerning morbidity and mortality, with only some including laboratory confirmation of influenza infection. consistent with other countries, australian children <5 years old experience the highest laboratoryconfirmed influenza notification rate (3.4 times the rest of the population in 2008), the highest rate of general practice consultations with influenza-like illness (ili) ( 50 000/100 000 population in 2008), and the greatest morbidity [3] . in the united states, children <5 years comprised 28% of ili presentations in the centers for disease control and prevention (cdc) national network during the 2007-2008 influenza season [4] . the value of these data using ili surveillance depends on a reliable and robust definition of ili as well as a clear understanding of how ili activity relates to influenza. the definition of ili varies between countries and surveillance systems, but it usually includes the presence of fever and symptoms of acute respiratory tract infection [5] [6] [7] . influenza-like illness is a poor predictor of actual influenza infection in adults, despite attempts to improve the accuracy of the definition [8] , but there is limited data on the reliability of ili in predicting influenza infection in children [9] [10] [11] [12] . in this study, we have used data collected from young children recruited as part of a community and hospital influenza surveillance program to assess the clinical predictors of influenza infection in children 5 years old. we developed age-specific ili definitions and tested their diagnostic accuracy against existing definitions and parental opinion. the western australia influenza vaccine effectiveness (waive) study is an observational study designed to measure influenza vaccine effectiveness in children. the study design has been described elsewhere [13, 14] . in brief, children 6-59 months of age were eligible for participation if they presented with symptoms suggestive of an acute respiratory infection to selected general practices (2008 only), emergency departments (eds; 2008 and 2009), and pediatric inpatient facilities (2008 and 2009 ). the recruitment period was for the duration of the winter influenza season as determined by local population surveillance. children were eligible for enrollment in 2008 whether they had a history of fever (by parental report) or a measured temperature >37.8°c on presentation in addition to any acute respiratory symptoms within 72 hours before recruitment. to enhance recruitment in 2009, children were enrolled whether they had a history of fever or a measured temperature of >37.5°c on presentation plus the presence of any acute respiratory symptoms within 96 hours of recruitment. the study received approval from the relevant local human research ethics committees. on enrollment by trained research staff, parents were asked to complete a questionnaire detailing demographics, medical history, and presenting symptoms. temperature measured at enrollment was recorded by research staff. per-nasal swabs (copan diagnostics inc., murrieta, ca) placed in viral transport medium or per-nasal aspirates were collected. influenza testing on nasal swabs was performed by polymerase chain reaction (pcr) directed at hemagglutinin and matrix gene targets in multiplex real-time assay [14, 15] , and by conventional cell cultures [14] . in addition, samples underwent pcr directed at other common respiratory viruses including respiratory syncytial virus (rsv), parainfluenza viruses 1-3, human metapneumovirus, rhinoviruses, adenoviruses, coronaviruses (other than severe acute respiratory syndrome coronavirus), and bocavirus [14] . presence of influenza a or b detected by pcr or culture was collapsed into 1 dichotomous dependent outcome: influenza present or absent. the predictor variables of interest fell into 2 groups: (1) demographic factors (age, sex, race [indigenous or other], deprivation quintile, influenza vaccination status, prematurity [<37 completed weeks gestation], birth weight, past medical history, child care usage, household composition, and household smokers); and (2) symptomatology (recorded temperature and presence or absence of parentally reported: cough, coryza, wheeze, breathing difficulties, earache, sinusitis, sore throat, irritability, rash, diarrhoea, vomiting, lethargy, poor feeding, sleep disturbance, fever, and pallor). vaccination status was verified via the australian childhood immunisation register (acir) [16] . where acir conflicted with parental report, the family doctor was contacted to establish the number of trivalent influenza vaccine (tiv) doses given previously. because they are not normally distributed, categorical variables were created for age, household composition, and duration of fever and respiratory symptoms. postal addresses were geo-coded before conversion to deprivation quintiles using data produced by the australian bureau of statistics [17] . statistical analysis was performed using spss 16.0.0 (spss inc., chicago, il). after initial univariate analysis, variables were analyzed simultaneously within both groups (demographic factors and symptoms) by forced entry into a multivariable logistical regression model. factors found to be significant (p < .05) at the group stage were then entered into a model encompassing both groups. various definitions of ili were then constructed based on the significant predictors of influenza infection. sensitivity, specificity, and positive predictive value (ppv) and negative predictive value (npv) and their respective binomial 95% confidence intervals (cis) were then calculated. in addition, positive and negative likelihood ratios (lr+ and lr-) were also calculated. ninety-five percent cis for likelihood ratios were calculated in the manner described by simel et al [18] . new ili definitions were compared with definitions used by the cdc ( presence of fever (100 o f [37.8°c ]) and a cough and/or sore throat in the absence of a known cause other than influenza [5] ) and parents' response to the questionnaire question: "do you think your child has influenza ('flu')?" nine hundred and forty-four subjects were recruited (315 subjects in 2008 and 629 subjects in 2009). the majority were recruited within the hospital setting (general practice, 153 subjects; eds, 619 subjects; inpatient wards, 172 subjects). twenty-five subjects were withdrawn (specimens not processed, 8; invalid consent, 8; incorrect age, 3; other, 6), resulting in a total sample population of 919. altering the eligibility criteria in 2009 did not result in any additional recruits that were ineligible by 2008 criteria: 17 subjects had a temperature recorded >37.5°c yet <37.8°c; however, all of these subjects also had a history of fever within the previous 72 hours. respiratory viruses were identified in 711 subjects (77.4%). rhinovirus (n = 239), rsv (n = 210), and influenza virus (n = 179) were most frequently detected. of those in whom influenza was detected, 131 had influenza a (including 97 subtyped as influenza a/h1n1 2009) and 48 had influenza b. cultures proved positive for 59% (77 of 131) of those with influenza a and 58% (28 of 48) of those with influenza b. influenza was detected in 23.5% (74 of 315) of recruits in 2008, compared with 17.4% (105 of 604) of eligible recruits in 2009 (p = .03). the median age was 22 months and 526 of 919 (57%) were male. chronic comorbidities were uncommon: 140 children (15%) had a comorbid condition, of which asthma was the most common (89 of 919, 10%). premature births (<37 weeks gestation) accounted for 122 of 919 (13%) subjects. nearly one-half of all children enrolled (430 of 919, 47%) had received the recommended schedule for seasonal influenza vaccination (ie, 2 doses of tiv in the first year of vaccination followed by 1 dose in subsequent years [19] ) with a further 126 of 919 (14%) having received 1 dose in total ( table 1) . the most common symptoms reported by parents were cough (794 of 919, 86%), coryza (801 of 919, 87%), poor feeding (686 of 919, 75%), sleep disturbance (657 of 919, 71%), and irritability (607 of 919, 66%), although wheeze and respiratory distress were less prevalent (410 of 919, 45% and 413 of 919, 45%, respectively) ( table 2 ). for those that tested positive for influenza, the average temperature at enrollment was 39.5°c (range, 37-43.0; standard deviation . five hundred and ninety-eight parents recorded a response to the question, "do you think your child has influenza ('flu')?" seventy-eight of the 333 (23%) answering "yes" proved positive for influenza, whereas 24 of the 265 (9%) answering "no" were positive for influenza (p < .001). age >2 years, lack of tiv, lower birth weight, sharing a home with 2 or more other children, and being cared for by a single adult were significant demographic predictors of influenza infection on univariate analysis (table 3) . with the exception of the number of adult household members, all of these variables were significant when entered simultaneously within this group, and so they were entered into the final model. the symptoms that were significant univariate predictors of influenza infection were as follows: raised temperature, fever for >3 days, presence of cough, absence of wheeze, respiratory distress, and rash. fever, presence of cough, and absence of wheeze remained significant when entered simultaneously within the symptom group, and so they were retained for the final model. with the exception of number of children in the house, all variables entered into the final model remained significant (see table 3 ). based on the results of the regression equation, ili was defined in this population using various combinations of the following criteria: presence of cough, absence of wheeze, and incremental thresholds of fever (see figure 1 and table 4 ). a complete data set for each of these variables was available for 798 cases, which were used for subsequent calculations of the performance of ili definition. the presence of cough alone was highly sensitive (93%; 95% ci, 84-96) yet lacked specificity in diagnosing influenza infection (14%; 95% ci, [12] [13] [14] [15] [16] [17] . if the definition of ili comprised cough and the absence of wheeze, sensitivity was reduced (60%; 95% ci, 52-68) but specificity improved (59%; 95% ci, 55-63). the addition of fever 38°c to this definition resulted in a small the above analyses were repeated with data stratified according to vaccination status (no tiv vs any tiv) and age group (2 years vs >2 years) (see supplementary digital content). the observed ppvs in the unvaccinated group were significantly greater than the vaccinated group. subgroup analysis of each vaccination group stratum by year of illness was undertaken with no appreciable difference noted (data not shown). for those 2 years, an ili definition that included fever 38°c, cough, and the absence of wheeze had a significantly lower sensitivity and ppv and a higher specificity compared with those >2 years. the reliability of parental opinion was unaffected by vaccination status or age group of child. young children and infants with seasonal influenza can present with a wide variety of symptoms and may not yet be developmentally capable of verbalizing symptoms to their caregivers [20] . current definitions of ili are largely derived from adult studies and lack validation within the pediatric setting [5] [6] [7] . fever is a common presenting symptom for young children with influenza. when all children with acute respiratory illness (irrespective of presence or absence of fever) are tested, 95% of those who test positive for influenza had a history of fever [21, 22] . systematic review of the literature has failed to find any particular combination of additional symptoms that can reliably predict influenza infection [8] . only 2 studies in this systematic review enrolled preschool children: in 1 study, 18 of 610 cases with fever and respiratory illness were <5 years old [9] , whereas the other did not stratify by age [10] . later, ohmit and monto [11] reported a ppv of 64% in a subset of 221 children <5 years old presenting with cough and fever (>38.2°c). the generalizability of these figures is limited, however, because the study design excluded children with rsv infection, which would have falsely elevated the incidence of influenza infection in the study population [11] . a further prospective pediatric study assessing the predictive nature of an ili diagnosis in children with fever and symptoms suggestive of respiratory infection (n = 128; age <17 years) did not stratify results by age, making it less generalizable to younger children [12] . this study recruited young children presenting with acute respiratory symptoms during 2 successive winter influenza seasons, which included the first wave of pandemic influenza a/h1n1 2009. in this population, we have found that an existing ili definition in common usage (cdc definition) proved highly sensitive yet lacked specificity in identifying those with influenza infection. a surveillance system that uses this ili definition in isolation would therefore grossly overestimate influenza prevalence by virtue of the large number of false positives generated. in contrast, we found that an ili definition comprising fever 38°c, cough, and absence of wheeze achieved a greater balance between sensitivity and specificity in the study population. however, ili was a poor predictor of influenza infection regardless of which definition was tested. children with ili had a 20%-30% probability of actually having influenza infection, whereas those without ili (regardless of definition tested) had a 10%-15% probability of testing positive for influenza. each definition tested had positive and negative likelihood ratios approaching 1, indicative of little appreciable change in the odds of an individual having disease if they met ili definition criteria or not. furthermore, parental prediction of influenza infection in their children compared favorably with use of an ili definition, underlining the poor diagnostic accuracy of ili overall. decisions regarding the investigation and management of children suspected of having influenza take into account a number of factors, including the known incidence of influenza at that time, the likelihood of severe disease, and the availability of antiviral medication [2] . however our findings would suggest that, due to the inaccuracy of syndromic definitions, clinicians should maintain a low threshold for influenza testing in children with possible influenza. this is especially important for children with moderate to severe illness and/or those requiring hospitalization. similarly, diagnostic testing is required to obtain accurate influenza surveillance in young children. ideally, this should include testing by highly sensitive and specific methods such as pcr, although immunofluorescent antigen detection tests are sufficiently sensitive for use on nasopharyngeal aspirates. the rapid antigen tests have been used in some studies [23, 24] , but they are less sensitive than laboratorybased tests (especially for influenza a/h1n1 2009); their performance is influenced by specimen type, test brand used, and the virus type and subtype; and they do not identify influenza a subtypes. these tests have been successfully incorporated into public health influenza surveillance systems in the past [25] ; however, they need to be reassessed now that influenza a/h1n1 2009 is circulating. the eligibility criteria for our study included a history of fever. this methodology is similar to a number of previous studies [9] [10] [11] [12] and highlights the need for external validation of any clinical predictor tool that is intended to be used in unselected populations. however, our findings remain relevant to the general pediatric population given that the vast majority of children with influenza present with fever [21, 22] . the diagnostic accuracy of ili differed between those vaccinated and those unvaccinated (reflecting the lower prevalence of influenza in those who were vaccinated). the pediatric population studied had high vaccination rates with tiv as a result of a state-wide campaign introduced in 2008 that provides free seasonal influenza vaccination to this age group. the high prevalence of influenza vaccination in our study population contributes to the generalizability of our findings to other countries where influenza vaccination is readily available to children <5 years of age. the reliability of ili as defined by fever (38°c), cough, and absence of wheeze was age dependent, with less sensitivity and greater specificity in those 2 years of age or younger. this is a reflection of proportionally higher numbers of children 2 years old with wheeze irrespective of influenza status (data not shown), presumably as a result of a bronchiolitic-type illness. this age-dependent manifestation of influenza infection further highlights the problems inherent in applying adult-derived ili definitions to the pediatric population. because the presence of wheeze was determined by the parents, the calculated prevalence is likely to be different from one based on a clinical definition [26] . however, because this bias is constant between those who tested positive for influenza and those who were negative, it is not expected to have affected the results. also, in practice, assessing children with an ili usually relies on parental history as well as clinical findings, making our findings relevant to real-life circumstances. to our knowledge, this is the first study attempting to construct a definition of ili for children aged 5 years and under using prospectively gathered data from a general pediatric population presenting with symptoms suggestive of acute respiratory tract infection. we have demonstrated that when predicting influenza infection in younger children, an ili definition constructed using age-specific data and comprising presence of fever (38°c), cough, and absence of wheeze results in a greater balance between sensitivity and specificity compared with a definition of ili used by current surveillance systems. the diagnostic accuracy of influenza virus surveillance systems would be enhanced by developing age-specific ili definitions aimed at the pediatric age group and/or by incorporating diagnostic testing into the system. author contributions. n. t. c. and z. v. w. analyzed the data and revised the manuscript. p. c. r., d. w. s., a. d. k., s. w., h. k., d. c., and p. v. e. devised the waive study and revised the manuscript. c. c. b. supervised the present study, analyzed the data, and revised the manuscript. financial support. the waive study is funded by the western australia department of health. potential conflicts of interest. n. t. c., p. c. r., and c. c. b. are members of the vaccine trials group, telethon institute for child health research. the vaccine trials group has received funding from vaccine manufacturers for conducting clinical trials, although not in relation to this study. p. c. r. has served on a scientific advisory board regarding influenza vaccines for csl ltd., has received travel support from baxter and glaxosmithkline to present at scientific meetings, and received institutional funding for investigator-led who global influenza surveillance network 2010-11 influenza antiviral medications: summary for clinicians annual report of the national influenza surveillance scheme -8 influenza season overview of influenza surveillance in the united states heterogeneous case definitions used for the surveillance of influenza in europe working towards a simple case definition for influenza surveillance does this patient have influenza? evaluation of clinical case definitions of influenza: detailed investigation of patients during the 1995-1996 epidemic in france diagnosing influenza: the value of clinical clues and laboratory tests symptomatic predictors of influenza virus positivity in children during the influenza season clinical predictors of influenza in children lessons from the first year of the waive study investigating the protective effect of influenza vaccine against laboratoryconfirmed influenza in hospitalised children aged 6-59 months vaccine effectiveness against laboratory-confirmed influenza in healthy young children. a case-control study duplex real-time reverse transcriptase pcr assays for rapid detection identification of pandemic (h1n1) 2009 and seasonal influenza a/h1, a/h3, and b viruses australian childhood immunisation register seifa: socio-economic indexes for areas likelihood ratios with confidence: sample size estimation for diagnostic test studies the australian immunisation handbook 9th edition clinical presentation of influenza in unselected children treated as outpatients the underrecognized burden of influenza in young children performance of six influenza rapid tests in detecting human influenza in clinical specimens accuracy and interpretation of rapid influenza tests in children enhancing public health surveillance for influenza virus by incorporating newly available rapid diagnostic tests what do parents of wheezy children understand by "wheeze"? [see comment epidemiological research from glaxosmithkline and csl ltd. d. w. s. is a director of 2 not-for-profit organizations (the influenza specialist group and the asia-pacific alliance for the control of influenza) that receive funding from vaccine manufacturers. all other authors report no potential conflicts.all authors have submitted the icmje form for disclosure of potential conflicts of interest. conflicts that the editors consider relevant to the content of the manuscript have been disclosed. key: cord-344070-17oac3bg authors: silverman, justin d; hupert, nathaniel; washburne, alex d title: using ili surveillance to estimate state-specific case detection rates and forecast sars-cov-2 spread in the united states date: 2020-04-03 journal: nan doi: 10.1101/2020.04.01.20050542 sha: doc_id: 344070 cord_uid: 17oac3bg detection of sars-cov-2 infections to date has relied on rt-pcr testing. however, a failure to identify early cases imported to a country, bottlenecks in rt-pcr testing, and the existence of infections which are asymptomatic, sub-clinical, or with an alternative presentation than the standard cough and fever have resulted in an under-counting of the true prevalence of sars-cov-2. here, we show how publicly available cdc influenza-like illness (ili) outpatient surveillance data can be repurposed to estimate the detection rate of symptomatic sars-cov-2 infections. we find a surge of non-influenza ili above the seasonal average and show that this surge is correlated with covid case counts across states. by quantifying the number of excess ili patients in march relative to previous years and comparing excess ili to confirmed covid case counts, we estimate the syndromic case detection rate of sars-cov-2 in the us to be less than 13%. if only 1/3 of patients infected with sars-cov-2 sought care, the ili surge would correspond to more than 8.7 million new sars-cov-2 infections across the us during the three week period from march 8 to march 28. combining excess ili counts with the date of onset of community transmission in the us, we also show that the early epidemic in the us was unlikely to be doubling slower than every 4 days. together these results suggest a conceptual model for the covid epidemic in the us in which rapid spread across the us are combined with a large population of infected patients with presumably mild-to-moderate clinical symptoms. we emphasize the importance of testing these findings with seroprevalence data, and discuss the broader potential to use syndromic time series for early detection and understanding of emerging infectious diseases. the ongoing sars-cov-2 pandemic continues to cause tremendous morbidity and mortality around the world [1, 2] . regional preparation for the pandemic requires forecasting the growth rate of the epidemic, the timing of the peak, the demand for hospital resources, and the degree to which current policies may curtail the epidemic, all of which benefit from accurate estimates of the true prevalence of the virus within a population [3] . confirmed cases are thought to be 35 underestimates of true prevalence due to some unknown combination of patients not reporting for testing, testing not being conducted, and false-negative test results. estimating the true prevalence informs the scale of upcoming hospital, icu and ventilator surges, the proportion of individuals who are susceptible to contracting the disease, and estimates of key epidemiological parameters such as the epidemic growth rate and the fraction of infections which are sub-clinical. 40 the current literature suggests that the predominant symptoms associated with covid are fever, cough and sore-throat; that is, patients often present with an influenza-like illness (ili) yet test negative for influenza [4, 5] . with many covid patients having a similar presentation as patients with influenza, existing surveillance networks in place for tracking influenza could be used to help track covid. ili correlates with known patterns of sars-cov-2 spread across states within the us, suggesting the surge is unlikely to be due to other endemic respiratory pathogens, yet is orders of magnitude larger than the number of confirmed covid cases reported. together this suggests that the 50 true prevalence of sars-cov-2 within the us is much larger than currently appreciated and that even the highest symptomatic case detection rates are likely lower than 3% corresponding to approximately 9 million new ili cases due to sars-cov-2. our analysis provides empirical corroboration of previous hypotheses of substantial undocumented cases [6] yet places the estimated undocumented case rate an order of magnitude higher than prior reports [6] . moreover, these 55 updated prevalence estimates predict that epidemic doubling times greater than 3.5 days [7, 8] would be unable to account for the magnitude of the ili surge. we test our hypothesis of sub 3-day doubling times in the us by analyzing both state and national covid surveillance data, finding a broad agreement of doubling time less than 3.5 days in both confirmed covid case counts and documented covid deaths. our findings highlight probable trajectories of the us 60 epidemic and provide evidence for a conceptual model for covid spread in the us in which more rapid spread than previously reported is coupled with a larger undiagnosed population to give rise to currently observed trends. we identified excess ili cases by first subtracting cases due to influenza and then subtracting the seasonal signal of non-influenza ili ( figure 1 ). many states, including washington, new york, oregon, pennsylvania, maryland, colorado, new jersey, and louisiana, have had a recent surge in number of non-influenza ili cases far in excess of seasonal norms. for example, in the second week of march, 2020, oregon saw 50% higher non-influenza ili than it had ever seen since the 70 inception of the ilinet surveillance system within the us. we find that with 95% probability approximately 4% of all outpatient visits in oregon during this time were for ili that could not be explained by either influenza or the normal seasonal variation of respiratory pathogens. we find that as the seasonal surge of endemic non-influenza respiratory pathogens declines, this excess ili correlates more strongly with state-level patterns of newly confirmed covid cases suggesting that 75 this surge is a reflection of ili due to sars-cov-2 (pearson ρ = 0.8 and p < 10 −10 for the last two weeks; figure s1 ). to equate this surge to state-wide or national case counts, we assume that the average number of patients seen per week by sentinel providers is representative of their respective states that week. using this assumption, the total excess non-influenza ili across the us was approximately 9.2 million excess individuals in the week starting march 15th, 2020 compared to 80 the same week in 2019 (95% credible interval of 8.0-10.1 million). the rate at which sars-cov-2+ patients with ili symptoms are identified as having covid varies by state and over time ( figure s2 ). our estimated symptomatic case detection rates have been increasing over the month of march, which can be expected given increases in testing capacity across 85 the us since the february 28 detection of community transmission in washington state. for the latest week ending march 14, covid cases in the states with the highest estimated symptomatic case detection rate (washington, nevada, and michigan) are only capturing approximately 1% of ili surges in those states ( figure 2 ). across the entire us we find the average symptomatic case detection rate to be 0.75% (95% credible interval 0.59%-1.0%). the true prevalence of sars-cov-2 is unknown. however, if we assume the excess non-influenza ili is almost entirely due to sars-cov-2, an assumption that becomes more valid as the virus becomes more prevalent, we can use the excess non-influenza ili to understand the constraints and mutual dependence of exponential growth rates, the rate of subclinical infections, and the time 95 between the onset of infectiousness and a patient reporting as ili figure 3 . with a january 15 start date of the us epidemic [9] , allowing early stochasticity from start-time to the onset of regular exponential growth, we find that it's impossible to explain the ili surge with an epidemic whose doubling time is longer than 3.5-days, as such slow growth scenarios fail to reach the observed excess ili even when accounting for possible asymptomatic cases. 100 we tested our hypothesized constraint that doubling times across the us would be shorter than 3.5 days by analyzing the doubling time confirmed covid cases and deaths due to covid. we find that both confirmed case data and covid fatality data support our estimated doubling-time bound derived from ili data. across the entire us, the doubling rate for deaths due to covid is 2.520 days (±0.001, p-value of test that doubling rate is less than 3.5 days approximately 0) 105 and the doubling time for confirmed cases is 2.455 days (±0.001, p-value of test that doubling rate is less than 3.5 days approximately 0). under a 4-day lag from the onset of infectiousness to reporting as ili, these doubling times capture the ili surge with clinical rates of 4.4% and 3.3%, respectively. here we use outpatient ili surveillance data from around the us to estimate the prevalence of sars-cov-2+. we find a clear, anomalous surge in ili outpatients during the covid epidemic that correlates with the progression of the epidemic across the us. the surge of non-influenza ili outpatients is much larger than the number of confirmed case in each state, providing evidence of large numbers of symptomatic probable covid cases that remain undetected. moreover, this 115 excess ili surge allows us to develop a bound for the slowest epidemic doubling time, and the lowest clinical rate that would still be consistent with the observed surge. we test whether or not observed epidemic doubling times are shorter than our upper-bound and find that the observed doubling time for both confirmed cases and deaths is faster than 3 days within the us. together, the surge in ili and analysis of doubling times suggest that sars-cov-2 has spread rapidly throughout the 120 us since it's january 15th start date and is likely accompanied by a large undiagnosed population of potential covid outpatients with presumably milder clinical symptoms than would be thought based on prior studies of sars-cov-2+ inpatients. our study has several limitations. first, the observed ili surge may represent more than just sars-cov-2 infected patients. a second epidemic of a non-seasonal pathogen that presents 125 with ili or changing patient behaviors causing higher rates of presentation for typical seasonal ili could confound our estimates of ili due to sars-cov-2. alternatively, it is also possible that our use of ili data has underestimated the prevalence of sars-cov-2 within the us. while early clinical reports focused on cough and fever as the dominant features of covid [5] , other reports have documented digestive symptoms as the complaint affecting up to half of patients 130 with laboratory-confirmed covid [11] , and alternative presentations, including asymptomatic or unnoticeable infections, could result in ili surges underestimating sars-cov-2 prevalence. additionally, our models have several limitations. first we assume that ili prevalence within states can be scaled to case counts at the state level. this is based on the assumption that the average number of cases seen by sentinel providers in a given week is representative of the 135 average number of patients seen by all providers within that state in a given week. errors in this assumptions would cause proportional errors in our estimated case counts and symptomatic case detection rate. second, our epidemic models are crude, us-wide seir models varying by growth rate alone and as such do not capture regional variation or intervention-induced changes in transmission. our models were used to estimate growth rates from ili for testing with covid 140 data and to estimate the mutual dependency of growth rate, the lag between the onset of infection and presentation to a doctor, and clinical rates; these models were not intended to be fine-grained forecasts for municipality hospital burden and other common goals for covid models. finer models with regional demographic, and case-severity compartments are needed to translate our range of estimated prevalence, growth rate, and clinical rates into actionable models for public 145 health managers. while an ili surge tightly correlated with covid case counts across the us strongly suggests that sars-cov-2 has potentially infected millions in the us, laboratory confirmation of our hypotheses are needed to test our findings and guide public health decisions. our conceptual model for the epidemic with the us makes clear and testable predictions. our model would suggest rela-150 tively high rates of community seropositivity in states that have already seen an ili surge. a study of ili patients from mid-march who were never diagnosed with covid could test our model's predictions about the number and regional prevalence of undetected covid cases presenting with ili during that time. if seroprevalence estimates are consistent with our estimated prevalence from these ili analyses, it would strongly suggest lower case severity rates for covid and indicate the value of ili and other public time-series of outpatient illness in facilitating early estimates of crucial epidemiological parameters for rapidly unfolding, novel pandemic diseases. since not all novel pandemic diseases are expected to present with influenza-like symptoms, surveillance of other common presenting illnesses in the outpatient setting could provide a vital tool for rapidly understanding and responding to novel infectious diseases. 160 in what follows, let i index state i and let t index week t (with t = 0 referring to october 3, 2010; the start of ilinet surveilance). our analysis centers around decomposing the probability of testing positive for covid, δ into the product of the symptomatic case detection rate for ili patients, δ s , and the probability that a covid patients presents to the clinic with ili, δ c . since 2010 the cdc has maintained ilinet for weekly influenza surveillance. each week approximately 2,600 enrolled providers distributed throughout all 50 states as well as puerto rico, the district of columbia and the us virgin islands, report the total number of patient encounters n it and the total number of which met criteria for influenza-like illness (ili -defined as a temperature for influenza y f lu it . therefore ilinet data can be thought of as a weekly state-level time-series representing the superimposed prevalence of various viruses which can cause ili. ilinet data was obtained through the cdc fluview interactive portal [13] . in addition to ilinet data, us state population data for the 2020 year was downloaded from https://worldpopulationreview.com/states/. the number of primary care providers in each state 180 per 100,000 residents b was obtained from becker's hospital review [14] . covid confirmed case counts were obtained from the new york times' database maintained at https://github.com/nytimes/covid-19-data. this dataset contains the daily cumulative confirmed case count for covid for each state z il for day l. within the ilinet dataset, new york city and new york were summed into a combined new york variable representing both new york city and the surrounding state. due to incomplete data in one or more of the data-sources described above the virgin islands, puerto rico, the commonwealth of the northern mariana islands, and florida were excluded from subsequent analysis. in addition, daily cumulative confirmed covid cases were converted to weekly counts of new cases bỹ to subtract influenza signal from y it we assume that the population of patients with ili within a state are the same population that are potentially tested for influenza. this assumption allows us to calculate the number of non-influenza ili cases as the resulting time-seriesỹ it are shown in figure s4 . 4 . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. we identified ili surges inỹ it by training a model onỹ it for all data prior to july 21, 2019. we then used this model to predict the prevalence of non-influenza ili (π it ) for dates after and including july 21, 2019. we calculated the ili surge as the difference between the observed proportion of non-influenza iliỹ it /n it andπ it . more specifically, to account for variation in the number of reporting providers, we trained the following binomial logistic-normal model we made the following prior specifications: we set the bandwidth parameter for the squared exponential kernel as ρ = 3 representing a strong local correlation in time that died off sharply beyond 3 weeks, α = 1 representing a signal to noise ratio of of approximately 1, υ = 1 and 195 ξ = 1 representing weak prior knowledge regarding the overall scale of variation in the the latent space. finally, we set θ = −2.197 representing an off-season prevalance of 0.1% non-influenza ili. samples from the posterior predictive density p(π it |y i1 , . . . ,ỹ it , n i1 , . . . , n it ) were collected using the function basset from the r package stray [15]; a total of 4000 such samples were collected in this analysis. we define the prevalence of non-influenza ili in excess of normal seasonal variation to exclude variation attributable to unseasonably high rates of other ili causing viruses (such as the outbreak of rsv in washington state in november-december 2019) we only investigate y * it for sets of 1 or more weeks (t, t + 1, . . . , t + k) such that the following criteria are met: 1. the 2nd order moving average of y * i(t+k) < y * i(t+k+1) must remain positive for all k 205 2. at least one week in (t, t + 1, . . . , t + k) corresponds to a week after february 23, 2020 together these criteria for attributing excess non-influenza ili to covid reflect our assumption that we expect covid must be increasing within communities since febuary 23, 2020. as covid new case countsz it represent the number of confirmed cases in an entire state and ilinet data represents the number of cases seen by a select number of enrolled providers, we must estimate scaling factors w i to enable comparison of ilinet data to confirmed case counts at the who state level. let π * it denote the probability that a patient with ili in state i has covid as estimated from ilinet data. let p i denote the population of state i and let b i denote the number of primary care providers per 100,000 people in state i. we simulated the number of covid cases (excess ili meeting criteria above) as that is we translate the inferred proportion of individuals with ili due to covid to the state level by considering the average number of patients seen by each provider in the study ( nit dit ) and the number of primary care providers in state i ( bipi 10 5 ). notably to account for potential errors in these scaling factors, we add propogate uncertainty into our calculation by using monte-carlo simulation of the average number of patients seen by each provider in the study. 215 5 . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. assuming that the majority of sars-cov-2 testing within the us has been directed by patient symptoms [16] , the pool of newly diagnosed sars-cov-2+ patients is a subset of the pool of sars-cov-2+ patients who are identified as having ili. therefore, we calculate the probability that a symptomatic sars-cov-2+ patient will be identified as having sars-cov-2 as δ s it =z ij /π * it 220 ( figure s2 ). to account for the contribution of assymptomatic sars-cov-2+ patients we use a recent analysis of cohort surveillance from the diamond princess [10] . monte-carlo simulations were used to propagate error from our uncertainty regarding potential asymptomatic infections affecting the 225 clinical rate δ c into our calculation of posteriors for epidemic trajectories. to match posterior estimates from mizumoto et al. [10] , we use quantile matching to parameterize δ c ∼ beta(α, β) to achieve a mean of .179 and a 95% probability set of (.155, .202). we include this contribution to the overall detection rate δ it as δ it = δ s δ c it . for a given epidemic growth rate, the clinical rate was estimated by comparing the excess ili across the us y † t = i y † it to the number of infections one might expect under simulated epidemic parameterized by the growth rate. an seir model, was parameterized for the us to a timescale of units days by setting ζ = 3.23 × 10 −5 corresponding to a crude birth rate of 11.8 per 1000 per year, a baseline mortality rate ω b = 2.38 × 10 −7 corresponding to 8.685 per 1000 per year, an infectious mortality rate ω i = 2.62 × 10 −7 , incubation period γ −1 of 3 days, infectious period ν −1 of 10 days, and β parameterized to ensure i(t) grew with a specified exponential growth rate early in the epidemic. a total of 14,000 simulations were run, 235 each having a unique exponential growth rates, r, drawn uniformly from r ∈ [log(2)/7, log(2)/1.5], reflecting doubling times in the range of 1.5 ≤ t 2 ≤ 7 days. each simulation was initialized with (s, e, i, r, t) = (3.27 × 10 8 , 0, 1, 0, 0) where time 0 was january 15. to simulate the stochastic time it took from the first case to the onset of regular exponential growth, a gillespie algorithm was used from the initial conditions until either t = 50 240 (march 5, 2020) or e(t) + i(t) = 100. the output from gillespie simulations was input as an initial value into the system of differential equations and integrated until the august 5, 2020. the number of infected individuals on a given day was the last observed i(t) for that day, and a weekly pool of infected patients was computed by a moving sum over the number of infected individuals every day for the past week, i w (t) = k=6 k=0 i t−k . defining y t = i y † it as the national excess ili, the clinical rate was for a given time delay t d it takes from the onset of infectiousness to a patient reporting to the doctor with ili. an estimate of clinical rate as a function of exponential growth rate was obtained by a generalized additive model log(δ c ) = s(log(r)))+ where s(.) denotes a piece-wise cubic spline. simulated epidemic trajectories were matched to posterior samples of y t through kernel density estimation. for t corresponding to the weeks starting on march 8 and 15, 2020, the density each simulated trajectory i(t) was evaluated against the density v t ∼ log-normal(0, 0.1) 6 . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. . https://doi.org/10.1101/2020.04.01.20050542 doi: medrxiv preprint where noise v t was added to account for additional sources of error in our calculations of scaling 250 factors, and p is a density estimated via kernel density estimation. the two weeks in march were chosen as the ili surge had the largest correlation with covid case counts during these two weeks. doubling times for confirmed cases and deaths due to covid were estimated using poisson regression with a log-link with an intercept and a covariate for the day of the epidemic. poisson 255 regression parameters r were converted to doubling times d using d = log(2)/r. . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. non−influenza ili proportion figure 1 : the excess non-influenza ili is extracted from all non-influenza ili by identifying the amount of non-influenza ili in excess of seasonal norms (blue point and error bars represent the posterior median and 95% credible set for ili not explained by non-covid endemic respiratory pathogens). a binomial logistic-normal non-linear regression model was fit to non-influenza ili data from 2010-2018 (grey lines). the model predicted the expected amount of non-influenza ili in the 2019-2020 season (grey ribbons represent the 95% and 50% credible sets; the black line represents the posterior median). observed non-influenza ili outside of the 95% credible set was attributed to covid based on the observed correlation across states with newly confirmed covid case rates ( figure s1 ). a number of regions are not represented due to insufficient laboratory influenza data to complete our analysis (see methods for full details). . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. : assuming the non-influenza excess ili for the week starting march 15 consists entirely of patients with covid, the probability that a symptomatic covid+ patient will be detected varies by state but even the highest symptomatic case detection rates are likely below 3%.in figure s2 we show how the symptomatic case detection rate varies over time across states. . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. . figure 3 : (a) 14,000 random seir models predict a range of current number of covid cases in the us; less transparant trajectories are better agreements with the ili surge under an assumption that 17% of infections are subclinical [10] . the inset plot shows covid forecasts comparable to weekly ili counts (red dots) by summing i(t) over the prior week. a t 2 = 3.5 day doubling time (solid black lines) is the slowest possible growth rate which can account for the observed data with a 4-day lag between the onset of infectiousness and reporting as ili. faster doubling times, such as t 2 = 1.6 (dashed black lines) can explain observed ili data but require a larger rate of subclinical infections to do so. (b) different growth rates imply different clinical rates in order to account for the excess ili surge, and the relationship between growth rate and clinical rate varies with the unknown time it takes from the onset of infectiousness to presentation an ili outpatient. the estimated 2.4-2.5 day doubling times across the us, for example, produce an estimated 3-4% clinical rate. . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. figure s3 . 12 . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. figure s1 : excess ili correlates strongly with patterns of newly confirmed covid cases. this correlation is strongest for the last two weeks of data, when other seasonal respiratory pathogens are at their lowest. 13 . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. figure s2 : since march 1, 2020, the case-detection of symptomatic covid patients has increased by a factor of ≈ 100. this likely represents increased awareness of community transmission within the us combined with increased availability of testing. still, the symptomatic case detection rate remains below 1% for most states with many states with detection rates closer to 0.1%. 14 . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted april 3, 2020. 15 . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted april 3, 2020. 30 35 40 45 50 2 7 12 17 22 27 30 35 40 45 50 2 7 12 17 22 27 30 35 40 45 50 2 7 12 17 22 27 30 35 40 45 50 2 7 12 17 22 27 30 35 40 45 50 2 7 12 17 22 27 30 35 40 45 50 2 7 12 17 22 27 30 35 40 45 50 2 7 12 17 22 27 30 35 40 45 50 2 7 12 17 22 figure s4 : once the signal attributable to influenza is extracted, the proportion of patient encounters in which patient had non-influenza ili (ỹ it /n it ) displays strong seasonal trends. the most notable deviations from these trends occur around febuary to march of the 2019-2020 flu season and align with the onset of the covid epididemic within the us. 16 . cc-by 4.0 international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted april 3, 2020. . https://doi.org/10.1101/2020.04.01.20050542 doi: medrxiv preprint a novel coronavirus from patients with pneumonia in china coronavirus disease 2019 (covid-19): situation report fundamental principles of epidemic spread highlight the immediate need for large-scale serological surveys to assess the stage of the sars-cov-2 epidemic epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in wuhan, china: a descriptive study clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in wuhan, china substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (sars-cov2) nowcasting and forecasting the potential domestic 275 and international spread of the 2019-ncov outbreak originating in wuhan, china: a modelling study first case of 2019 novel coronavirus in the united states estimating the asymptomatic proportion of coronavirus disease 2019 (covid-19) cases on board the diamond princess cruise ship, yokohama, japan key: cord-305460-wln758og authors: alqahtani, amani salem; tashani, mohamed; heywood, anita elizabeth; almohammed, abdulrahman bader s.; booy, robert; wiley, kerrie elizabeth; rashid, harunor title: tracking australian hajj pilgrims’ health behavior before, during and after hajj, and the effective use of preventive measures in reducing hajj-related illness: a cohort study date: 2020-05-04 journal: pharmacy (basel) doi: 10.3390/pharmacy8020078 sha: doc_id: 305460 cord_uid: wln758og this study assessed australian hajj pilgrims’ knowledge, attitude and practices throughout their hajj journey to understand their health behaviors, use of preventative measures and development of illness symptoms. a prospective cohort study with data collection at three phases (before, during and after hajj) was conducted among australian pilgrims between august and december 2015. baseline data were collected from 421 pilgrims before hajj, with 391 providing follow-up data during hajj and 300 after their home return. most participants (78% [329/421]) received one or more recommended vaccines; travel agents’ advice was the main factor affecting vaccination uptake. most participants (69% [270/391]) practiced hand hygiene with soap and sanitizers frequently, followed by disposable handkerchief use (36% [139/391]) and washing hands with water only (28% [111/391]). during hajj 74% (288/391) of participants reported one or more illness symptoms, 86% (248/288) of these symptoms were respiratory. cough was less often reported among pilgrims who received vaccinations, cleaned their hands with soap or alcoholic hand rubs, while a runny nose was less common among those who frequently washed their hands with plain water but was more common among those who used facemasks. this study reveals that most australian hajj pilgrims complied with key preventative measures, and that tour group operators’ advice played an important role in compliance. pilgrims who were vaccinated and practiced hand hygiene were less likely to report infection symptoms. the hajj pilgrimage attracts over two million pilgrims every year to saudi arabia, making it one of the biggest annual human mass gatherings on the planet [1] . consequently, with such large numbers of pilgrims from around the world in close proximity to one another, there is amplified risk of transmission of infectious diseases among those present [2] . to minimize the risks, saudi ministry of health (moh) require a valid vaccination certificate against meningococcal disease for all pilgrims, plus yellow fever and polio vaccines for pilgrims coming from or transiting through endemic countries. additionally, vaccinations against influenza, diphtheria, pertussis, tetanus, mumps and measles are recommended, particularly for pilgrims susceptible to more severe disease [3] . other relatively inexpensive preventative measures such as hand hygiene and facemask use are also recommended [4] . studies have consistently shown that there is an extensive variation in vaccine uptake among pilgrims based on their country of origin, demographics and vaccines received. similarly, hand hygiene and other preventative practices also vary among pilgrims, making it more difficult for researchers to ascertain whether vaccine uptake and health behaviors overall have improved in comparison to previous years or studies [5] [6] [7] . although there is some research linking knowledge, attitudes, and beliefs of hajj pilgrims to their use of preventive measures, no studies have explored their knowledge, preparedness and preventive practices over the course of their travel (i.e., following the same cohort before, during and after hajj) [5, [8] [9] [10] [11] . to address these research gaps, we conducted a cohort study to explore australian hajj pilgrims' knowledge about the risk of diseases during hajj, assess their preparedness and use of preventive measures at three times points (before, during and after hajj) , investigate the factors affecting their preventive health behavior, and determine the number of reported infections during and after hajj. this was a prospective cohort study undertaken between august and december 2015 among australian hajj travelers aged 18 years or older. the participants were residents of greater sydney, new south wales, australia. this region was chosen because it contains the largest muslim population in new south wales, australia. the participants were assessed before, during and after returning from their pilgrimage to makkah, saudi arabia. this study was approved by the university of sydney human research ethics committee (hrec) (project no: 2014/599). all overseas pilgrims travelling to saudi arabia require a visa which can only be acquired through approved hajj travel agents [12] . these tour operators play a pivotal role in pilgrims' preparation leading up to hajj, relaying important travel instructions prior to departure through 'pre-hajj seminars' which almost all prospective pilgrims attend [13] . this made these seminars an ideal recruitment points for a representative sample. the list of accredited australian hajj tour operators and their addresses were collected from the saudi embassy in canberra, australia. a selection of participating travel agencies was decided on the basis of the number of hajj visas allocated for a given tour operator and the operators with the highest visa quotas were approached first. three questionnaires were used for data collection: (1) a pre-hajj questionnaire completed prior to departure; (2) a 'hajj diary' completed daily over six days during the peak hajj period; and (3) a post-hajj questionnaire, completed upon return to australia. the surveys were mainly in english, with the availability of arabic, turkish and urdu translations for those who preferred it. we attended eleven seminars held by the hajj travel agents in sydney from 1 august to 6 september 2015 to explain our research. once pilgrims consented to participate in the study, their characteristics were obtained using a self-administered questionnaire documenting socio-demographic details, their travel itineraries, vaccination records, and presence of chronic medical conditions such as hypertension, diabetes, bronchial asthma and hyperlipidemia. barriers to, and facilitators of vaccine uptake and data on risk perception of diseases at hajj, such as influenza, pneumonia, blood borne infections and their preparedness for using preventive measures during hajj were also collected. we did not ask about the receipt of meningococcal vaccine because our previous survey showed that all australian pilgrims receive this as a mandatory visa requirement [13] . we travelled to makkah, saudi arabia, during hajj period (15-30 september 2015) and identified and met the study participants upon their arrival in mina, greater makkah. the participants were given a diary for self-completion daily over the six consecutive days of hajj (the peak hajj days). they were followed from 21-26 september 2015. each participant was asked to record the following in the diary for each day: actual use of preventative measures including, facemask use, hand sanitizer use, hand washing after touching an ill person and use of disposable handkerchiefs. as the data were collected electronically, no respondent could submit their responses with vitally important information missing. any respondent who used a preventive measure ≥5 days during the peak hajj days considered to be 'frequently compliant', those who used the preventive measure <5 days were considered to be 'infrequently compliant' and those who did not use the preventive measure at all were considered 'non-compliant'. any development of symptoms suggestive of respiratory infection including cough, subjective fever sore throat, rhinitis and other symptoms including vomiting, diarrhea and nausea was also recorded. we considered those who complained of cough, subjective fever and sore throat to meet the criteria of influenza-like illness (ili). the respondents participated in a computer-aided telephone interview (cati) seven to 10 days after their coming back to australia (until 26 december 2015). they were asked about any development of symptoms suggestive of respiratory infection such as cough, sore throat, runny nose, fever, and also about some constitutional or gastrointestinal symptoms like vomiting, diarrhea and nausea after their home return from hajj. facilitators of and barriers to using preventive measures during hajj were also explored. for each participant we made three call attempts before being classified as lost to follow-up. a consecutive sampling strategy was used to ensure a representative sample of hajj pilgrims living in nsw. based on results from previous research, and considering an error margin of 5% to be acceptable for this survey, a sample size of 350 pilgrims was deemed to be sufficient. considering a loss to follow-up rate of 20%, 420 participants were targeted, representing 12% of australian pilgrims attending hajj in 2015. statistical analysis was performed using the spss v.23.0 (spss, inc., chicago, il, usa). chi-squared tests and pearson correlation were used to assess variables and establish associations and correlations. factors with p values < 0.25 in univariate analysis were entered into multivariable regression models. binary logistic regression, using the backward wald method, controlling for factors was used to investigate variables related to health behavior. p values less than 0.05 were considered statistically significant in multivariable models. a three-point likert scale was used to measure the pilgrims' perception about the risk of diseases during hajj and the effectiveness of the preventive measures. a total of 421 pilgrims were recruited in the first stage of the study (before hajj); out of those, 391 (93%) were followed during hajj; and finally, 300 (71%) were reached after their return to australia ( figure 1 ). of 421, 46% were female and their mean age was 42.2 (standard deviation [sd] ± 11.2) years. regression models. binary logistic regression, using the backward wald method, controlling for factors was used to investigate variables related to health behavior. p values less than 0.05 were considered statistically significant in multivariable models. a three-point likert scale was used to measure the pilgrims' perception about the risk of diseases during hajj and the effectiveness of the preventive measures. a total of 421 pilgrims were recruited in the first stage of the study (before hajj); out of those, 391 (93%) were followed during hajj; and finally, 300 (71%) were reached after their return to australia (figure 1 ). of 421, 46% were female and their mean age was 42.2 (standard deviation [sd] ± 11.2) years. over a quarter (28%) reported having one or more pre-existing medical conditions. over one third of participants (39%) had a university degree or higher qualification and two thirds (66%) were employed. the pilgrims intended to stay in saudi arabia for a median of 25 (range: 10-45) days, and the majority (81%) were attending hajj for the first time. the participants' further demographic details are presented in table 1 . the majority of participants (78%, [329/421]) received one or more recommended vaccines (i.e., vaccines that are recommended but not compulsory); of those, 43% (180/421) received only the influenza vaccine, 33% (139/421) received influenza plus other recommended vaccines while the remaining 2% (10/421) received only "other than influenza" vaccines. overall, the coverage of influenza, pharmacy 2020, 8, 78 6 of 15 pneumococcal and pertussis vaccine was, respectively, 76% (319/421), 25% (107/421) and 21% (88/421) ( table 2) . on the other hand, 22% (92) did not receive any recommended vaccine. * some pilgrims cited more than one reason. thirty-one percent (129/421) were in an at-risk category; of those, 20% (26/129) were aged ≥65 years and 80% (103/129) aged <65 but had one or more chronic medical condition (e.g., diabetes). the influenza vaccination rate among at-risk group was 76% (98/129), while the uptake of pneumococcal vaccine was 21% (27/129) . no significant differences of influenza and pneumococcal vaccine uptake were noted between 'at risk' and not 'at risk' groups. participants reported different sources of vaccination advice including, hajj (table 2 ). multivariate analysis showed that being aged between 36 and 64 years was significantly associated with the receipt of recommended vaccines compared to being aged ≤35 years or ≥65 years (adjusted odds ratio [aor] = 2.1, 95% confidence interval [ci] = 1.2-3.6, p < 0.01). moreover, those who had a university qualification or higher education were more likely to receive vaccine than those who had lower level of education (aor = 4.1, 95% ci = 1.2-13.1, p < 0.01). . those who were very concerned about blood borne diseases (aor = 2.1, 95% ci = 1.1-4.3, p = 0.02) and pneumonia (aor = 1.8, 95% ci = 1.1-3.2, p < 0.01) were twice as likely to receive hepatitis b and pneumococcal vaccines respectively. yet, no association was found between the level of concern about influenza and the receipt of influenza vaccine. (table 3 ). in addition, 19% (76/391) of pilgrims used antibiotics during hajj. there was no significant association between pilgrims' intended use of non-pharmacological measures and their actual use during hajj (table 4) . in multivariate analysis, no demographic factors were associated with using hand hygiene with soap and sanitizers, but those who were concerned about developing pneumonia during hajj (aor = 2.1, 95% ci = 1.1-4.3, p = 0.04) were more likely to practice hand hygiene with soap and sanitizers during hajj compared to those who were not concerned. those aged ≥65 years (aor = 3.1, 95% ci = 1.1-8.8, p = 0.02) were more likely to practice hand hygiene with alcoholic hand rubs than those aged <65 years. moreover, males were more likely to wash their hands with water only (aor = 1.9, 95% ci = 1.2-2.9, p < 0.01) and wash hands after touching an ill person (aor = 2.4, 95% ci = 1.3-4.6, p < 0.01) but were less likely to use disposable handkerchiefs (or = 0.4, 95% ci = 0.3-0.7, p < 0.01) compared to females. on the first day of hajj, some pilgrims were symptomatic; 5% had a cough, 5% had a sore throat, and 3% had a runny nose. however, on the 4th day up to 16% became symptomatic. the reported onset of daily symptoms among pilgrims during the peak days of hajj is presented in figure 2 . on the first day of hajj, some pilgrims were symptomatic; 5% had a cough, 5% had a sore throat, and 3% had a runny nose. however, on the 4th day up to 16% became symptomatic. the reported onset of daily symptoms among pilgrims during the peak days of hajj is presented in figure 2 . overall, 74% (288/391) of participants reported one or more illness symptoms throughout the hajj journey. eighty-six percent (248/288) of these symptoms were respiratory, including cough, 45% (176/391); sore throat, 44% (171/391); runny nose, 26% (103/391); and fever, 15% (59/391). ili was only reported among 10% (40/391) of participants. additionally, 16% (64/391) reported diarrhea, 12% (46/391) nausea and another 5% (21/391) reported vomiting. twenty six per cent (103/391) of respondents did not report any symptom during hajj. over half (52% [157/300]) reported symptoms after returning from hajj; of those, 37% (111/300) reported cough (41%, [45/111] of these also had cough during hajj), 25% (76/300) sore throat (43%, [33/76] of them reported sore throat during hajj), 16%, (47/300) runny nose (30%, [14/47] reported runny nose during hajj) and 11% (32/300) had fever (16%, [5/32] had fever during hajj). ili was reported in 10% (29/300) of participants; of those, 10% (3/29) also reported ili during hajj. moreover, 5% (15/300) reported diarrhea, 3% (8/300) nausea and another 3% (8/300) reported vomiting. as shown in table 5 , cough was less likely to occur among those who were vaccinated against influenza and those who used alcoholic hand rubs. runny nose was less likely to occur who frequently washed their hands with plain water but was more common among those who used facemasks. vomiting was less likely to be reported among those who washed their hands frequently with soap and sanitizers during hajj, but more common in those who used disposable handkerchiefs. no association was found between use of any of the preventive measures and reduction in fever and sore throat. overall, 74% (288/391) of participants reported one or more illness symptoms throughout the hajj journey. eighty-six percent (248/288) of these symptoms were respiratory, including cough, 45% (176/391); sore throat, 44% (171/391); runny nose, 26% (103/391); and fever, 15% (59/391). ili was only reported among 10% (40/391) of participants. additionally, 16% (64/391) reported diarrhea, 12% (46/391) nausea and another 5% (21/391) reported vomiting. twenty six per cent (103/391) of respondents did not report any symptom during hajj. over half (52% [157/300]) reported symptoms after returning from hajj; of those, 37% (111/300) reported cough (41%, [45/111] of these also had cough during hajj), 25% (76/300) sore throat (43%, [33/76] of them reported sore throat during hajj), 16%, (47/300) runny nose (30%, [14/47] reported runny nose during hajj) and 11% (32/300) had fever (16%, [5/32] had fever during hajj). ili was reported in 10% (29/300) of participants; of those, 10% (3/29) also reported ili during hajj. moreover, 5% (15/300) reported diarrhea, 3% (8/300) nausea and another 3% (8/300) reported vomiting. as shown in table 5 , cough was less likely to occur among those who were vaccinated against influenza and those who used alcoholic hand rubs. runny nose was less likely to occur who frequently washed their hands with plain water but was more common among those who used facemasks. vomiting was less likely to be reported among those who washed their hands frequently with soap and sanitizers during hajj, but more common in those who used disposable handkerchiefs. no association was found between use of any of the preventive measures and reduction in fever and sore throat. this cohort study captured and compared the health behavior, knowledge, attitudes and practices of australian hajj pilgrims regarding preventative measures against communicable diseases throughout the course of hajj travel (before, during and after the journey). vaccinated pilgrims and those who washed their hands and used alcohol rubs were less likely to develop respiratory symptoms during hajj. influenza vaccination coverage was relatively higher (76%) among australian pilgrims, compared to that reported in studies from other countries, but was lower compared to that in australian pilgrims in 2014 (80%) and in 2013 (83%) [5, 14] . in contrast, coverage of other recommended vaccines, such as the pneumococcal vaccine, was suboptimal (25%), as has been reported in previous studies involving australian hajj pilgrims and other international pilgrims [15, 16] . the coverage of influenza vaccine among pilgrims with high-risk conditions was 76%, while that of pneumococcal vaccine was only 21%; compared to the influenza vaccination coverage among pilgrims with high-risk conditions from other countries, australian hajj pilgrims had a higher vaccination rate [5] . the low pneumococcal vaccine uptake is a concern because pneumococcal diseases, including invasive pneumococcal disease, are major causes of morbidity and mortality in the extremes of age and in individuals with chronic medical conditions worldwide, including hajj pilgrims [17, 18] . of note, while the influenza vaccine is highly recommended by the saudi moh for hajj attendance, particularly for pilgrims aged ≥65 years [4] , there is no formal recommendation regarding the pneumococcal vaccine. this study found that a large number of participants were willing to use non-pharmacological preventative measures to reduce the prevalence of illnesses during hajj. however, there was no significant correlation between their intention to use measures and their actual use during hajj. although the level of concern for pneumonia and diarrhea was higher among first time hajj goers and pilgrims <65 years of age, these factors were not shown to be significant in their actual use of preventive measures during hajj. pilgrims who were concerned about catching diseases and those who joined hajj for the first time were more likely to accept the use of non-pharmacological measures before hajj. however, these intentions were not associated with actual use of preventive measures. previous studies have not explored if demographic factors were associated with an intention to use non-pharmacological protective measures before travel and their actual use during hajj. nonetheless, earlier studies did conclude that health education prior to departure was significantly associated with greater compliance with preventative practices, particularly the use of facemasks and hand sanitizers [19] [20] [21] [22] . demographic factors had minimal association with pilgrims' health behavior such as vaccine uptake, and the factors associated with pilgrims' willingness to use preventive measures were not associated with their actual use during hajj. consequently, these findings pose questions of other possible causes that might affect pilgrims' behaviors during hajj. in this study, several factors were identified as influencing pilgrims' health behavior, including disease risk perception, awareness of recommendations, the influence of people around them, age and medical history, the source of their travel health advice, and the lived experience of using preventive measures during hajj. these factors can be stratified into five categories: individual, interpersonal, organizational, community-associated and policy-related. relying on a single factor does not fully explain the interplay and dynamics between pilgrims' health behaviors and their influencers; rather, consideration of multiple factors at different levels may help to understand these cross-cutting relationships. these dimensions fit within the 'social-ecological model' which emphasizes the connectivity and relationship among multiple factors affecting health behavior, as shown in figure 3 . the overlapping circles in the model illustrate how factors at one level influence factors at another level. this framework is based on evidence that no single factor can explain why only some people comply with preventive measures while remain influenced by multiple inter-related factors. the core of the model is at the individual level, surrounded by the outer four bands representing the interpersonal, organizational, community and policy levels ( figure 3 ). it is important to understand and find the most common or influential reasons and the link between these levels to enhance pilgrims' health behavior uptake in regards to each measure. in addition, while some interventions such as vaccines are known to be clinically effective, their acceptability to pilgrims is a vital part of their overall effectiveness as a public health intervention. effective behavioral change is only made when changes interconnect between the individual, community, organizational and policy levels. this includes addressing underlying and related factors such as individual beliefs, culture, and miscommunication between organizations as well as government policies. this can help promote lasting changes in practices; on an individual level, cultural and religious beliefs; in attitudes and perception; in communication between the health care providers such as gps and the community; providing the service suppliers such as the travel agent with up-to-date health recommendations and also improving the communication between the international and local health policies (figure 3 ) [23] . future studies could be grounded in 'social ecological' theory, in order to further articulate the inter-level relationships between health behavior factors identified in this study, encouraging new insights to promote the health interventions among hajj pilgrims. care providers such as gps and the community; providing the service suppliers such as the travel agent with up-to-date health recommendations and also improving the communication between the international and local health policies (figure 3 ) [23] . future studies could be grounded in 'social ecological' theory, in order to further articulate the inter-level relationships between health behavior factors identified in this study, encouraging new insights to promote the health interventions among hajj pilgrims. during hajj, about 63% of participants reported developing one or more respiratory symptoms; a cough and sore throat were the commonest symptoms similar to what was found in earlier studies involving australian pilgrims who attended hajj in 2014 [24] , and among french, iranian and malaysian pilgrims who attended hajj between 2003 and 2012 [25] [26] [27] . during hajj, ili was reported by 10% of pilgrims in this study, a similar rate was reported previously among australian hajj pilgrims in 2013 [28] . in other studies involving hajj pilgrims the reported incidence of ili varied (from 10% to 70%) [29] . the risk of disease was equally high among returning pilgrims, with 30% to 40% of them reporting cough, sore throat and runny nose, and 10% suffering from ili as found in other studies [29] [30] [31] . countries should continue to monitor the pilgrims and their contacts after the pilgrims' return from hajj to estimate the risk of post-hajj infections. influenza vaccine effectiveness was studied in several studies. in two studies conducted in 2003 and 2012, a reduction in ili was observed, while in a few other earlier studies no significant reduction was noted [5] . a synthesis of published and raw data from eleven hajj years between 2005 and 2014 showed that the rate of ili decreased among hajj pilgrims as the vaccination rate increased (relative risk 0.2, p < 0.01) [32] . subsequently, a 'test-negative' case-control study using data from individual hajj years involving participants from multiple countries shows trivalent influenza has an effectiveness of 43.4% (95% ci 11.4% to 63.9%, p = 0.01) against laboratory-confirmed influenza [33] . the varying results may due to heterogeneity in defining ili or may be because of ili symptoms representing non-influenza infections or even due to confounding factors such as use of facemasks or hygienic interventions [30, 31, 34] . there are some limitations; 29% respondents failed to complete the study. this is a little higher than the expected loss to follow up (20%), however as this study was conducted over three months during hajj, about 63% of participants reported developing one or more respiratory symptoms; a cough and sore throat were the commonest symptoms similar to what was found in earlier studies involving australian pilgrims who attended hajj in 2014 [24] , and among french, iranian and malaysian pilgrims who attended hajj between 2003 and 2012 [25] [26] [27] . during hajj, ili was reported by 10% of pilgrims in this study, a similar rate was reported previously among australian hajj pilgrims in 2013 [28] . in other studies involving hajj pilgrims the reported incidence of ili varied (from 10% to 70%) [29] . the risk of disease was equally high among returning pilgrims, with 30% to 40% of them reporting cough, sore throat and runny nose, and 10% suffering from ili as found in other studies [29] [30] [31] . countries should continue to monitor the pilgrims and their contacts after the pilgrims' return from hajj to estimate the risk of post-hajj infections. influenza vaccine effectiveness was studied in several studies. in two studies conducted in 2003 and 2012, a reduction in ili was observed, while in a few other earlier studies no significant reduction was noted [5] . a synthesis of published and raw data from eleven hajj years between 2005 and 2014 showed that the rate of ili decreased among hajj pilgrims as the vaccination rate increased (relative risk 0.2, p < 0.01) [32] . subsequently, a 'test-negative' case-control study using data from individual hajj years involving participants from multiple countries shows trivalent influenza has an effectiveness of 43.4% (95% ci 11.4% to 63.9%, p = 0.01) against laboratory-confirmed influenza [33] . the varying results may due to heterogeneity in defining ili or may be because of ili symptoms representing non-influenza infections or even due to confounding factors such as use of facemasks or hygienic interventions [30, 31, 34] . there are some limitations; 29% respondents failed to complete the study. this is a little higher than the expected loss to follow up (20%), however as this study was conducted over three months in three different settings (before, during and after hajj) in two countries (australia and saudi arabia), it was very challenging to follow all participants, some of whom may have taken a side trip to other countries at the end of the hajj. a french study revealed that over a quarter of the pilgrims intended to delay their return to france after hajj, a situation that makes difficult to ensure maximum follow up [35] . information and recall bias may have occurred due to the anecdotal nature of some survey questions, especially those of the post-hajj survey. although our results cannot be generalized to all pilgrims, this study is the first of its kind to assess pilgrims' kap continuously throughout the hajj journey to understand their health behaviors, experience of using preventative measures and the development of acute respiratory infections, and other symptoms of infections. in the meantime, studies conducted in france and malaysia have shown that compliance to some preventive measures such as hand hygiene and face mask use has increased, while the uptake of recommended vaccines including influenza and pneumococcal vaccines still remains low [36] [37] [38] ; therefore, awareness campaigns should be continued to tackle respiratory infections including the ongoing covid-19 pandemic that, thus far, has taken a toll of over 200,000 people across the world (as of 27 april 2020) and affected many muslim countries including saudi arabia [39] . to mitigate the epidemic, the saudi arabian authority has already temporarily cancelled umrah (minor pilgrimage) visit to makkah [40] ; the decision on whether this year's hajj pilgrimage (late july to early august) should be cancelled or not remains to be decided and may depend on the progress of the pandemic [41] . our findings mean that preventive measures like hand washing and use of alcoholic hand rubs could be implemented readily during hajj, and tour operators may play important roles in improving compliance. in conclusion, this study reveals that most australian hajj pilgrims complied with key preventative measures, and that tour group operators' advice played an important role in compliance. pilgrims who complied with preventive measures were less likely to suffer from infection symptoms. researchers and policy makers should work together to explore ways to educate the pilgrims and their tour operators about the importance of vaccination and using simple but inexpensive preventive measures, such as hand hygiene, that may halt the spread of highly contagious infectious diseases, e.g., covid-19, pandemic influenza and drug-resistant pathogens in mass gatherings. transmission of respiratory tract infections at mass gathering events preparing australian pilgrims for the hajj health conditions for travellers to saudi arabia for the pilgrimage to mecca (hajj)-2015 vaccinations against respiratory tract infections at hajj vaccination in hajj: an overview of the recent findings non-pharmaceutical interventions for the prevention of respiratory tract infections during hajj pilgrimage protective measures against acute respiratory symptoms in french pilgrims participating in the hajj of 2009: table 1 predictors of protective behaviors among american travelers to the hajj pilgrims' knowledge about acute respiratory infections health risks at the hajj exploring australian hajj tour operators' knowledge and practices regarding pilgrims' health risks: a qualitative study hajj research team. influenza vaccination among australian hajj pilgrims: uptake, attitudes, and barriers pneumococcal vaccine uptake among australian hajj pilgrims in 2011-2013 attitude and practice (kap) survey concerning antimicrobial use among australian hajj pilgrims burden of clinical infections due to s. pneumoniae during hajj: a systematic review the potential for pneumococcal vaccination in hajj pilgrims: expert opinion using health educators to improve knowledge of healthy behaviour among hajj 1432 (2011) pilgrims hand hygiene compliance and effectiveness against respiratory infections among hajj pilgrims: a systematic review effectiveness of an education health programme about middle east respiratory syndrome coronavirus tested during travel consultations uptake and effectiveness of facemask against respiratory infections at mass gatherings: a systematic review an ecological perspective on health promotion programs pilot use of a novel smartphone application to track traveller health behaviour and collect infectious disease data during a mass gathering: hajj pilgrimage influenza viral infections among the iranian hajj pilgrims returning to shiraz, fars province circulation of respiratory viruses among pilgrims during the 2012 hajj pilgrimage the prevalence of acute respiratory symptoms and role of protective measures among malaysian hajj pilgrims acute febrile respiratory infection symptoms in australian hajjis at risk of exposure to middle east respiratory syndrome coronavirus prevalence of influenza at hajj: is it correlated with vaccine uptake? health related experiences among international pilgrims departing through king abdul aziz international airport the prevalence and preventive measures of the respiratory illness among malaysian pilgrims in 2013 hajj season changes in the prevalence of influenza-like illness and influenza vaccine uptake among hajj pilgrims: a 10-year retrospective analysis of data influenza vaccine effectiveness among hajj pilgrims: a test-negative case-control analysis of data from different hajj years pilot randomised controlled trial to test effectiveness of facemasks in preventing influenza-like illness transmission among australian hajj pilgrims in 2011 travel reported by pilgrims from marseille, france before and after the 2010 hajj: table 1 hajj in the time of covid-19 respiratory tract infections among french hajj pilgrims from abubakar baaba, a. assessment of knowledge, attitude and practice towards prevention of respiratory tract infections among hajj and umrah pilgrims from malaysia uptake of recommended vaccines and its associated factors among malaysian pilgrims during hajj and umrah saudi arabias measures to curb the covid-19 outbreak: temporary suspension of the umrah pilgrimage the cancellation of mass gatherings (mgs)? decision making in the time of covid-19 this article is an open access article distributed under the terms and conditions of the creative commons attribution (cc by) license key: cord-297609-6g39lu1y authors: wertheim, heiman f l; nadjm, behzad; thomas, sherine; malik, suhud; nguyen, diep ngoc thi; vu, dung viet tien; van nguyen, kinh; van nguyen, chau vinh; nguyen, liem thanh; tran, sinh thi; phung, thuy bich thi; nguyen, trung vu; hien, tran tinh; nguyen, uyen hanh; taylor, walter; truong, khanh huu; ha, tuan manh; chokephaibulkit, kulkanya; farrar, jeremy; wolbers, marcel; de jong, menno d; van doorn, h rogier; puthavathana, pilaipan title: viral and atypical bacterial aetiologies of infection in hospitalised patients admitted with clinical suspicion of influenza in thailand, vietnam and indonesia date: 2015-10-13 journal: influenza other respir viruses doi: 10.1111/irv.12326 sha: doc_id: 297609 cord_uid: 6g39lu1y background: influenza constitutes a leading cause of morbidity and mortality worldwide. there is limited information about the aetiology of infection presenting clinically as influenza in hospitalised adults and children in south-east asia. such data are important for future management of respiratory infections. objectives: to describe the aetiology of infection presenting clinically as influenza in those hospitalised in south-east asia. methods: respiratory specimens archived from july 2008 to june 2009 from patients hospitalised with suspected influenza from indonesia, thailand and vietnam were tested for respiratory viruses and atypical bacteria by polymerase chain reaction. results: a total of 1222 patients’ samples were tested. of 1222, 776 patients (63·5%) were under the age of 5. viruses detected included rhinoviruses in 229 of 1222 patients (18·7%), bocaviruses in 200 (16·4%), respiratory syncytial viruses in 144 (11·8%), parainfluenza viruses in 140 (11·5%; piv1: 32; piv2: 12; piv3: 71; piv4: 25), adenovirus in 102 (8·4%), influenza viruses in 93 (7·6%; influenza a: 77; influenza b: 16) and coronaviruses in 23 (1·8%; oc43: 14; e229: 9). bacterial pathogens were mycoplasma pneumoniae (n = 33, 2·7%), chlamydophila psittaci (n = 2), c. pneumoniae (n = 1), bordetella pertussis (n = 1) and legionella pneumophila (n = 2). overall, in-hospital case fatality rate was 29 of 1222 (2·4%). conclusion: respiratory viruses were the most commonly detected pathogens in patients hospitalised with a clinical suspicion of influenza. rhinovirus was the most frequently detected virus, and m. pneumoniae, the most common atypical bacterium. the low number of detected influenza viruses demonstrates a low benefit for empirical oseltamivir therapy, unless during an influenza outbreak. influenza is a common reason for primary care consultation and constitutes a leading cause of hospitalisation, morbidity and mortality worldwide. 1 non-influenza viruses are the most common causes of illness that clinically resemble influenza, particularly in children. data on the epidemiology and disease burden of influenza-related disease in south-east asia (sea) are emerging, and this, in turn, is shedding more light on the epidemiology of other viral and bacterial aetiologies of influenza-like illnesses (ilis) in hospitalised adults and children in this region. 2 the insight provided by this information is important not only for future prevention strategies, treatment and clinical management of respiratory infections, but also to guide future studies in this region. recently, many diagnostic advances have been made in order to help to determine the aetiology of acute respiratory infections. these advances have been made possible by molecular diagnostic techniques that are now in widespread use. furthermore, research in this field has gained much attention and funding due to the outbreaks of severe acute respiratory syndrome coronavirus (sars-cov) and avian influenza viruses a/h5n1 and a/h7n9 in sea and middle eastern respiratory syndrome coronavirus (mers-cov) in the middle east. generally, the causative agents of ili remain undiagnosed in routine clinical practice due to the expense of testing and the slow turnaround time for most diagnostic polymerase chain reactions (pcrs), considered too long to impact the management of usually self-limiting respiratory tract infections. however, with the increasing use of antiviral drugs, such as neuraminidase inhibitors for influenza, testing for ili pathogens may become more important for rational antiviral drug treatment and to prevent overuse of antibiotics, thereby helping to control costs and prevent the emergence and spread of antimicrobial drug resistance. during the past few years, several novel causative agents of ili have been identified in respiratory specimens, including human metapneumovirus, new human coronaviruses (nl63, hku1, sars-cov, mers-cov), rhinovirus c and continuously emerging novel lineages and subtypes of influenza virus a capable of infecting humans (h5n1, h5n6, h1n1pdm09, h3n2v, h7n7, h7n9, h9n2, h6n1, h10n8). viruses for which disease causation has not (yet) been established have also been identified, including human bocavirus and respiratory polyomaviruses, and likely there are many more to come. [3] [4] [5] [6] the most comprehensive data on the causes of ili originate from developed countries, with some data available from sea. recently, studies have been published from asian countries, [7] [8] [9] [10] [11] [12] providing much needed, valuable information and showing a similar spectrum of viral pathogens. this study was designed to determine the viral and atypical bacterial aetiologies of ilis in patients in sea using molecular techniques. the detection of influenza viruses and respiratory pathogens other than influenza viruses in specimens that were sent for routine influenza testing can provide valuable insights into the epidemiology of ili across all age groups in three sea countries over the same time period. furthermore, this study could address the potential occurrence of multiple simultaneous infections in the same patient. we performed a laboratory-based surveillance study, in which we tested all archived respiratory specimens from hospitalised patients over 1 year (july 2008 to june 2009) with clinically diagnosed ili meeting the following criteria: age ≥1 year, signs of ili according to treating physician, duration of ili symptoms ≤10 days and respiratory specimens sent for influenza testing. clinicians at all participating hospitals received regular training in who protocols for the management and recognition of influenza/ili. archived respiratory specimens (nose swab, throat swab, nasopharyngeal aspirate, nasal wash, tracheal aspirate, bronchoalveolar lavage) were obtained from 12 sites in three countries across sea (three sites in indonesia, four sites in thailand, five sites in vietnam). influenza testing capacity was created for a randomised clinical study comparing two dosages of oseltamivir for severe influenza by sites in the south east asia infectious diseases clinical research network, as described elsewhere. 13 the specimens used for this study were stored at à80°c until further testing. where more than one specimen type was available for each patient, they were prioritised in the sequence endotracheal aspirate/bronchoalveolar lavage > nasopharyngeal aspirate > throat swab > nasal swab/ washing. this retrospective study for additional testing on archived specimens after full anonymisation of specimens was approved by the institutional review boards of each site. commercially available multiplex polymerase chain reaction (pcr)/gel electrophoresis assays (seeplex â rv12 ace detection and seeplex â pneumobacter ace; seegene, south korea [14] [15] [16] ) were used to detect 12 major respiratory viruses and six respiratory bacterial pathogens: influenza viruses a and b, respiratory syncytial viruses (rsvs) a and b, rhinoviruses a/b, coronaviruses oc43/hku1 and 229e/ nl63, adenovirus, parainfluenza viruses 1-3, human metapneumovirus, mycoplasma pneumoniae, chlamydophila pneumoniae, legionella pneumophila serotype 1 and bordetella pertussis. additional in-house real-time [reverse-transcriptase (rt)] pcrs were used to detect the following viruses: rhinoviruses a, b and c, enteroviruses, parainfluenza virus 4, bocavirus, parechoviruses and chlamydophila psittaci, as described elsewhere. [17] [18] [19] basic demographic data (country, sex, age) were collected as well as date of disease onset, date of admission, type of specimen and intensive care unit (icu) admission. for the analysis results of streptococcus pneumoniae and haemophilus influenzae, pcrs were not taken into account, because of the high positivity rates resulting from nasopharyngeal carriage of 25á4% and 31á4%, respectively, in this study (data not shown). the baseline demographic characteristics (age category, sex, country of admission and proportion of icu admissions) were summarised as frequency and proportion. the frequencies of different pathogens were summarised per country, sex and age category to determine the most frequent viruses and bacteria in each stratum. disease outcomes including severity (severe defined as icu admission), death, duration of hospitalisation and duration of illness were examined by regression analysis. the dependence of binary outcomes (severe disease and death) on the presence of pathogens was analysed by logistic regression, adjusted for age and sex. kaplan-meier survival analysis was used to visualise the number of hospitalisation days between each age category (patient deaths were censored). cox regression was used to analyse whether viruses were found more often if the duration of illness before presentation to hospital was shorter. regression analysis was based on the data from vietnam and thailand only as relevant clinical data were missing from indonesia. the statistical software sas version 9.2 (sas institute inc., cary, nc, usa) was used for all computations. a two-sided p value of 0á05 or less was interpreted as statistically significant. a total of 1222 patient samples were collected at participating sites. table 1 summarises the demographic data. the median number of days from illness onset to specimen collection was 4 (iqr: 3-6 days). for 155 patients, no exact admission date was known. the majority of patients in this study were under the age of 5 (776, 63á5%). most patients were enrolled in vietnam (n = 826, 67á6%). the median age was higher in indonesian patients as compared to those from vietnam and thailand (16 year, versus 4 year and 2 year, respectively). one hundred and 67 patients (13á7%) were admitted to the intensive care unit (icu); icu admission data from 232 patients were unknown (indonesia: 223; thailand: 2; vietnam: 7). sixtyseven patients (5á5%) were mechanically ventilated and the overall reported in-hospital mortality was 2á4% (n = 29). detected bacterial pathogens were m. pneumoniae (n = 33, 2á7%), c. psittaci (n = 2), c. pneumoniae (n = 1), b. pertussis (n = 1) and l. pneumophila (n = 2). fluctuations in frequencies were found in particular months for several viruses, with rsv occurring from july to november 2008 and rhinovirus peaking in march 2009 ( figure 2 ). due to differences in age group representation, the number of viral agents detected per patient was lower in indonesia (60/225, 0á27) than in vietnam (800/826, 0á97) and thailand (151/171, 0á88). likewise, the proportion of identified viral agents detected that were rsv was higher in thailand and vietnam than in indonesia (24/151, 15á9%; 117/800, 14á6%, respectively, compared with 3/60, 5%) whilst coronavirus was a greater proportion of identified viral agents in indonesia (7/60, 11á7%) than thailand (3/151, 2á0%) and vietnam (11/800, 1á4%). see table s1 for pathogens detected by country. there were 234 of 1222 patients (19á2%) who had at least two pathogens detected (bacterial and/or viral). of these, 170 patients were found to have two pathogens, 54 patients had three pathogens, nine patients had four pathogens, and one patient had five pathogens. the frequency of each pathogen is shown in table s2 . the pathogens most commonly detected in patients with two or more pathogens were bocavirus (n = 132) and rhinovirus (n = 123). the proportion of patients who had either a virus or a bacterium detected appeared to decrease with age, with the highest proportion of pathogens detected in the 0-4 age group, but increased again in those older than 65. most viral pathogens were more common in children, with parechovirus and parainfluenza virus 4 only found in children (table 2) . m. pneumoniae was found only in patients younger than 45 years, whilst both cases of c. psittacii occurred in patients 45 years and above. the icu admissions for each country are shown in table 1 . the number of patients who were mechanically ventilated was 42 of 826 (5á1%) in vietnam and 21 of 171 (12á3%) in thailand (data not available for indonesia). for the vietnamese data set, where data were most complete, icu admission was related to age; 15 of 616 children (2á4%) aged under 5 years, three of 36 (8á3%) aged 5-14 years, 69 of 106 (65á1%) aged 15-44 years, 33 of 45 (73á3%) aged 45-64 years and 20 of 23 (87%) aged over 65 years were admitted to icu (v 2 for trend p < 0á0005). logistic regression was performed on this data set including the covariates: pathogen type (virus, bacterial, both and neither), age and sex. the analysis showed that only higher age was associated with icu admission (adjusted or for icu admission in those aged >64 years compared to those aged <5 years: 243á79, p < 0á001, table 3 ). no specific pathogen was associated with icu admission. there were only sufficient data available to analyse the duration of hospitalisation for patients in vietnam. the median duration of hospitalisation was 5 days (iqr: 3-7 days). multivariate cox regression analysis for hospitalisation days by pathogen type (virus, bacterial, both and neither), sex and age was performed, and both pathogen type (p = 0á0004) and age category (p = 0á015) had a significant effect. it was found that patients in the >65-year group were hospitalised longer, but did not differ by sex. influenza a virus-positive patients were admitted for a median of 4 days (iqr: 3-8 days) versus 5 days (iqr: 3-8 days) for patients where influenza a was not detected. a similar pattern was seen for coronavirus e229: median of 3 days (iqr: 2-4 days) for positive patients versus 5 days (iqr: 3-8 days) in the negative group. patients who were m. pneumoniae-positive were admitted longer: a median of 9 days (iqr: 7-12 days) versus a median of 5 days (iqr: 3-8 days). there were 29 deaths and a viral pathogen was detected in 12 (41%) of these: influenza virus a (n = 3), rhinovirus (n = 3), bocavirus (n = 1), coronavirus (n = 2), parainfluenza virus (n = 1), rsv (n = 1) and one combination of influenza a with adenovirus and rhinovirus. data concerning the cause of death were not collected in this study; consequently, conclusions about causation cannot be drawn. influenza-like illnesses are a major cause of mortality and morbidity worldwide, especially in younger and elderly age groups. it has been suggested that a large proportion of iliattributable deaths globally occur in africa and sea. 20 data regarding the aetiology of respiratory infections in sea are limited, with some of the available data suggesting that viruses account for a large proportion of these infections. 21 in this 1-year study, we used molecular techniques to detect viruses and atypical bacteria from samples collected from patients hospitalised with ili. the majority of patients enrolled in this study were enrolled in vietnam and were under the age of 5. for ili diagnostics and surveillance, nose and throat swabs are considered the sampling methods of choice, whereas for the diagnostics of h. influenzae and s. pneumoniae, gram-staining and bacterial culture of representative purulent sputum, blood culture or urinary antigen tests (for s. pneumoniae) would be the testing method of choice, which were not done. detection rates of these two pathogens were therefore not included in the analysis. in this study, a potential pathogen was identified in 60á6% of patients. this rate is higher than the previously published rates of between 35% and 47% of organisms identified, in studies looking at the aetiology of respiratory tract infections in different asian countries. 9, 22 a total of 58á6% of the study cohort were found to be positive for viruses compared to just 3á2% of the cohort who had an atypical bacterial pathogen alone detected. more recent studies form asian countries using molecular techniques have shown the rates of virus detection reaching 72%. 10 the most common viruses detected in our study were rhinoviruses, accounting for 32% of the patients in whom viral pathogens were detected and 18á7% of the cohort as a whole, similar to other studies. 1, 23 variation between the three countries in the proportion of patients in whom viral pathogens were detected, and variations in the relative importance of different viruses, largely represents the differing age groups sampled, with the younger patients in vietnam and thailand having higher a proportion of rsv detected. bocaviruses were first described in 2005, and since then, numerous studies have described their detection in the human respiratory tract 3 and their possible association with disease, but causation has not convincingly been proven. 24 in this study, this virus was detected in 16á4% of patients, which is similar to other studies where detection rates ranged from 1á5% to 21á5%. 25, 26 these previous studies as well as this study have detected bocavirus, predominantly from young children rather than from adults. in this study as in other studies, bocaviruses were commonly (136/200, 68% bocavirus isolates in our study, table s2 ) detected in combination with other pathogens, most commonly rhinovirus. atypical bacteria were rarely (3á2%) detected. this rate of detection is lower than previously published data, where detection varied from 13% to 26%. 27, 28 however, as we did not include serological diagnostic methods for atypical bacteria, the true prevalence in our patients may have been higher. the most commonly detected atypical bacteria in this study were m. pneumoniae, followed by c. psittaci and l. pneumophila. the results from our study are similar to the aetiological studies of community-acquired pneumonia in singapore and malaysia, in which m. pneumoniae was the commonest atypical pathogen detected. 29 very little is known about the epidemiology of c. psittaci in sea, and indeed this pathogen has only very recently been described as causing human disease in vietnam. 30 the type of pathogen detected was significantly associated with the duration of hospitalisation. atypical bacteria and in particular infection with m. pneumoniae was associated with longer duration of hospitalisation. the age of the patients was also significantly associated with longer duration of hospitalisation, with the >65 years age group requiring longer admissions. this probably corresponds with increasing comorbidities and decreasing immunity with age, similar to other regions in the world. no clear seasonality was seen with the various pathogens detected, but rates of infection appeared to be higher in the winter and spring months and a study over a single year may miss seasonal trends. rates of influenza virus a increased in indonesia and vietnam in april 2009 coinciding with an epidemic of seasonal h3n2 influenza in vietnam, prior to the first sporadic cases of h1n1pdm09 detected at the end of may 2009. 31 the strengths of this type of study include the large numbers of patients who were enrolled prospectively from three different countries in south-east asia allowing an overview of the viral and atypical bacterial causes of ilis in the region. this study also used rt-pcr techniques; however, an important limitation was the lack of microscopy and sputum cultures for common bacteria associated with ilis such as s. pneumoniae and h. influenzae type b and serology for atypical pathogens. also, limited clinical data were recorded for these cases, making it difficult to assess the clinical significance of the various pathogens detected and the virulence of each pathogen. although swabs from all patients meeting our eligibility criteria were analysed, we have no data concerning the number of patients who may have been admitted with ili but not swabbed. the lack of clinical data also limits comparison between sites as differing admission criteria may impact on the distribution of pathogens. similarly, both the lack of clinical data and the absence of data concerning overall admission limit our ability to interpret the findings relating to icu admission; the finding that adult patients were overrepresented amongst those admitted to icu could represent differential routine sampling of adults and children on such units; however, we feel this is unlikely. there was a lack of data related to patients and co-morbidities, such as immunosuppression and copd, which could also have had an impact in the severity of infection seen. furthermore, a lack of suitable control patients also limits our ability to determine the attributable fraction of ili that is likely to be caused by each virus and also prevents useful analysis of the relevance of the high rates of nasopharyngeal carriage of s. pneumoniae and h. influenzae type b. establishing the aetiology of ili is becoming increasingly important, not only to establish the burden of various pathogens in order to improve management and prevention, but also to guide antiviral treatment and to try to limit the inappropriate use of antimicrobials to reduce the emergence and spread of antibiotic resistance. our study, along with overwhelming evidence from many other studies from these and other countries, suggests that viral causes of ili are much more common than bacterial causes, but that the viral aetiology may vary in space and time. in recent years, the emergence of sars-cov, mers-cov and avian influenza viruses a/h5n1 and a/h7n9 has highlighted the risks posed by respiratory viral infections in humans and sea has become an area of increasing interest for monitoring pathogens. very little has been published from this region regarding the incidence and prevalence of viral or atypical organisms that can contribute to ilis, making studies looking at this increasingly important. additional supporting information may be found in the online version of this article: table s1 . frequency of detected pathogens by country. table s2 . frequency of pathogens found in a combination with another pathogen. prospective study of the incidence, aetiology and outcome of adult lower respiratory tract illness in the community etiology and management of community-acquired pneumonia in asia cloning of a human parvovirus by molecular screening of respiratory tract samples viral infections of the lower respiratory tract: old viruses, new viruses, and the role of diagnosis respiratory infections unique to asia masstag polymerasechain-reaction detection of respiratory pathogens, including a new rhinovirus genotype, that caused influenza-like illness in new york state during viral pathogens associated with acute respiratory infections in central vietnamese children association between nasopharyngeal load of streptococcus pneumoniae, viral coinfection, and radiologically confirmed pneumonia in vietnamese children identification of viral and atypical bacterial pathogens in children hospitalized with acute respiratory infections in hong kong by multiplex pcr assays viral etiologies of acute respiratory infections among hospitalized vietnamese children in ho chi minh city a prospective three-year cohort study of the epidemiology and virology of acute respiratory infections of children in rural india fatal respiratory infections associated with rhinovirus outbreak effect of double dose oseltamivir on clinical and virological outcomes in children and adults admitted to hospital with severe influenza: double blind randomised controlled trial use of the seeplex rv detection kit for surveillance of respiratory viral outbreaks in comparison of the seeplex reverse transcription pcr assay with the r-mix viral culture and immunofluorescence techniques for detection of eight respiratory viruses detection of 12 respiratory viruses with two-set multiplex reverse transcriptase-pcr assay using a dual priming oligonucleotide system rapid detection of human parechoviruses in clinical samples by realtime pcr development of an internally controlled real-time pcr assay for detection of chlamydophila psittaci in the lightcycler 2.0 system development and evaluation of a four-tube real time multiplex pcr assay covering fourteen respiratory viruses, and comparison to its corresponding single target counterparts epidemiology and etiology of childhood pneumonia non-flu aetiology of ili in 3 se asian countries ª 2015 the authors. influenza and other respiratory viruses published by viral etiology of acute lower respiratory tract infections in hospitalized young children in northern taiwan respiratory viral infections detected by multiplex pcr among pediatric patients with lower respiratory tract infections seen at an urban hospital in delhi from aetiological role of viral and bacterial infections in acute adult lower respiratory tract infection (lrti) in primary care human bocavirus-the first 5 years high frequency of human bocavirus 1 dna in infants and adults with lower acute respiratory infection human bocavirus: a novel parvovirus epidemiologically associated with pneumonia requiring hospitalization in thailand atypical bacterial pathogen infection in children with acute bronchiolitis in northeast thailand acute bronchitis in the community: clinical features, infective factors, changes in pulmonary function and bronchial reactivity to histamine an asian study on the prevalence of atypical respiratory pathogens in communityacquired pneumonia first report of human psittacosis in vietnam a clinical virological and epidemiological analysis the investigators were involved in all aspects including protocol design, study execution and oversight and the writing of this publication. richard molenkamp and bob de wever, dept of medical microbiology, academic medical center, amsterdam, the netherlands, are kindly acknowledged for sharing of protocols and positive controls for setting up in-house real-time (rt-) pcrs for respiratory viruses and atypical bacteria. the authors declare that they have no conflict of interests. the study was conducted by the seaicrn (http://www.seaicrn.org/) and sponsored by the national institute of allergy and infectious diseases. site support was also provided by the wellcome trust (london, uk) through its major overseas programmes. h wertheim, md de jong, hr van doorn and p puthavathana contributed to study design, study implementation, analysis and interpretation of the data, writing of the manuscript, and reading and approval of the final version. b nadjm involved in analysis and interpretation of the data, writing of the manuscript, and reading and approval of the final version. s thomas, dung vtv and uyen hn participated in analysis and interpretation of the data and reading and approval of the final version. n agustiningsih and diep ntn contributed to study implementation, analysis and interpretation of the data, and reading and approval of the final version. s malik, kinh vn, chau vvn, liem tn, sinh tt, thuy btp, trung vn, tran th, khanh ht, tuan mh and kulkanya c involved in study implementation and reading and approval of the final version. w taylor and j farrar contributed to study design, study implementation and reading and approval of the final version. m wolbers involved in study design, study implementation, analysis and interpretation of the data, and reading and approval of the final version. key: cord-272878-6f0q661e authors: schnepf, nathalie; resche-rigon, matthieu; chaillon, antoine; scemla, anne; gras, guillaume; semoun, oren; taboulet, pierre; molina, jean-michel; simon, françois; goudeau, alain; legoff, jérôme title: high burden of non-influenza viruses in influenza-like illness in the early weeks of h1n1v epidemic in france date: 2011-08-17 journal: plos one doi: 10.1371/journal.pone.0023514 sha: doc_id: 272878 cord_uid: 6f0q661e background: influenza-like illness (ili) may be caused by a variety of pathogens. clinical observations are of little help to recognise myxovirus infection and implement appropriate prevention measures. the limited use of molecular tools underestimates the role of other common pathogens. objectives: during the early weeks of the 2009–2010 flu pandemic, a clinical and virological survey was conducted in adult and paediatric patients with ili referred to two french university hospitals in paris and tours. aims were to investigate the different pathogens involved in ili and describe the associated symptoms. methods: h1n1v pandemic influenza diagnosis was performed with real time rt-pcr assay. other viral aetiologies were investigated by the molecular multiplex assay respifinder19®. clinical data were collected prospectively by physicians using a standard questionnaire. results: from week 35 to 44, endonasal swabs were collected in 413 patients. overall, 68 samples (16.5%) were positive for h1n1v. in 13 of them, other respiratory pathogens were also detected. among h1n1v negative samples, 213 (61.9%) were positive for various respiratory agents, 190 in single infections and 23 in mixed infections. the most prevalent viruses in h1n1v negative single infections were rhinovirus (62.6%), followed by parainfluenza viruses (24.2%) and adenovirus (5.3%). 70.6% of h1n1v cases were identified in patients under 40 years and none after 65 years. there was no difference between clinical symptoms observed in patients infected with h1n1v or with other pathogens. conclusion: our results highlight the high frequency of non-influenza viruses involved in ili during the pre-epidemic period of a flu alert and the lack of specific clinical signs associated with influenza infections. rapid diagnostic screening of a large panel of respiratory pathogens may be critical to define and survey the epidemic situation and to provide critical information for patient management. in order to monitor the spread of influenza and alert health handlers, several epidemiological tools have been developed. in france, a network of 1300 general practitioners, ''réseau sentinelles'', working throughout the country, provides real-time clinical data used to evaluate regional and national influenza spreading [1, 2] . the criteria used by this network to define clinical influenza-like illness (ili) are the occurrence of a sudden fever above 39uc with myalgia and respiratory signs. in general no formal viral diagnosis is carried out. the groupes régionaux d'observation de la grippe (grog) is a second french network that surveys the emergence and the spread of the influenza viruses [3, 4] . this network is based on clinical surveillance of acute respiratory infections and laboratory analysis of nasal specimens collected from adults and children by volunteer general practitioners and pediatricians. according to the sentinel network's criteria, french health authorities proclaimed that flu epidemic level was reached during the second week of september 2009 (week 37) [5, 6] . on the contrary, data provided by the grog showed only sporadic h1n1v activity until the last week of october (week 44) [6, 7] . thus, it became rapidly obvious that a variety of viruses were circulating in the community and that an overestimation of myxovirus infection was at stake [8, 9, 10, 11] . as a better knowledge of the epidemic status was a key feature for national healthcare organization, hospital preparedness, patient management and disease control, unambiguous viral diagnosis appeared critical. in france, data on viral aetiologies associated with ili were at best sporadic and correlations with clinical symptoms were often lacking. extensive molecular assays to screening for respiratory viruses were not available countrywide for routine diagnosis. therefore the epidemiological pattern of respiratory pathogens with overlapping seasonality was poorly known. the aim of the present study was to investigate respiratory pathogens involved in ili during the early weeks of the 2009-2010 h1n1v diffusion in france (weeks 35 through 44) and describe the associated symptoms in paediatric and adult populations. this study was a non-interventional study with no addition to usual proceedures. biological material and clinical data were obtained only for standard viral diagnostic following physicians' prescriptions (no specific sampling, no modification of the sampling protocol, no supplementary question in the national standardized questionnaire). data analyses were carried out using an anonymized database. according to the french health public law (csp art l 1121-1.1), such protocol does not require approval of an ethics committee and is exempted from informed consent application. in the two academic hospitals, saint-louis hospital (sls) in paris and tours hospital (trs), influenza-like illness (ili) was defined as a patient suffering from at least one general symptom (fever above 38uc, asthenia, myalgia, shivers or headache) and one respiratory symptom (cough, dyspnoea, rhinitis or pharyngitis), in agreement with the guidelines from the french institut de veille sanitaire (invs), a governmental institution responsible for surveillance and alert in all domains of public health [12] . criteria for severe clinical presentation were temperature below 35uc or above 39uc despite antipyretic, cardiac frequency above 120/min, respiratory frequency above 30/min, respiratory distress, systolic arterial pressure below 90 mmhg or altered consciousness. predisposing factors of critical illness were children younger than one year old, pregnant women, diabetes, chronic pre-existing disease (such as respiratory, cardiovascular, neurologic, renal, hepatic or hematologic diseases) and immunosuppression (associated with hiv infection, organ or hematopoietic stem cells transplantation, receipt of chemotherapy or corticosteroids) [13, 14] . a cluster of suspected influenza infections was defined as at least three possible cases in a week in a closed community (household, school,…) [15] . in the two institutions, the prescription of h1n1v molecular testing was recommended for patients with ili and with either a severe clinical presentation, an underlying risk factor of complications or a condition which was not improving under antiviral treatment. investigation of grouped suspected cases was also recommended. from week 35 (last week of august) to 44 (last week of october), 413 endonasal swabs were collected in 3 ml of universal transport medium (copan diagnostics inc, murrieta, ca) from adults and children seen in emergency rooms for suspected ili (table 1 ) and sent to sls and trs laboratories for h1n1v detection. the two microbiology laboratories participated in the reference laboratories network for the detection of pandemic influenza h1n1v. clinical data were collected at the time of medical attention and reported by clinicians on a national standardized questionnaire provided by invs [1, 12] . this questionnaire included the presence or absence of the main general and respiratory symptoms associated with ili (fever, asthenia, myalgia, shivers, headache, cough, rhinitis, pharyngitis, sudden onset) [12] . total nucleic acid was extracted from 400 ml of universal transport medium using the easymag system (biomérieux, marcy l'etoile, france) in sls or the ez1 advanced xl (qiagen, courtaboeuf, france) in trs, according to the manufacturers' instructions (elution volume: 100 ml in sls or 90 ml in trs). before extraction, 5 ml of an internal amplification control (iac) which contained an encephalomyocarditis virus (emc) rna transcript was added into the sample. pandemic h1n1v infection was diagnosed by real-time reverse transcription-pcr (rt-pcr) assay on a 7500 real time pcr system (applied biosystems, foster city, ca) according to the protocol of the centers for disease control (cdc) [16] . other respiratory infections were investigated by a multiplex molecular assay based on the multiplex ligation-dependent probe-amplification (mlpa) technology (respifinder19h, pathofinder, maastricht, the netherlands) that allows the detection and differentiation of 14 respiratory viruses, including influenza virus a (infa), influenza virus b (infb), rhinovirus (rhv), parainfluenza viruses 1 to 4 (piv-1 to piv-4), human metapneumovirus (hmpv), adenovirus (adv), respiratory syncytial virus a (rsva), respiratory syncytial virus b (rsvb) and human coronaviruses 229e, oc43 and nl63 (cor-229e, cor-oc43, cor-nl63) [17] . the test allows also the detection of h5n1 influenza a virus and of four bacteria: chlamydophila pneumoniae (cp), mycoplasma pneumoniae (mp), legionella pneumophila (lp) and bordetella pertussis (bp). the amplified mlpa products were analyzed on an abi 3100 genetic analyzer (applied biosystems, foster city, ca). fragment sizing analysis was performed with the genemarker software (softgenetics, llc, state college, pa). further testing for h1n1v was carried out with simplexa tm influenza a h1n1 (2009) (focus diagnostics, cypress, california) when the cdc real time rt-pcr assay was negative for h1n1 and the respifinder19h assay was positive for influenza a. if this latter assay was negative, h3n2 typing was performed as previously described [18] . data from our study are summarized as frequencies and percentages for categorical variables. quantitative variables are presented as medians, 25th and 75th percentiles. to compare those variables according to the viral infection status, fisher tests by using cdc reference assay, h1n1v was detected in 66 samples out of 413 (16.6%), more frequently in sls (38 samples) than in trs (28 samples) (p,10 24 ). overall, weekly percentage of h1n1v positive endonasal swabs remained under 10% until week 41 and increase significantly after (p trend ,0.0001) ( figure 1 ). rate of h1n1v detection reached 30% in sls at week 42 and in trs at week 44. overall, this rate was in agreement with results provided by the grog network, showing an earlier start of h1n1v epidemic in paris area [7, 19] . all 413 nucleic acid extracts were analyzed using the respifinder19h assay ( figure 2 ). sixty six patients tested h1n1v positive with cdc real time rt-pcr assay were confirmed with the multiplex assay. thirteen were also co-infected by one or two other respiratory pathogens (multiple infections) ( figure 2 ). three of the 347 h1n1v negative samples could not be studied with the multiplex assay because they contained rt-pcr inhibitors (no amplification of the internal control). two hundred and fifteen (62.5%) of the remaining 344 h1n1v negative samples were found positive for at least one respiratory pathogen ( figure 2 ). two hundred and twelve were positive for non influenza pathogens (189 single infections and 23 mixed infections with two, three or four viruses) and three additional single infections by influenza a were identified in sls, including two by pandemic h1n1v and one by seasonal h3n2, as determined after molecular typing (data not shown). overall, 68 patients (16.5%) were then positive for h1n1v, one for h3n2 and 212 for non influenza pathogens. there were 245 single infections (55 with h1n1v and 190 with other respiratory pathogens) and 36 mixed infections (13 with h1n1v and 23 without h1n1v) ( figure 2 ). among h1n1v negative single infections, the most prevalent viruses were rhinovirus (62.6%, 119 patients), followed by parainfluenza viruses 1 to 4 (24.2%, 46 patients), adenovirus (5.3%, 10 patients), human coronavirus 229e, oc43 and nl63 (3.2%, 6 patients) and respiratory syncytial virus a and b (2.6%, 5 patients) (figure 2 ). in addition, respifinder19h assay identified three patients with bacterial infection, two with mycoplasma pneumoniae (one 25 years old female in sls and one 39 years old female in trs) and one with bordetella pertussis (one 60 years old male in sls). no single infection by influenza b, hmpv, chlamydophila pneumoniae or legionella pneumophila was identified ( figure 2 to analyze if viral co-infections occurred more frequently for some viruses, we carried out a two by two comparisons, that showed a higher proportion of co-infection only for adv (p = 0.05). non-influenza respiratory viruses presented a different epidemic profile compared to h1n1v. overall, in both hospitals, weekly rate of non-h1n1v respiratory viruses whether alone or involved in co-infection increased between week 37 and 39 (from 51.4% to 81.3%) and then consistently decreased ( figure 3 ). rhv infections that represented nearly half of non-h1n1v viral infections (141 out of 213, 66.2%) were a significant contributing factor. in both hospitals, emergence of h1n1v cases was associated with a rapid decline of rhv rate of infection from 50-60% down to less than 20% with a one to two weeks gap between sls and trs. data on age ( in both institutions, 85.5% (106/124) children younger than 15 years of age were infected by at least one respiratory pathogen ( table 2 ). h1n1v infected patients were not significantly younger than h1n1v non infected patients (27 years old vs. 25 years old, p = 0.80) (figure 4) . however, 70.6% (48/68) of h1n1v cases were identified in patients under 40 years old (22 in sls and 26 in trs) and no case was observed in patients older than 65 years ( table 2) . piv infection occurred in very young patients (median (figure 4) . consequently, piv and adv were more frequently detected in the younger population of trs versus sls (p,10 24 and p,10 23 respectively). in contrast, although individuals with rhv infection were slightly younger than individuals without (median age = 24 vs. 29 for patients without rhv, p = 0.05) (figure 4) , influenza-like illness associated with rhv was more frequent in sls than in trs (p = 0.012). finally, patients with viral multiple infection were significantly younger than those with single infection (median, idr: 4, 2-18.5 vs. 25, 6-43) and rates of mixed infection at the time of medical attention, 383 (92.7%) standardized clinical questionnaires were collected out of 413 patients. four of them could not be exploited because they were too incomplete. a review of the 379 workable questionnaires showed that 90.8% (344/379) of the patients included in this study fulfilled the criteria of ili as defined above, and 52.5% had either a severe clinical presentation or an underlying risk factor of complications (45.9%, 174/379), or were in a suspected cluster of grouped cases (6.6%, 25/379). overall, most patients have fever (93.9%) and cough (86.1%) ( table 3) . other classical clinical signs associated with ili such as asthenia, myalgia, shivers, headache, rhinitis or pharyngitis were less frequent. a sudden onset was also described in 59.2% of cases. only 32.5% of the patients had a temperature above 39uc; the age of these patients ranged from zero to 86 years, with a median age of 32 years and a mean age of 34 years (data not shown). in h1n1v infected patients (including single and multiple infections), the main symptoms were also fever (98.2%) and cough (89.5%) ( we then compared clinical characteristics between patients positive for h1n1v, patients positive for other respiratory pathogens and negative for h1n1v and patients without any detection of respiratory pathogens (as detected with respifin-der19h) ( table 3 ). there was no difference between the three groups except for fever, cough, pharyngitis. however for these latter symptoms, the comparison between patients positive for h1n1v and those positive for other respiratory pathogens or between patients positive for h1n1v and those without any detection of respiratory pathogens, showed no difference except for pharyngitis, which was less frequent in patients positive for h1n1v than in patients positive for other respiratory pathogens ( table 3) . as rhv was the most frequent aetiology in ili, we also compared clinical symptoms observed in patients with a single infection by rhv or by h1n1v (data not shown). there was no difference except that rhinitis and pharyngitis were significantly more frequent in rhv infection (62.7% vs. 34.1% [p = 0.006] and 39.0% vs. 10.0% [p = 0.001], respectively). viral multiple infection (including samples with h1n1v) was not associated with a different clinical presentation. fever and cough were observed in over 90% of the patients (90.6% and 90.3%, respectively), but only 33.3% of these patients had a temperature above 39uc, which was not different from patients with single viral infection (28.6%). our results highlight the high frequency of non-influenza viruses involved in acute respiratory infections during the epidemic period of a flu alert as defined by the réseau sentinelles according to ili definition (a sudden fever above 39uc accompanied by myalgia and respiratory signs). these data extent previous observations in europe reporting high prevalence of rhv infections before seasonal influenza [4, 20] or in 2009, before h1n1v pandemic influenza [1, 8, 9, 11, 21] . we confirm that rhv represent the most frequent aetiology of acute respiratory table 2 . age of patients with respiratory samples positive for h1n1v, positive for other respiratory pathogens or negative. infections both in adult and paediatric populations and may represent more than 50% of cases. we show that other viral infections than influenza and rhv may represent up to 30% of aetiologies. we observed differences between the two hospitals, with a higher frequency of parainfluenza and adv infections in tours in contrast with a higher frequency of rhv in paris, likely explained by the higher proportion of paediatric samples collected in tours. however, despite the distance between the two institutions (about 250 km) and differences between the two populations, both presented similar patterns of high frequency of non-influenza viruses in acute respiratory infections before the flu epidemic wave and a decline when influenza reached epidemic levels. in the two cities, high frequencies of rhv were seen at the same level with a likely different evolution speed, with sudden increase and decrease in sls and more progressive variation in trs. in both institutions, there was a decrease in the proportion and number of rhv diagnoses roughly in parallel with the increase of influenza diagnoses. indeed, h1n1v exceeds 20% of positive detection's rate only when rhv dropped under 40%. these data are thus consistent with negative interaction of the two epidemics at the population level. it was previously hypothesised that rhv epidemic could interfere with the spread of pandemic influenza [20, 21, 22] . few in vitro data support this hypothesis. it has been reported that interferon and other cytokines production by rhv infected cells induced a refractory state to virus infection these data include the three patients whose respiratory samples could not be studied with the multiplex assay because of rt-pcr inhibitors. of neighbouring cells [23] . further work is needed to confirm in vitro and in vivo such negative interactions and if viral interference are really translated to a population level. analysis of rhinovirus and influenza epidemics in previous years should also help to determine if similar interferences were observed with seasonal influenza and to elaborate modelling and prediction of the spread of influenza according to respiratory viruses' circulation. systematic extensive screening of respiratory viruses at a national level should be implemented for this purpose. very few rsv infections were observed in contrast to usual epidemiology which was characterized the last four past years by a start of epidemics in weeks 44-45 [1] . it has been confirmed by other laboratories and the french invs that the 2009-10 rsv epidemic was delayed and had a lower impact compared with the previous winter season [1, 24] . delayed and reduced rsv spread may be due to viral interference between rsv and influenza. another possible explanation is better prevention behaviour about respiratory infections as recommended by a national campaign including recommendations for hands washing after sneezing and the use of mask [1] . influenza infections were mainly detected in patient under 40 years old and no case was found in patients older than 65. these results corroborate previous data suggesting that past seasonal h1n1 infections or vaccination may give partial crossed protection [10, 13, 25] . we have previously shown that the neutralizing titers against pandemic h1n1v virus correlate significantly with neutralizing titers against a seasonal h1n1 virus, and that the h1n1v pandemic influenza virus neutralizing titer was significantly higher in subjects who had recently been inoculated by a seasonal trivalent influenza vaccine [26] . viral co-infections were predominantly seen in paediatric patients, as previously described [4, 27, 28, 29] , both in influenza and non-influenza cases at a similar rate. no evidence of more pronounced respiratory impact was seen in these patients. our results showed the lack of specific clinical signs associated with proven h1n1v infections. clinical characteristics did not differ between influenza infections or other viral infections. in particular, the proportion of patients with fever above 39uc was not higher in h1n1v positive patients. in addition, the patients without any evidence of respiratory viral infections did not have different symptoms. these patients may have been infected with other virus not included in the multiplex assay (human bocavirus, coronavirus hku1) [9, 10, 11] or were seen too late at the time of viral shedding was cleared [30] . however, to determine how specific the symptoms are for influenza would require to assess also the distribution of respiratory pathogens (h1n1v and other respiratory viruses) and related symptoms in patients presented at the emergency departments in sls and trs with respiratory syndromes, but not tested for h1n1v. in addition, despite some underlying conditions that were associated with complications not previously observed in seasonal influenza, most illnesses caused by the h1n1v virus were acute and self-limited [13, 31] . the higher proportion of non influenza viruses reported in ili in 2009 was thus most likely a consequence of more frequent visits to a doctor for respiratory tract infections than usually observed for fear of the flu pandemic. the general lack of difference in symptoms in the particular context of h1n1v pandemic has therefore to be considered with caution and does not rule out that more significant differences may arise in future influenza epidemics with other influenza viruses. our data confirm that it may be virtually impossible to recognize symptoms heralding h1n1v infections and virological data should be helpful along with clinical reports to monitor influenza epidemic [10] . molecular multiplex detection has recently emerged as a potent diagnostic tool to determine acute respiratory infections' aetiologies [11, 32, 33] . these data show that sensitive molecular multiplex detection of respiratory viruses is feasible and efficient for the detection of virus involved in acute respiratory infections and provides insights into their epidemic profile. our results confirm the performance of respifinder19h assay to detecting respiratory viruses in the general population as recently shown in transplant patients with ili [34] . respifinder19h confirmed all h1n1 infections detected by the cdc reference assay and was able to identify two additional h1n1 cases suggesting a high sensitivity of this multiplex assay to detect influenza a infections. in conclusion, our results highlight that successive and mixed outbreaks of respiratory viral infections may affect influenza epidemiology and can lead to misinterpret the early development of a flu epidemic. rapid diagnostic screening of a large panel of respiratory pathogens may be critical to define and survey the epidemic situation and to provide critical information for patient management. impact of the 2009 influenza a(h1n1) pandemic wave on the pattern of hibernal respiratory virus epidemics virtual surveillance of communicable diseases: a 20-year experience in france a new influenza surveillance system in france: the ile-de-france ''grog''. 1. principles and methodology surveillance of community-acquired viral infections due to respiratory viruses in rhone-alpes (france) during winter 1994 to 1995 bulletins hebdomadaires surveillance de la grippe en france. bulletins hebdomadaires a variety of respiratory viruses found in symptomatic travellers returning from countries with ongoing spread of the new influenza a(h1n1)v virus strain frequency of detection of upper respiratory tract viruses in patients tested for pandemic h1n1/09 viral infection novel virus influenza a (h1n1sw) in south-eastern france rapid detection of respiratory tract viral infections and coinfections in patients with influenza-like illnesses by use of reverse transcription-pcr dna microarray systems surveillance de la grippe en france clinical aspects of pandemic 2009 influenza a (h1n1) virus infection severe hospitalised 2009 pandemic influenza a(h1n1) cases in france modified surveillance of influenza a(h1n1)v virus infections in france cdc protocol of real-time rt-pcr for influenza a (h1n1) respifinder: a new multiparameter test to differentially identify fifteen respiratory viruses application of a fluorogenic pcr assay for typing and subtyping of influenza viruses in respiratory samples bulletins épidémiologiques de la grippe a (h1n1) rhinoviruses, a(h1n1)v, rvs: the race for hivernal pandemics rhinoviruses delayed the circulation of the pandemic influenza a (h1n1) 2009 virus in france does viral interference affect spread of influenza? the interferon response circuit: induction and suppression by pathogenic viruses situation épidémiologique de la bronchiolite portrait of a year-old pandemic detection of extensive cross-neutralization between pandemic and seasonal a/ h1n1 influenza viruses using a pseudotype neutralization assay multiplex real-time pcr for detection of respiratory tract infections frequent detection of viral coinfection in children hospitalized with acute respiratory tract infection using a real-time polymerase chain reaction detection of multiple respiratory pathogens during primary respiratory infection: nasal swab versus nasopharyngeal aspirate using real-time polymerase chain reaction prospective evaluation of a novel multiplex real-time pcr assay for detection of fifteen respiratory pathogens-duration of symptoms significantly affects detection rate improving the clinical diagnosis of influenza-a comparative analysis of new influenza a (h1n1) cases molecular diagnosis of respiratory viruses detection of respiratory viruses by molecular methods comprehensive diagnostics for respiratory virus infections after transplantation or after potential exposure to swine flu a/h1n1: what else is out there? we gratefully acknowledge the contribution of the members of the two virology laboratories and sandrine picco for their excellent technical assistance in the detection of h1n1v pandemic virus and other respiratory viruses, and catherine scieux for her help in epidemiological data analysis. key: cord-288372-48wao8a0 authors: dia, ndongo; richard, vincent; kiori, davy; cisse, el hadj abdoul khadir; sarr, fatoumata diène; faye, abdourahmane; goudiaby, déborah g; diop, ousmane m; niang, mbayame n title: respiratory viruses associated with patients older than 50 years presenting with ili in senegal, 2009 to 2011 date: 2014-04-08 journal: bmc infect dis doi: 10.1186/1471-2334-14-189 sha: doc_id: 288372 cord_uid: 48wao8a0 background: in africa, especially in west africa, studies about the prevalence and diversity of respiratory viruses (influenza and others) in elderly people are largely lacking. in studies done elsewhere, it is well established that older people, when compared with younger adults, are at greater risk of significant morbidity and mortality from complications arising from influenza. the main aim of this study was to determine the prevalence and the diversity of respiratory viruses associated with ili cases in adults over 50 years old in senegal. methods: the recruitment period of this study was from january 2009 to december 2011. 232 patients aged 50 years and above presenting ili cases were enrolled. nasal-pharyngeal and/or oral pharyngeal swabs were collected from patients. rna was extracted from 200 μl of each sample followed by a two-step real-time rt-pcr. the anyplex™ ii rv16 detection kit was used for viral detection. the kit enabled the simultaneous detection of the presence of 16 respiratory viruses. results: 150 viruses were detected: influenza viruses (44.7%) and rhinoviruses (26.7%) were the most prevalent. we detected 13 human parainfluenza viruses (8.7%), 7 human respiratory syncytial viruses (4.7%), 6 coronaviruses (4%), 5 human metapneumoviruses (3.3%), 5 human adenoviruses (3.3%) and 1 human bocavirus (0.7%). 14 cases (6%) of dual virus infections and one triple viral detection case were encountered. 56 (56.6%) viruses detected were found in the 50-64 year old age group, 59 (76.6%; p < 0.001) from 65–74 year old age group and 35 (62.5%) were detected in the ≥75 year old age group. the viral co-infections were more frequent in the 65-74 age group (9/15). conclusions: this pilot study demonstrates a variety of respiratory viruses in the elderly. it also highlights a high prevalence of these viruses in this age group. we speculate from these results that the impact of respiratory viruses other than influenza on the elderly has been considerably underestimated. a more exhaustive study seems necessary in order to provide a more complete picture of the burden of respiratory viruses on morbidity among adults over 50 years old in the sub-saharan context. viral aetiology, prevalence and diversity data in people with influenza like illness (ili) and/or acute respiratory illness (ari) in africa, (especially in west africa), are scarce and often limited to the influenza viruses' infection. following the last influenza pandemic episode [1] , few global and pediatric studies were conducted in some countries of the sub-region [2] [3] [4] , and only a limited number of studies have described the etiology of ili due to viruses including non-influenza respiratory virus [5] [6] [7] . however, no study has been conducted to describe the prevalence and the diversity of respiratory viruses (influenza and others) in west african elderly people. in studies done elsewhere, it is well established that older people, when compared with younger adults, are at greater risk of significant morbidity and mortality from complications arising from influenza [8, 9] . for example in the united states alone, up to 40% of non-pneumonic lower respiratory illnesses in the elderly have been associated with respiratory viral infection [10] , and an estimated 54,000 deaths annually have been attributed to the influenza and respiratory syncytial viruses (rsv) [11] . it should be highlighted that in senegal the number of elderly people in consultation in healthcare centers for influenza like illness (ili) is very low. indeed, routine influenza monitoring in senegal showed that samples from people above 50 years old represent only 3.7% of the total, over a 16 year surveillance period [4] . some practices such as auto-medication and the use of traditional medicine to treat ili largely explain this situation with the socio-economic situation being another contributing factor. thus the main aim of this study was to determine the prevalence and the diversity of respiratory viruses associated with ili cases in adults over 50 years old. the recruitment period of this prospective observational study was from january 2009 to december 2011 inclusive. all 232 patients aged 50 years and above presenting with ili during this period were enrolled in the study. it should be noted that samples were collected in the context of flu monitoring. an influenza sentinel surveillance system for outpatients with ili was established in 1996 in senegal and became part of the who global influenza surveillance and response system (gisrs). it is coordinated locally by the national influenza center (nic) at the institut pasteur de dakar. trained medical personnel were asked to screen all outpatients who were attended at the sentinel sites for signs and symptoms of ili. the symptoms of influenza are similar to those arising from other viral respiratory pathogens. the inclusion criteria, according to the cdc case definition, were sudden onset of fever (≥38°c) with cough or sore throat fewer than 3 days in duration. nasal-pharyngeal and/or oral-pharyngeal swabs were collected from each enrolled ili case, placed in cryovials containing 3 ml of viral transport medium (universal transport medium, copan diagnostics inc., murrieta, ca, usa) and stored at 4°c on site. if nasal-pharyngeal and oral-pharyngeal swab specimens were collected from the same patient, both swabs were placed in the same cryovial. upon arrival at the laboratory the specimens were separated into 3 aliquots for analyses. the first aliquot was used for molecular analysis for the detection of influenza viruses (real-time reverse transcription polymerase chain reaction or rrt-pcr detection), the second was used for influenza virus isolation, and the third was stored at −80°c for further analysis. the latter was used in the present study. for each patient who met the case definition criteria, a form collecting demographic and clinical data was completed.the questions included information on date of enrollment and symptom onset, sex, age, clinical symptoms, previous treatments, vaccination status for influenza, and whether or not the patient was hospitalized. ribonucleic acid (rna) extraction was performed from 200 μl of each sample using the qiaamp viral rna kit (qiagen, valencia, ca, usa) according to the manufacturer's instructions. each rna sample was eluted with 100 μl nuclease-free water before rna quantification with a nanodrop apparatus (nanodrop lite, thermo scientific). a two-step real-time rt-pcr was performed using the cfx96 real-time pcr detection system (bio-rad). the revertaid first strand cdna synthesis kit (thermo scientific) was used. first 1 ng of rna was mixed with 1 μl of random hexamer primer and nuclease free water for a final volume of 10 μl. it was then incubated at 65°c for 5 minutes and immediately put on ice in order to remove the secondary structures in gc-rich rna. for the cdna synthesis step, 4 μl of 5x reaction buffer, 1 μl of rnase inhibitor (20 u/μl), 2 μl of dntp mix (10 mm) and 1 μl of revertaid m-mulv reverse transcriptase (200 u/μl) were added and incubated for 5 minutes at 25°c followed by 60 minutes at 42°c and 70°c for 5 minutes. the cdna product could be used directly for the next step (pcr amplification) or stored at −80°c until use. for viral detection, the anyplex™ ii rv16 detection kit (seegene) was used. the kit enabled simultaneous detection of influenza a virus, influenza b virus, human respiratory syncytial virus a, human respiratory syncytial virus b, human adenovirus, human metapneumovirus, human coronavirus 229e, human coronavirus nl63, human coronavirus oc43, human parainfluenza virus −1, −2, −3, −4, human rhinovirus a/b/c, human enterovirus and human bocavirus. reactions are duplicated in two panels (a and b) for detection of the 16 viruses. the total reaction volume was 20 μl for each sample (for each panel), containing 4 μl 5x rv16 a (or 5x rv16 b), 4 μl of 8-mop solution, 4 μl of 5x anyplex pcr master mix (mix well by inverting 5 times) and 8 μl of cdna product. pcr was assessed after 95°c for 15 minutes for transcriptase reverse enzyme inactivation, 50 cycles of 95°c for 30 seconds, 60°c for 60 seconds and 72°c for 30 seconds. 1 additional cycle of 55°c for 30 seconds was added for completion. the fluorescence is detected with a melting curve step, 55°c-85°c (5 seconds/0.5°c). fisher's exact test was used to verify whether the associated proportions were statistically supported and a p-value < 0.05 was considered statistically significant. we used the 50-64 year' old group as the reference. the r.15.1 tool was used to perform the analyses. the senegalese national ethical committee of the ministry of health approved the surveillance protocol as less than minimal risk research, and written consent forms were not required. throughout the study, the database was shared with the epidemiology department at the senegalese ministry of health and prevention for appropriate public health action. a total of 232 patients above 50 years old were enrolled into the study, 129 (55.8%) were women and 102 (44.2%) were men (table 1 ). patients' ages ranged from 51 to 97 years, with a mean age of 66 years. ninety-nine (42.7%) enrolled patients were between 51 and 64 years old, 77 (33.2%) between 65 and 74 years old and 56 (24.1%) were 75 years old or older. fever was the most reported clinical symptom, in 213 (92.2%; 213/232) of the enrolled patients, followed by cough (80.2%; 186/232), rhinitis (75.9%; 176/232), myalgia (53.9%; 125/232) and pharyngitis (44%; 102/232). in all, 132 (56.9%) out of the 232 patients were found to be infected with at least one of the viruses of interest. a total of 150 (64.6% of patients) viruses were detected. of these viruses, influenza viruses (44.7%; 67/150) and rhinoviruses (26.7%; 40/150) were the most prevalent viruses detected (table 2) . we detected 13 human parainfluenza viruses (3 piv1, 4 piv3, 6 piv4) (8.7%), 7 human respiratory syncytial viruses (6 rsv a and 1 rsv b) (4.7%), 6 coronaviruses (4 coronaviruses nl63, one coronaviruses 229e and one coronavirus oc43) (4%), 5 human metapneumoviruses (3.3%), 5 human adenoviruses (3.3%) and one human bocavirus (0.7%). a total of 14 cases (6%) of dual virus infections and one triple viral detection case were encountered. influenza viruses (13 cases) and rhinoviruses (8 cases) were the most common type of virus found in samples with coinfections. regarding the number of viruses detected per age group, 56 (56.6%; 56/99) were from the 50-64 age group, 59 (76.6%; 59/77; p < 0.001) from the 65-74 year old age group and 35 (62.5%; 35/56) were detected in the older than 75 year old age group ( table 2) . the viral coinfections are more frequent in the 65-74 year old age group (9/15) followed by the ≥75 years group (4/15). taking into account the clinical symptoms and viral detection, cough, rhinitis, pharyngitis or headache were in similar proportion in viral positive and non-positive patients: cough was observed in 79.4% of positive patients and 77% in the negative patients group (p = 0.17), rhinitis 79.4% and 67% (p = 0.04), pharyngitis in 45.8% and 42% (p = 0.74) respectively. in contrast myalgia symptoms are significantly higher among viral-positive patients: 76.3% versus 25% (p < 0.001). the pattern of the virus detection throughout the study period is showed in the figure 1 . influenza viruses (a and b) were mostly detected from july to august (between weeks 28 and 43), which corresponds to the rainy season in senegal. a minor detection peak is also registered at the beginning of the year. rhinoviruses and parainfluenza viruses showed homogeneous detection levels throughout the study period. grouped, the remaining viruses seemed to have a similar temporal pattern to that of influenza viruses. the gap observed between weeks 18 and 22 correspond with a lack of samples from patients from our targeted age group. the present study is the first description of the etiology of respiratory viruses associated with patients with ili in a cohort of elderly people in the west african context. the results obtained showed that 132 samples of the study population out of 232 contained at least one of the targeted respiratory viruses. the frequency of virus detection (56.9%) among the elderly with ili in our study is consistent with that of several studies already conducted. huo et al. (2012) [12] , in a similar study in china detected at least one respiratory virus in 53% of patients 60 years old or older, and munoz et al. (2000) [13] 49% in elders in a long term care facilities in ontario during the 1998-1999 period. hasman et al. (2009) [14] detected at least one respiratory virus in 68% patients in a study conducted in usa, without any precision about the ages of the 154 adults. in others studies frequencies are lower. for example nicholson et al. (1997) [15] 43% (211/497) of viral detection among the elderly between 60 and 90 years of old, 36% (185/512) detected in the elderly over 65 years old in a study in china [16] and 22.2% in a recent study in japan [17] . it is important to note that the technical approach used explains some discrepancies in rates of detection: primarily in their sensitivity and secondly in the number of targeted viruses. alternatively differences in rates of detection could be due to true geographical differences in overall burden, differences in study populations (outpatients or hospitalized patients) and to the studies sample collection periods. overall, the viral detection rate in the present study is very high as elderly people are often protected by pre-existing antibodies from previous illnesses, maybe illnesses suffered even decades back [18] [19] [20] . indeed, because of pre-existing systemic and mucosal antibodies, elderly adults have been observed to have lower amounts of respiratory secretions and lower viral loads compared to children [10] . consistent [21] noted that the age distribution differed significantly between positive and negative patients, with positive patients being younger than negative patients (or = 0.98, ic 0.95-0.99; p = 0.0226). of the 150 viruses detected in the elderly, influenza a virus was the most common viral pathogen. combined with influenza b viruses, influenza viruses represented 45% (67/150) of viruses detected. this influenza detection rate was expected as the enrollment of patients was directed towards patients with ili. these results are in agreement with previous findings in the elderly [14, 17, 22] . our study revealed a high detection rate of rhinoviruses (40/150; 26.7%). rhinovirus is the most common respiratory pathogen in all age groups [23] . in a previous study, nicholson et al. (1997) [15] showed that rhinoviruses were responsible for a greater disease burden (activities restriction, duration of illness) than that of influenza in elderly subjects representing 52% of detected viruses. in another study published by greenberg (2002) [24] , rhinovirus was the most prevalent pathogen (121 isolates; 53%) of the 231 identified in upper respiratory episodes. these findings are in concordance with the high rhinovirus detection rate in the present study. with lower prevalence, piv, rsv, hcov, hmpv, enteroviruses, adenoviruses and bocaviruses were identified from elderly patient' specimens and contributed collectively to 28.7% of all ili cases in our study. these results show the high diversity of respiratory viruses circulating in the elderly population. this viral diversity supports previous results [10, 12, 24] and often in similar distributions with those of the present study. co-infections were relatively common in this study especially in the 65-74 years old age group (11.7%; 9/77). the rate found in this age group was in line with the findings of hasman et al. (2009) [14] (11%) and huo et al. (2012) [12] , 11.7%. huo and colleagues, in agreement with our results noted that co-infections were found most commonly in adults older than 60 years of age. focusing on clinical symptoms, with the exception of myalgia, our study showed no significant differences between viral-positive and viral-negative patients with ili. viral circulation observed during the study period showed different patterns depending on the viral types. if we consider influenza viruses, we observed a circulation peak during the period starting in week 35 and ending in week 44. this period corresponds to the middle of the rainy season in senegal. this result is further supported by a recent study conducted by mbayame and colleagues [4] . these authors established clearly the seasonality of influenza viruses in senegal after many years of surveillance with a regular circulation during the year and a peak in the middle of the rainy season (july-august-september). the slight peak of influenza observed at the beginning of the year (february) is the result of the shift caused by the recent pandemic episode. the pandemic occurred in early 2010 in senegal with a peak in february [25] . rhinoviruses showed a regular yearly circulation with peaks along the year corresponding to any rain season influence. the remaining respiratory viruses (piv, rsv, hcov, hmpv, enterovirus, adenovirus and bocavirus) were more likely associated with ili peak during the rainy season. this co-circulation with influenza viruses was also seen in a previous pediatric study in senegal [6] . further studies (multiple year surveillance) are needed in order to properly define the temporal patterns of non-influenza virus circulation in senegal. our study did have several limitations. the first weakness is the small number of samples treated in this study. a more exhaustive sampling would give a better representation of the different targeted viruses in the ili cases among the elderly population in senegal. unfortunately after 16 years of influenza sentinel monitoring we noted that the number of elderly presenting at healthcare centers for ili consultation is rather low compared to other age groups (children and young adults). the absence of nursing home services as in industrial countries, the use of traditional medicine (especially among the elderly) and economic constraints do not facilitate such studies in the west african context. it is worth noting that this was a retrospective study, the database contained limited information on disease outcome and atypical clinical symptoms in ili patients which were not reported. thus the association between viral infections (or co-infections) and severe signs could not be established. as in previous studies it appears that co-infections were associated with more severe signs than mono-infections [26, 27] . without such data we could not measure the burden of targeted respiratory viruses in older patients with ili. another limitation is that our study is only focused on outpatient' cases; it would be interesting to investigate hospitalized patient cases (severe cases). a final limitation was that the study included mainly one geographic location, dakar, the capital city of senegal. despite the small number of samples included, the present pilot study demonstrates a variety of respiratory viruses in the elderly. it also highlights a high prevalence of these viruses in this cohort. from these results, it appears that the impact of respiratory viruses other than influenza was considerably underestimated. a more exhaustive study (increasing the number of elderly patients, with a better clinical picture and better documentation including disease outcomes, illness duration, hospitalizations etc.), relying on the new sentinel surveillance system (extension of sentinel sites in others geographical areas), seems necessary in order to provide a more complete picture of the burden of respiratory viruses on morbidity among adults over 50 years old in the sub-saharan context. influenza-like illness; rrt-pcr: real-time reverse transcription polymerase chain reaction; who: world health organization; cdna: complementary deoxyribonucleic acid; rna: ribonuleic acid; gisn: global influenza surveillance network; nic: national influenza center rsv: respiratory syncytial virus; hcov: human coronavirus; hmpv: human metapneumovirus emergence of a novel swine-origin influenza a (h1n1) virus in humans virological surveillance of influenza-like illness among children in ghana influenza viruses in nigeria, 2009-2010: results from the first 17 months of a national influenza sentinel surveillance system sentinel surveillance for influenza in senegal viral etiology of influenza-like illnesses in antananarivo viral etiology of respiratory infections in children under 5 years old living in tropical rural areas of senegal: the evira project viral etiology of influenza-like illnesses in cameroon the impact of influenza on the health and health care utilisation of elderly people influenza-related hospitalisation and death in australians aged 50 years and older new respiratory viruses and the elderly mortality associated with influenza and respiratory syncytial virus in the united states surveillance of 16 respiratory viruses in patients with influenza-like illness in nanjing current research on influenza and other respiratory viruses: ii international symposium aetiology of influenza-like illness in adults includes parainfluenzavirus type 4 acute viral infections of upper respiratory tract in elderly people living in the community: comparative, prospective, population based study of disease burden influenza a/h1n1 2009 pandemic and respiratory virus infections the post-infection outcomes of influenza and acute respiratory infection in patients above 50 years of age in japan: an observational study characterization of an avian influenza a (h5n1) virus isolated from a child with a fatal respiratory illness epidemiology of 2009 pandemic influenza a (h1n1) deaths in the united states outbreaks of 2009 pandemic influenza a (h1n1) among long-term-care facility residents-three states nationwide surveillance of 18 respiratory viruses in patients with influenza-like illnesses: a pilot feasibility study in the french sentinel network rates of hospitalizations for respiratory syncytial virus, human metapneumovirus, and influenza virus in older adults fields virology viral respiratory infections in elderly patients and patients with chronic obstructive pulmonary disease a subregional analysis of epidemiologic and genetic characteristics of influenza a(h1n1)pdm09 in africa respiratory viruses in children admitted to hospital intensive care units: evaluating the clart1 pneumovir dna array comparison of human metapneumovirus, respiratory syncytial virus and influenza a virus lower respiratory tract infections in hospitalized young children respiratory viruses associated with patients older than 50 years presenting with ili in senegal the authors thank dr abdou salam gueye, associate director for science at the us centers for disease control (cdc) in ivory coast, for his great help in the revisions of this paper, especially for the english language. many thanks to oumy niass (phd student, immunology unit, ipd) for her help in the statistical analyses. this work was funded by institut pasteur de dakar, senegal. the authors declare that they have no competing interests. the work presented here was carried out in collaboration between all authors. mnn and omd, defined the research and revised the manuscript; nd performed and coordinated technical work, wrote the draft and revisions of the paper; dk performed the main technical part of this work; eakc and af participated in the technical work; dg and eakc participated in data management and analysis; vr revised the manuscript and participated in the monitoring of the surveillance sites; fds participated in the monitoring of the surveillance sites. all authors have contributed to, seen and approved the manuscript. key: cord-002451-r7a0orh7 authors: chu, yanhui; wu, zhenyu; ji, jiayi; sun, jingyi; sun, xiaoyu; qin, guoyou; qin, jingning; xiao, zheng; ren, jian; qin, di; zheng, xueying; wang, xi-ling title: effects of school breaks on influenza-like illness incidence in a temperate chinese region: an ecological study from 2008 to 2015 date: 2017-03-06 journal: bmj open doi: 10.1136/bmjopen-2016-013159 sha: doc_id: 2451 cord_uid: r7a0orh7 objective: to assess the effects of winter/summer school breaks on occurrences of influenza-like illness (ili). methods: we jointly analysed ili surveillance data with the timing of school breaks in a temperate district in beijing, china from 2008 to 2015. ili incidence rate ratios (irrs) of schoolchildren (5–14 and 15–24 years of age) to adults (25–59 and >60 years of age) were used to measure the age shift of ili incidence before, during and after the 4-week winter/7-week summer breaks. serfling-based poisson regression model with adjustment for unmeasured confounders was built to further assess the effect of winter school breaks. results: ili incidences were consistently lower during winter breaks than before winter breaks for all age groups. irrs of younger schoolchildren aged 5–14 to adults were higher during winter school breaks than before breaks, while the opposite was true for the irrs of older schoolchildren aged 15–24 to adults. schoolchildren-to-adults irrs during summer breaks were significantly lower than before or after school breaks (p<0.001). conclusions: both winter and summer breaks were associated with reductions of ili incidences among schoolchildren and adults. our study contributes additional evidence on the effects of school breaks on ili incidence, suggesting school closure could be effective in controlling influenza transmission in developing countries. schoolchildren play a major role in the spread of influenza considering their high clinical attack rates, high social contact rates and increased viral shedding compared with adults. 1 reactive school closure at the initial phase of a pandemic is considered as an effective non-pharmaceutical intervention to mitigate the spread of influenza. both schoolchildren and their caregivers are expected to have a lower risk of infection after school closure by altering social mixing patterns. 2 several empirical studies have confirmed the effects of school closure on reducing influenza virus transmission after immediate closure of schools at an early stage of a pandemic, including studies from hong kong, 3 france, 4 germany, 5 eight european countries, 6 canada 7 and usa. 8 however, the assessment of effects of school closure on influenza transmission remains challenging. reactive school closure during an influenza pandemic was often accompanied by other control measures such as intensive screening, border control measures and improved sanitation, which may confound the assessment of effects of school closure on influenza incidence. to differentiate the effects of school closure from other accompanied interventions and changed health-seeking behaviour, some studies have evaluated the effects of school closure on influenza transmission by focusing on school breaks rather than reactive school closures. as school calendars are usually set at the beginning of each school term, observed changes of influenza incidence before, during and after school breaks are believed to be less confounded than school closures strengths and limitations of this study ▪ we assessed the impact of school breaks on influenza activity. ▪ age shift of influenza-like illness incidence from schoolchildren to adults during school breaks indicated a reduction of influenza transmission in schoolchildren. ▪ serfling-based poisson model was built to adjust for unmeasured confounders. ▪ residual confounding such as population mobility, change of health-seeking behaviour may exist. aimed to control a pandemic. 9 10 significant associations between winter school breaks and temporary reductions of influenza-like illness (ili) incidence have been found in chile 10 and argentina, 9 as children were less likely to be gathered during school breaks thus reducing the chance of influenza transmission. we hypothesised that the 4-week winter/7-week summer breaks in china might also be relevant to ili incidence reductions among schoolchildren and adults. our study aims to assess the effects of winter/summer school breaks on ili incidence by estimating the ili incidence ratio of schoolchildren to adults before, during and after the breaks in xicheng district, beijing, china. beijing is located in northeastern china and has a temperate and continental climate with four distinct seasons. xicheng district is located in the centre of beijing. weekly district-wide ili time series during 2008-2015 was obtained from beijing medical institutions in communicable disease surveillance and early warning system. ili was defined as fever (temperature of 38°c or greater) and a cough and/or a sore throat in the absence of a known cause other than influenza. the ili reporting system recorded weekly ili cases into five age groups (0-4, 5-14, 15-24, 25-59 and 60 years or older). beijing children's hospital, affiliated with capital medical university is one of the sentinel hospitals for ili surveillance, and accounts for more than half of the total paediatric outpatient visits in beijing. to render our data more representative of ili rates across the entire beijing population, we age-standardised our ili visit data in xicheng district to the age distribution of beijing's population. age-specific population sizes in xicheng district were collected from china's 2010 population census. we assumed that the age-stratified population remained constant over the study period. we ignored the immunised population as the influenza vaccination coverage was below 3% in the general population. school calendars for winter/summer breaks of primary and secondary schools in beijing were obtained directly from beijing municipal education commission. school calendars were usually announced ahead of each academic year. winter breaks were often in january and february, which lasts for 4 weeks and covered the 7-day statutory holidays for national chinese new year. summer breaks are in july and august, which last for 7 weeks. the exact starting dates of winter/summer school breaks varied slightly across years and schools. on average, winter breaks in beijing started at the 5th week and ended at the 8th week, while summer breaks started at the 29th week and ended at the 35th week. data analysis ili incidence rate ratios (irrs) of schoolchildren (5-14 and 15-24 years of age) to adults (aged 25-59 and 60 years or older) were used to measure the age shift of ili incidence before, during and after the 4-week winter/7-week summer breaks. in this study, we used 25 years of age as the cut-off point for schoolchildren and adults, because the ili surveillance system in beijing reported weekly ili numbers by age groups of 0-4, 5-14, 15-24, 25-59 and 60 years or older, without specific age for each case. owing to data availability, we grouped individuals under 25 years of age as schoolchildren, which was close to but might be a bit older than the true cut-off for schoolchildren. similar to previous literature, [10] [11] [12] we compared schoolchildren-to-adults irrs before, during and after the winter/summer breaks using two-sided z tests. as the reduction in the ili incidence during the winter break was maintained on average for 2 weeks after the end of the winter break, we allowed a 2-week window and defined before/after school breaks as 2 weeks apart from the break. we considered a 4-week window before and after the winter/summer break as a sensitivity analysis. in order to control the effects of unmeasured confounders of long-term and seasonal trends, we built a serfling-specified poisson regression model to assess the effect of winter breaks. a vector of dummy variables for each year was considered and sinusoidal terms of weeks were added in the poisson model to adjust for long-term and seasonal trends, respectively. as weekly adjacent ili visits were temporally correlated, we further added an autocorrelated regression term to adjust for autocorrelation in the poisson model. 13 details of the model can be found in online supplementary appendix and tables s1-s2. model fitting was evaluated by pseudo r 2 values. irrs of during/after to before winter breaks were used to evaluate the change of ili incidence before, during and after winter breaks in each age group. irrs of during/after to before winter breaks smaller than 1 indicated that ili visits during/after winter breaks were lower than those before winter breaks. we presented the estimates as weekly irrs, which was calculated by dividing the weekly ili incidence during/after winter breaks to an average of ili incidence 2 weeks before winter breaks. data analyses were conducted in r v.3.2.3 using package nlme. 14 from january 2008 to december 2015, center for disease control and prevention of xicheng district in beijing recorded 1 294 279 ili visits and a total of 41 269 165 clinic attendances. percentage of number of ili visits to number of total clinic attendances (ili%) had been widely used as an indicator for influenza activities. 15 figure 1a illustrates age-specific trends in weekly ili% throughout 8 years. peaks of influenza activities appear from end of december to middle of january across different age groups in the past years. summer peaks were observed for groups of age 0-4 and 5-14 without a consistent annual pattern. during the winter school breaks, ili incidence decreased substantially in age groups of 5-14 and 15-24, but modestly in age groups over 25 (figure 1b). during the summer school breaks, only ili incidence in age group of 5-14 reduced dramatically, while levelled off for other age groups (figure 1b). a decline of schoolchildren-to-adults irr during school breaks implied an age shift in patients with ili towards adults, and further suggested a reduction of influenza transmission among schoolchildren. irrs of younger schoolchildren aged 5-14 to adults was higher during winter school breaks than before breaks, while the opposite was true for the irrs of older schoolchildren aged 15-24 to adults (table 1, figure 1c) . during summer breaks, schoolchildren-to-adults irrs were significantly lower than those of 2-week before or after school breaks ( p<0.001). to be specific, irrs of age 5-14 to age above 60 had declined by 13.3% (95% ci 3.59%, 22.1%). the above conclusion was robust when we considered a 4-week window before and after winter/ summer break (see online supplementary table s3). since long-term trend and seasonal pattern of ili visits could be two confounders in the previous analysis of irrs, we used a poisson regression model to assess the effect of school breaks on ili incidence. the models generally well fitted the observed ilis with adjusted pseudo figure 1 age-specific ili incidence rates, adjusted average ili incidence rates, adjusted average ili irrs of influenza surveillance data by week in beijing, 2008-2015. (a) weekly ili incidence rates for five age groups. (b) adjusted average ili incidence per 10 000 persons by week. (c) adjusted average ili incidence rate ratio of schoolchildren-to-adult by week. the upper plot is irrs of age 5-14 to adults; the lower plot is irrs of age 15-24 to adults. ili, influenza-like illness; irr, incidence rate ratio. r 2 values for the five age groups (0-4, 5-14, 15-24, 25-59 and 60 years or older) ranging from 0.44 to 0.56. after incorporating the autocorrelation within error terms, we estimated irrs of during/after to before winter break and their cis (table 2) . during-to-before irrs were smaller than 1 for all age groups. however, the starting time and duration of the ili reduction varied across different age groups. in age group of 15-24, ili incidence dropped substantially by around 40% during the whole school break and lasted for 4 weeks thereafter. declines of during-to-before irrs for age groups of 0-4 and 5-14 were also observable but not necessarily smaller than declines in adults (table 2). for adults (age 25-59 and 60 years or older), ili reductions were statistically significant in the second and third week of the break and lasted for about 2 weeks (table 2). among all age groups, incidence of ili visits returned to regular patterns with none of after-before irrs statistically significant 3 weeks after the end of winter breaks (table 2) . consistent with other studies in china and other temperate regions, 16 we found peaks of influenza in beijing appeared synchronised across age groups in most winters and springs during 2008-2015. for age groups of 0-4 and 5-14, peaks of ili% had also been detected before summer breaks, which might be driven by final examination stress, poor classroom ventilation, incompletely developed immune system and other unfavourable factors related to young children. since ili visits could be caused by influenza, respiratory tract infection and other illness, ili% is a broad indicator of respiratory disease activity not entirely specific for influenza and could be affected by changes in health-seeking behaviour. 10 therefore, laboratory surveillance data are necessary for further conclusions on age-specific seasonal influenza activities from a public health perspective. in this study, our findings generally support that winter/summer school breaks could achieve temporary reductions in ili incidence rates, especially among schoolchildren (age of 5-14 and 15-24). in consistency with previous empirical studies in chile, usa and european countries, 10 17 the average reduction in schoolchildren-to-adult irrs in beijing lasted for up to 2 weeks (table 1) after school sessions resumed, and this allowed a few successive chains of transmission for influenza virus to reach full-scale transmission. 10 the finding that ili incidence rates decreased substantially in schoolchildren yet modestly in adults during winter breaks is consistent with past work on age-specific influenza transmission. 17 moreover, the reduction in ili during the school holidays went beyond the dips in ili ±4 weeks surrounding the holiday weeks (see online supplementary appendix), further supporting our suggestion of lower influenza transmission in schoolchildren during school breaks. the finding that irrs of schoolchildren aged 5-14 to adults increased significantly in winter breaks is inconsistent with findings of influenza activities in chile. 10 although ili incidence rates of age group 5-14 had been reduced by 26.3% (95% ci 24.9% to 27.7%) after entering winter breaks in 2008-2015, reductions in ili incidence rates for adults' groups were even larger and thus resulted in increased irrs of age group 5-14 to adult. in addition to different strains of influenza circulating in a certain year, there could be many other human behaviour factors playing a role in regulating seasonal age-specific ili visits in china. for example, under pressure of final examinations before winter *small p values indicate that the irr for the period before the break is significantly higher than that for the period during the break (or the irr for the period during the break is significantly higher than that for the period after the break). ili, influenza-like illness; irr, incidence rate ratio. breaks, some schoolchildren are unwilling to spare any time to see a doctor when they had influenza-like symptoms. during chinese spring festival vacation (covered by winter breaks), most of migrant workers in beijing went back to their hometown 18 19 and local people who stayed in the city were unwilling to seek medical advice for fear of missing family gatherings (see online supplementary figure s1). moreover, apparent drops of total outpatient visits were observed in every january (see online supplementary figure s1), suggesting the change of healthcare-seeking behaviour around the holidays. in fact, irrs of schoolchildren aged 5-14 to adults during spring festival were significantly higher than other weeks within winter breaks. 20 21 consequently, we infer the enhanced irrs of schoolchildren aged 5-14 to adults in winter breaks are due to integrated effects of spring festival, population migration and changes in health-seeking behaviour. we established an autocorrelated poisson regression model to control the effect of potential confounders that may distort age-specific ili incidence rates, such as seasonal change of ili visits and year. we found a significant reduction in ili incidence rates for age group of 5-14 when entering winter breaks, compared with those of 2 weeks before (irr 0.89, 95% ci (0.81 to 0.98)). however, the reduction is less significant when comparing with the rates throughout the entire break for the adults' groups. the phenomenon that reduction of ili incidence rates for age above 60 was larger than those for age 25-59 and close to those for age [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] (table 2) could be due to age-specific health-seeking behaviour, immune system characteristics and residual confounders of seasonal ili visits. although a serfling model has been adopted, our model is still relatively conservative in controlling seasonal pattern of ili because timing of peaks of influenza activities had severe overlaps with timing of winter breaks, and the overfitting could underestimate the effect of winter breaks on decrease of irrs. therefore, additional analysis that considers high variability in the temporal relationship between winter breaks and weekly ili rates across influenza seasons are needed and incorporating environmental factors or human behaviour data may better clarify the relationship between age-specific influenza activity and winter breaks. we quantified the reduction in irrs, and also analysed age-specific timing of the decline in ili incidence. similar to the study in argentina, 9 a stepwise trend of the effect on ili had been detected. age groups of 5-14 and 15-24 experienced the initial significant decrease in ili during the first week of winter breaks and lasted for 4 weeks. the reduction effect was caught up by other age groups, and lasted for 4 weeks for 0-4 age group and 2 weeks for adults' groups. the spread of influenza likely varies according to population subgroup. 22 overall, our study supports winter breaks as effective means to prevent the epidemic of winter-to-spring influenza activities. schoolchildren-to-adults irrs consistently decreased during summer breaks compared with 2-week before *irrs were used to estimate whether incidence of ili-associated visits in a particular week were lower, higher, or did not deviate from the expected seasonal ili patterns. each row represents a separate regression model. ili, influenza-like illness; irr, incidence rate ratio. summer breaks. according to our analysis, protective effect of summer breaks on schoolchildren of age 5-14 was more obvious than that on adults. we emphasise that results on effects of summer breaks on influenza activities should be interpreted with caution, because respiratory viruses other than influenza virus were active during summer that could make clinical surveillance less specific to influenza. 23 although our study sustains the positive effect of school breaks in controlling influenza activities, insufficient evidence is provided on the effect of school breaks in order to support the decision on duration of the break and the timing with respect to the influenza season. furthermore, the economic cost of school breaks that is mainly due to absenteeism of working parents who have to stay home to take care of their children remains difficult to estimate. 24 further analysis of the environmental or social factors influencing the transmission of seasonal and pandemic influenza is required in order to provide additional information for policymakers and public health officials to use when considering measures to control pandemic influenza. estimating household and community transmission parameters for influenza social contact networks for the spread of pandemic influenza in children and teenagers school closure and mitigation of pandemic (h1n1) 2009, hong kong closure of schools during an influenza pandemic social contacts of school children and the transmission of respiratory-spread pathogens estimating the impact of school closure on social mixing behaviour and the transmission of close contact infections in eight european countries effects of school closure on incidence of pandemic influenza in alberta on influenza and school closings: time for prospective studies effect of winter school breaks on influenza-like illness rates of influenza-like illness and winter school breaks characterizing the epidemiology of the 2009 influenza a/h1n1 pandemic in mexico spatial and temporal characteristics of the 2009 a/h1n1 influenza pandemic in peru introductory time series with r linear models with r. 2nd edn overview of influenza surveillance in the united state characterization of regional influenza seasonality patterns in china and implications for vaccination strategies: spatio-temporal modeling of surveillance data estimating the impact of school closure on influenza transmission from sentinel data a study on the patterns of migration in chinese large and medium cities areal differentiation of inter-provincial migration in china and characteristics of the flow field spatial pattern of severe acute respiratory syndrome in-out flow in 2003 in mainland china analysis on the multi-distribution and the major influencing factors on severe acute respiratory syndrome in beijing effects of school closures, 2008 winter influenza season, hong kong influenza-associated hospitalizations in the united states estimating the costs of school closure for mitigating an influenza pandemic competing interests none declared.provenance and peer review not commissioned; externally peer reviewed. open access this is an open access article distributed in accordance with the creative commons attribution non commercial (cc by-nc 4.0) license, which permits others to distribute, remix, adapt, build upon this work noncommercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. see: http:// creativecommons.org/licenses/by-nc/4.0/ key: cord-347079-1zbsbcdd authors: silverman, justin d.; hupert, nathaniel; washburne, alex d. title: using influenza surveillance networks to estimate state-specific prevalence of sars-cov-2 in the united states date: 2020-06-22 journal: sci transl med doi: 10.1126/scitranslmed.abc1126 sha: doc_id: 347079 cord_uid: 1zbsbcdd detection of sars-cov-2 infections to date has relied heavily on rt-pcr testing. however, limited test availability, high false-negative rates, and the existence of asymptomatic or sub-clinical infections have resulted in an under-counting of the true prevalence of sars-cov-2. here, we show how influenza-like illness (ili) outpatient surveillance data can be used to estimate the prevalence of sars-cov-2. we found a surge of non-influenza ili above the seasonal average in march 2020 and showed that this surge correlated with covid-19 case counts across states. if 1/3 of patients infected with sars-cov-2 in the us sought care, this ili surge would have corresponded to more than 8.7 million new sars-cov-2 infections across the us during the three-week period from march 8 to march 28, 2020. combining excess ili counts with the date of onset of community transmission in the us, we also show that the early epidemic in the us was unlikely to have been doubling slower than every 4 days. together these results suggest a conceptual model for the covid-19 epidemic in the us characterized by rapid spread across the us with over 80% infected patients remaining undetected. we emphasize the importance of testing these findings with seroprevalence data and discuss the broader potential to use syndromic surveillance for early detection and understanding of emerging infectious diseases. the ongoing severe acute respiratory syndrome-coronavirus 2 (sars-cov-2) pandemic continues to cause substantial morbidity and mortality around the world [1, 2] . regional preparation for the pandemic requires estimating the growth rate of the epidemic, the timing of the epidemic peak, the demand for hospital resources, and the degree to which current policies may curtail the epidemic, all of which benefit from accurate estimates of the true prevalence of the virus within a population [3] . confirmed cases are thought to be underestimates of true prevalence due to some unknown combination of patients not reporting for testing, testing not being conducted, and false-negative test results. estimating the true prevalence of sars-cov-2 would inform the scale of upcoming surges in hospital demand, the proportion of individuals who remain susceptible to contracting the disease, and estimates of key epidemiological parameters such as the epidemic growth rate and the fraction of infections that are subclinical. the current literature suggests that the predominant symptoms associated with covid-19 are fever, cough, and sore throat; that is, patients often present with an influenza-like illness (ili) yet test negative for influenza [4, 5] . as covid-19 often presents with similar symptoms to influenza, existing surveillance networks in place for tracking influenza could be used to help track covid-19. outpatient ili surveillance has proven to be a useful tool for assessing the impact of influenza [6, 7] . when combined with the number of providers and patients in a given region, ili surveillance allows estimation of influenza prevalence and severity [8, 9, 10, 11, 12, 13, 14] . studies of outpatient ili have repeatedly demonstrated that confirmed influenza case rates underestimate disease burden, likely due to preferential testing of more severe cases [8, 14, 9, 13 ] . together these features suggest that ili surveillance could provide a crucial tool for estimating covid-19 prevalence within the us. here, we quantified the baseline prevalence of non-influenza ili in the us over the past 10 years and identified a recent surge of non-influenza ili starting the first week of march, 2020. this surge of excess ili correlated with known patterns of sars-cov-2 spread across states within the us yet was orders of magnitude larger than the number of confirmed covid-19 cases reported by the end of march. using influenza surveillance networks to estimate statespecific prevalence of sars-cov-2 in the united states admitted to the hospital to decrease. however, although the daily number of ili visits to emergency departments across new york city increased in march 2020, the proportion of those patients who went on to be admitted also increased by as much as 3-fold compared to the baseline rate prior to march ( fig. s3a ). this observation suggests that patients with mild ili presented less often to hospital emergency departments. such a decrease in care-seeking behavior for mild ili, if similar across the us, could deflate the estimated size of the ili surge in the later weeks of march by a factor of approximately 3. if non-ili patients were less likely to seek medical care, then we would expect that the number of patients complaining of other symptoms not typically associated with covid-19 (for example vomiting) would also decrease compared to prior years. in the month of march, the daily number of patients presenting with vomiting decreased by as much as a factor of 3 compared to the baseline rate in prior years (fig . s3b ). assuming that all non-ili conditions were similarly decreased during march, this would suggest that our estimates of the ili surge could be inflated by as much as a factor of 3. this assumption is conservative as it assumes that even individuals with severe conditions (such as severe trauma) would avoid seeking health-care in response to covid-19 at the same rate as those with more mild conditions such as vomiting. however, the potential 3-fold decreased care-seeking behavior for non-ili conditions cancels out the potential 3-fold decreased care-seeking behavior of mild ili, suggesting that our estimates of prevalence based on the ili surge may be insensitive to recent changes in care-seeking behavior (fig. s3c). overall these estimates suggest a conceptual model in which health care utilization for both mild ili and non-ili conditions declined at similar rates as covid-19 increased in the us. to estimate the proportion and magnitude of the march 2020 us ili surge attributable to sars-cov-2 infections, we made the following three assumptions: (1) that the patient population reported by sentinel providers is representative of their state each week; (2) that changes in care-seeking behavior of ili patients is occurring at a similar rate as that of other non-ili patients; and (3) that the total number of patients in the us who require medical care over the course of a year has not substantially changed since 2018. our first assumption is common and underlies prior studies which have used ili to estimate influenza prevalence [8, 14] . our second assumption is supported by our new york city analysis which suggests that both mild ili and non-ili conditions have seen similar changes in healthcare seeking behavior. our third assumption is based on the observation that the increasing need for health-care between march 8 and march 28, 2020 due to covid-19 is likely small compared to the approximately 1 billion outpatient encounters that occur annually [18, 19] . these assumptions together with surveys describing the average number of patients seen by providers [19] , the number of providers in each state [20] , and the total number of outpatient visits per year [21, 18] , allowed us to estimate that, if outpatient clinics remained open during the covid-19 epidemic, we would expect that there would have been approximately 2.8 million patient encounters with ili due to covid-19 between march 8 to march 28, 2020 (95% credible set 2.6 million to 3.0 million). not all patients infected with sars-cov-2 will present to a health-care provider with ili. although we cannot directly measure the rate of such sub-clinical cases, a number of prior studies on asymptomatic rates of covid-19 and the careseeking behavior of ili patients in the us suggest a lowerbound on the subclinical rate of patients with ili. a recent study of passengers on the diamond princess cruise-ship accounted for a right-censoring of patients sampled and estimated that 18% of patients infected with sars-cov-2 are asymptomatic for the course of their infection (95% credible set 16% to 20%). this estimate likely represents an underestimate given that the majority of passengers were over 60 years old, a demographic thought to have a lower asymptomatic rate than younger individuals [22] . beyond asymptomatic individuals, a large study of adult health-care seeking behavior in the united states found that, of a random sample of over 17,000 individuals with ili, 40% of those went on to seek health care [23] . together these additional contributions from sub-clinical cases correspond to a mean clinical rate of 32% (the overall rate at which sars-cov-2 cases seek medical care) and a lower bound of 8.7 million sars-cov-2 infections between march 8th and march 28th (95% credible set 8.0 million to 9.4 million). prevalence estimates for each state within this time-period are shown in fig. s4 . we define the syndromic case detection rate as the number of confirmed covid-19 cases in a week divided by the size of the ili surge that week. the syndromic case detection rate varied by state and over time ( fig. s5 ). our estimated syndromic case detection rates increased over the month of march; this was expected given increases in testing capacity across the us since the february 28 detection of community transmission in washington state. for the week ending march 14, covid-19 cases in the states with the highest estimated syndromic case detection rate (washington, nevada, and michigan) only captured approximately 1% of ili surges in those states. in the last week of the month ending on march 28, the syndromic case detection rate across the us increased to 12.5% (95% credible interval 9.5%-18.3%). the true prevalence of sars-cov-2 is unknown at the time of this writing. however, if we assume the excess noninfluenza ili is almost entirely due to sars-cov-2, an assumption that becomes more valid as sars-cov-2 becomes more prevalent, we can use the excess non-influenza ili to define lower bounds on the exponential growth rate of the us sars-cov-2 epidemic. by estimating the number of patients visiting clinics for covid-19 in the us in march, we can also identify the mutual dependence of exponential growth rates, the rate of sub-clinical infections, and the time between the onset of infectiousness and a patient reporting as ili (fig. 2) . using stochastic susceptible, exposed, infectious and recovered (seir) simulations of us covid-19 epidemics with a january 15 start date [24] , we find that an initial epidemic doubling time longer than 4 days is unlikely to explain the ili surge. doubling times longer than 4 days fail to produce enough infected individuals to match the observed excess ili. doubling time faster than 4 days can explain the observed excess ili with a clinical rate that depends on the growth rate. here, we define the clinical rate as the proportion of infected individuals who present to a health care provider. in keeping with our sub-4 day doubling times, we found that across the entire us, new deaths due to covid-19 doubled every 3.01 days over the month of march (±0.001, p-value of test that doubling rate is less than 4 days approximately 0). if there was only a 1-day lag from onset of infectiousness to presentation with ili and the entirety of the first week of the us ili surge is comprised of patients with covid-19, then an epidemic starting january 15th and growing at the rate of deaths in the us would imply a 12% clinical rate ( fig. 2a) . a four-day lag between the onset of infectiousness and presentation with ili yields a clinical rate of 25% among the 87% of simulations which could account for the ili surge. the 25% overall clinical rate estimated from a january 15 start date and the doubling time of us covid-19 deaths is in close agreement with the 32% clinical rate we estimated independently based on a 18% asymptomatic rate and 40% symptomatic clinical rate. although our epidemic model suggests the first week of the ili surge is consistent with the us epidemic start date and growth rate, the ili surge across the us peaked the week ending march 21, much earlier than our epidemic models, suggesting the epidemic in the us differed from the seir model through some combination of factors. such factors could include successful interventions, even faster decreases in care-seeking than observed in new york, heterogeneity in susceptibility [25] , or an early epidemic doubling faster than every 3 days. faster growth rates require lower clinical rates to explain the ili surge. epidemic curves growing at the rate of deaths in italy, doubling every 2.65 days, could better match the curvature of the ili surge by peaking around mid to late march, but would imply a clinical rate of 4.7% the second week of march with a 4-day lag between onset and recorded as ili ( fig. 2b and c). if the entirety of the ili surge was attributable to covid-19, the slowest-possible doubling time for the us epidemic which can explain the ili surge would be a doubling time of of 4 days. any evidence of significant secondary introductions, super-spreading, or rapid transmission events in early transmission chains will decrease these estimated clinical rates [26] . evidence of slow initial spread would increase the estimated clinical rates. last, estimating the infection fatality rate from the ili surge requires knowing the clinical rate and the delay from clinical presentation with ili to death. if patients present with ili at the onset of their illness, exhibit a 16 day median lag between onset and death [27] , and have a 32% clinical rate as estimated from the 18% asymptomatic rate and 40% clinical rate of symptomatic covid-19 cases, then the observed ili surge corresponds to an infection fatality rate of 0.29%. we stress that estimating the infection fatality rate from this ili surge is highly sensitive to both the lag from presentation with ili to death and the clinical rate ( fig. s6 ). consequently, the ili surge is compatible with fatality rates ranging from 0.07% to 1.4% depending on the unknown sub-clinical rate and lag from presentation with ili to death. under the cdc planning scenarios specifying a 4-day lag from onset of symptoms to presentation to the doctor with ili [28] and a 15 day lag from onset to death, the resulting 11-day lag from ili to death produces ifr estimates of 0.57% (0.51-0.68% 95% credible set) for the unadjusted ili surge and 0.19% (0.17-0.22% 95% credible set) for the ili surge adjusted to account for asymptomatic and subclinical cases. we use outpatient ili surveillance data from around the us to estimate the prevalence of sars-cov-2. we found a clear, anomalous surge in ili outpatients during the covid-19 epidemic that correlated with the progression of the epidemic in multiple states across the us. the surge of non-influenza ili outpatients was much larger than the number of confirmed case in each state, providing evidence of large numbers of probable symptomatic covid-19 cases that remained undetected. this result is also consistent with ili excess observed in france in late-february/early-march [29] . additionally, this finding predicts that the slowest epidemic doubling time that could explain the ili surge would be 4 days, and that this rate could only be achieved with unusually fast early transmission or super-spreading events and a clinical rate near 100%. consistent with this prediction, we found that deaths due to covid-19 within the us doubled every 3.0 days and note that this empirical growth rate for the us epidemic can account for the ili surge with a 25% clinical rate assuming a 4 day lag from the onset of infectiousness to presentation as an outpatient with ili. together, these results suggest that sars-cov-2 spread rapidly throughout the us since its january 15th start date and was likely accompanied by a large undiagnosed population of potential covid-19 outpatients with presumably milder distribution of clinical symptoms than estimated from prior studies of sars-cov-2+ inpatients. excess ili appears to have peaked during the week starting on march 15th, leading the observed ili dynamics to diverge from the overall epidemic dynamics implied by the growth rate of covid-19 deaths in the us. if the ili dynamics were proportional to the epidemic curve then the two could be related via a constant subclinical rate. however, the changing ratio between covid-19 prevalence estimated by the ili surge and the epidemic curves parameterized by the growth rate of us deaths suggests additional mechanisms may be behind the ili slowdown. mechanisms which can explain the difference between our simulated epidemic curves and the ili surge include effective social distancing, disproportionate reductions in ili care-seeking behavior relative to non-ili care-seeking behavior, or heterogeneity in susceptibility or contact structure not captured in our seir model [25] . our empirical estimate of the size of the ili surge has several potential limitations. first, the observed ili surge may represent more than just sars-cov-2 infected patients. a second epidemic of a non-seasonal pathogen that presents with ili could confound our estimates of ili due to sars-cov-2. however, this seems unlikely as additional viral surveillance through the us centers for disease control and prevention (cdc) suggests that between march 8 to march 28 other monitored respiratory viruses were at low prevalence [30] . nonetheless, were our approach to be used during winter months, additional steps would be needed to account for concomitant non-influenza seasonal pathogens. additionally, our assumption that changes in health-care seeking behavior are similar between mild ili and non-ili condition may be incorrect. although this assumption was supported by new york city emergency department surveillance data, it is possible that differential health-care seeking would be present in other locations or in the outpatient setting. last, it is also possible that our use of ili data has underestimated the prevalence of sars-cov-2 within the us. although early clinical reports focused on cough and fever as the dominant features of covid-19 [5] , other reports have documented digestive symptoms as the complaint affecting up to half of patients with laboratory-confirmed covid-19 [31] , and alternative presentations, including asymptomatic or unnoticeable infections, could result in underestimation of sars-cov-2 prevalence. additionally, our models have several limitations. first, we assumed that ili prevalence within states can be scaled to case counts at the state level. this is based on the assumption that the average number of cases seen by sentinel providers in a given week is representative of the average number of patients seen by all providers within that state in a given week. errors in this assumption would cause proportional errors in our estimated case counts and syndromic case detection rate. second, our us-wide seir models vary by growth rate alone and as such may not capture important heterogeneity in susceptibility or transmission as well as regional variation, intervention-induced changes in transmission, or clustering of infection outbreaks. our models were used to illustrate that the ili surge is consistent with an estimated growth rate and start date for the us epidemic and to specify the mutual dependency of growth rate, the lag between the onset of infection and presentation to a doctor, and clinical rates. finer models with regional demographic and case-severity compartments are needed to translate our range of estimated prevalence, growth rate, and clinical rates into actionable models for public health managers. last, our method of calculating the infection fatality rate relied on assumptions about the clinical rate and the delay from patients recorded as ili to death. our clinical rate required using patterns of care-seeking for typical seasonal causes of ili as did our delay from ili to death; consequently, neither should be relied on as a definitive source for covdi-19 and estimating the clinical rate and delay from ili to death for covid-19 specifically will reduce the large uncertainty around our iliestimated infection fatality rates. despite these potential limitations, the ili surge identified in syndromic surveillance time-series allowed early estimates of covid-19 prevalence, estimates that were not possible from confirmed case data due to early logistical delays in sars-cov-2 testing in the us. our prevalence estimates are supported by a serosurvey conducted in new york state. we estimated that over 8.3% of new york state residents were infected by sars-cov-2 by march 28; on april 23, 2020, new york state announced that 14% of residents had evidence of past infection by sars-cov-2 by march 29 at which time the cumulative pcr-confirmed case counts totaled only 0.3% of new york's population [32] . although an ili surge tightly correlated with covid-19 case counts across the us and consistent with the new york state serology strongly suggests that sars-cov-2 has potentially infected millions in the us, further laboratory confirmation of our hypotheses are still needed to guide public health decisions. our findings make testable predictions that one would find relatively high seroprevalence in other states that have already seen an ili surge and that seroprevalence of individuals infected in march across states is proportional to relative sizes of the states' ili surges. a study of ili patients from mid-march who were never diagnosed with covid-19 could produce a focused test of our predictions about the number and regional prevalence of undetected covid-19 cases presenting with ili during that time. if seroprevalence estimates beyond new york state continue to corroborate our prevalence estimates from syndromic surveillance, this would strongly suggest lower case severity rates for covid-19 than were assumed in late march by comparing pcr-confirmed case counts to deaths. further corroboration of our estimates of the magnitude of the ili surge would suggest ili and other public time-series of outpatient illness allow early and reliable estimates of crucial epidemiological parameters for rapidly unfolding, novel pandemic diseases. as not all novel pandemic diseases are expected to present with influenza-like symptoms, surveillance of other illnesses that commonly present in the outpatient setting could provide a vital tool for rapidly understanding and responding to novel infectious diseases. the goal of our study was to use publicly available data to estimate the number of patients seeking care for non-influenza ili in excess of seasonal trends during the three weeks spanning march 8 to march 28, 2020 and then use this ili surge to estimate covid-19 incidence in march and parameterize epidemiological model growth rates and clinical rates. the ili surge detection above produced an excess proportion of patients visiting outpatient providers for non-influenza ili in each week and each state. to scale up the proportion of patients to a national number of covid-19 cases, we estimated the number of patients per sentinel provider in the cdc dataset, normalized that number of patients per provider to a number of patients per doctor, and scaled that up by an estimated number of practicing doctors in the us. the result was an estimated number of covid-19 patients visiting doctors in each state for each week -we called this our "unadjusted" ili surge. the unadjusted ili surge is an under-estimate of covid-19 prevalence due to only clinical infections, those that seek medical care. we accounted for both asymptomatic infections and symptomatic but sub-clinical infections to produce an "adjusted" ili surge as our final estimate of covid-19 incidence in each state and each week. we then used the unadjusted and adjusted ili surges to estimate syndromic case detection and fatality rates. we also used the unadjusted ili surge as an empirical observation to evaluate epidemiological modelling of covid-19 growth rates and clinical rates in the us. throughout our methods, we use i to denote the index state i and let t index week t (with t=0 referring to october 3, 2010; the start of state-specific ilinet surveillance). since 2010 the cdc has maintained ilinet for weekly influenza surveillance. each week approximately 2,600 enrolled providers distributed throughout all 50 states as well within the ilinet dataset, new york city and new york were summed into a combined new york variable representing both new york city and the surrounding state. due to incomplete data in one or more of the data-sources described above the virgin islands, puerto rico, the commonwealth of the northern mariana islands, and florida were excluded from subsequent analysis. in addition, to match the weekly reporting of ili from ilinet, daily cumulative confirmed covid-19 cases were converted to weekly counts of new cases by ( ) to subtract influenza signal from it y we assumed that the population of patients with ili within a state are the same population that are potentially tested for influenza. this assumption allows us to calculate the number of non-influenza ili cases as mean imputation based on neighboring states was used to address missing values in laboratory influenza quantification. to assess the impact of this model for extracting noninfluenza ili signal, we calculated covid-19 prevalence without first removing signal from influenza, we found little change in our prevalence estimates ( fig. s7 ). this likely reflects that influenza also demonstrates strong seasonal patterns that can be addressed as discussed below. we to account for variation in the number of total patients, we modeled it y as binomial distributed. to account for correlation in non-influenza ili over time, we use a gaussian process model which assumes that weeks that are closer together will have more similar levels of non-influenza ili. the following model reflects these modeling choices: where  refers to a gaussian process. we made the following prior specifications: we set the bandwidth parameter for the squared exponential kernel as ρ=3 representing a strong local correlation in time that died off sharply beyond 3 weeks, α=1 representing a signal to noise ratio of approximately 1, ν=1 and ξ=1 representing weak prior knowledge regarding the overall scale of variation in the latent space. collected using the function basset from the r package stray [34] ; a total of 4000 such samples were collected, for each state, in this analysis. we defined the prevalence of non-influenza ili in excess of normal seasonal variation as to investigate whether our results were sensitive to the above model specification, we alternatively used the sample mean and variance from years 2010-2018 as an estimate of typical seasonal non-influenza ili. despite not accounting for the binomial count structure of ili data or correlations in the proportion ili between weeks, this simpler model resulted in nearly identical prevalence estimates ( fig. s8 ). still, we used the gp-derived estimates throughout this paper due to their better accounting for the known binomial count and week-to-week correlation structure of ili-causing pathogen prevalence. to exclude variation attributable to unseasonably high rates of other ili causing viruses (such as the outbreak of rsv in washington state in november-december 2019) we only investigated * it y for weeks after march 7th 2020 as only these later weeks had high correlation to the covid-19 confirmed case rate ( fig. s2 ). as new covid-19 case counts it z  represent the number of confirmed cases in an entire state and ilinet data represents the number of cases seen by a select number of enrolled providers, we had to estimate scaling factors i w to enable comparison of ilinet data to confirmed case counts at the state level. let * it π denote the probability that a patient with ili in state i has covid-19 as estimated from ilinet data. let i p denote the population of state i and let i b denote the number of primary care providers per 100,000 people in state i. we translated the inferred proportion of individuals with ili due to covid-19 to the state level by considering the average number of patients seen across all providers in the state in a 5-day work-week. in addition, we added a discount factor λ=0.55 to calibrate these estimates with prior reports regarding the total number of outpatient visits per year [18] . this yielded our estimated number of covid-19 cases (excess ili at the state level) as where m=20.2 is the mean number of patients seen by physicians per day [19] . to account for the contribution of sub-clinical sars-cov-2 infections we used a recent analysis of cohort surveillance from the diamond princess [35] . monte-carlo simulations were used to propagate error from our uncertainty regarding potential asymptomatic infections affecting the clinical rate assuming that the majority of sars-cov-2 testing within the us has been directed by patient symptoms [36] , the pool of newly diagnosed sars-cov-2+ patients is a subset of the pool of sars-cov-2+ patients who are identified as having ili. therefore, we calculated the probability that a sars-cov-2+ patient with ili who seeks medical care will be iden the exact lag from an outpatient being recorded as ili to death is unknown, but estimated lag times from onset to death and from hospitalization to death [27] can be used to understand the range of implied infection fatality rates from the ili surge. we calculated the infection fatality rate implied by the ili surge as a function of the unknown lag from patients being recorded as ili and death, and we repeat this calculation for both the raw and subclinical rate adjusted ili estimates. for a lag of l days from ili reporting to death, the infection fatality rate was estimated by dividing the magnitude of the adjusted or raw ili surge by all new deaths occurring within the dates (2020/03/08 + l, ..., 2020/03/28 + l). a plot of the fatality rate by lag for raw and unadjusted ili surges revealed a large range of fatality rates compatible with the ili surge and highly sensitive to the estimate of lag and clinical rates. one study [27] estimated a median 11.2 days from hospitalization to death and 16.1 days from symptom onset to death. for the raw ili surge estimate, 11 day and 16 day lag times would produce median infection fatality rate estimates of 0.57% and 0.89%, respectively, without adjusting for any subclinical infections; for the subclinical-adjusted ili surge estimate, these lag times would produce median infection fatality rate estimates of 0.19% and 0.29%, respectively. as of april 6, 2020, deaths from sars-cov-2 epidemic were still growing nearly exponentially as evidenced by nearly linear growth on a log y axis. early in the epidemic, estimating exponential growth rates by poisson regression with a log link function produces accurate estimates of the true growth rate [37] , and so we estimated growth rates for the us and italy by poisson generalized linear models predicting new deaths using date as a quantitative explanatory variable. us covid-19 deaths from march 5, 2020 to april 1, 2020, were summed by date to calculate national-level statistics. initially, april 2-5 were included but were found to have anomalously high leverage and were hence excluded from our analysis. we applied the same procedure to covid-19 deaths in italy, focusing on deaths from february 24 until march 12. we used the slope from poisson regression as the estimated exponential growth rate, which yielded a us growth rate of the following seir models [41, 42] combined with persistence of high loads of sars-cov-2 that can be cultured up to 7 days after symptom onset [43] , resulting in our use of a 7.3-11 day 95% credible interval for the infectious period. finally, we parameterized β to ensure i(t) grew with a specified exponential growth rate early in the epidemic. we ran a total of 2,000 simulations for each of the two growth rate distributions (us and italy) analyzed. growth rates were drawn at random from a normal distribution with standard deviation of 0. figure s1 . excess ili for each us state. figure s2 . excess ili correlates strongly with patterns of newly confirmed covid-19 cases. figure s3 . surveillance data from new york city emergency departments. figure s4 : prevalence of sars-cov-2 infections between march 8 and march 28, 2020. figure s5 . syndromic case detection rates by state. figure s6 : estimating the infection fatality rate (ifr) of covid-19 based on the unadjusted ili surge. figure s7 : investigating model sensitivity when ili is modeled without first removing signal from influenza. figure s8 . investigating model sensitivity when seasonal trends in non-influenza ili are identified using an alternative statistical model. 1. an early surge of ili visits across the us. the proportion of patients presenting with ili that could not be explained by influenza or typical seasonal variation (that is, excess ili) is shown for four states (blue line and ribbons represent the posterior median as well as 95% and 50% credible sets; results from all analyzed states are shown in fig. s1 ). ili that could not be attributed to influenza was calculated based on influenza laboratory surveillance data (2019-2020 flu season shown in red, prior seasons are shown in black). a time-series model was used to infer seasonal variation of non-influenza ili. excess ili was then calculated as the difference between non-influenza ili from 2019-2020 and the seasonal baseline of non-influenza ili. excess ili after march 7th is highlighted in darker blue as these data correlated strongly with observed covid-19 case counts ( fig. s2 ). . epidemiological models were either stochastic (simulated via tau-leaping) or deterministic (solved by numerical integration). in addition to our raw estimates of the ili surge size (unadjusted), we provide adjusted prevalence estimates accounting for sub-clinical cases by assuming an 18% asymptomatic rate and a 40% rate of health-care seeking of symptomatic ili patients (adjusted). epidemic trajectories were simulated using an seir model (black lines). the increasing gap between ili prevalence estimates and seir trajectories (orange) suggest the presence of additional factors such social distancing, changes in care-seeking behavior, or heterogeneity in susceptibility or transmission. (c) more generally, the size of the clinical population estimated from ili data imposes a dependence between epidemic doubling time, the clinical rate, and the lag between onset of infectiousness and ili reporting. combinations of these three variables that are consistent (black) or inconsistent (gray) are shown as well as a smoothed estimate of clinical rate as a function of doubling time. china novel coronavirus investigating and research team, a novel coronavirus from patients with pneumonia in china world health organization fundamental principles of epidemic spread highlight the immediate need for large-scale serological surveys to assess the stage of the sars-cov-2 epidemic epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in wuhan, china: a descriptive study clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in surveillance for influenza-united states, 1997-98, 1998-99, and 1999-00 seasons update: influenza activity in the united states during the 2017-18 season and composition of the 2018-19 influenza vaccine do family physicians make good sentinels for influenza? spanish influenza surveillance system, estimating the burden of seasonal influenza in spain from surveillance of mild and severe influenza disease estimating the burden of influenza-associated hospitalizations and deaths in chile during 2012-2014 burden of influenza-associated outpatient influenza-like illness consultations in china detection of excess influenza severity: associating respiratory hospitalization and mortality data with reports of influenza-like illness by primary care physicians influenza surveillance in new zealand in 2005 estimating influenza disease burden from population-based surveillance data in the united states national ambulatory medical care survey: 2016 national summary tables united health foundation, america's health rankings analysis of special data request for information on active state licensed physicians the u.s. health system in perspective: a comparison of twelve industrialized nations estimating clinical severity of covid-19 from the transmission dynamics in wuhan, china self-reported influenza-like illness during the 2009 h1n1 influenza pandemic-united states washington state 2019-ncov case investigation team, first case of 2019 novel coronavirus in the united states epidemic size and probability in populations with heterogeneous infectivity and susceptibility superspreading and the effect of individual variation on disease emergence high contagiousness and rapid spread of severe acute respiratory syndrome coronavirus 2. emerg influenza-like illness, the time to seek healthcare, and influenza antiviral receipt during the 2010-2011 influenza season-united states excess cases of influenza-like illnesses synchronous with coronavirus disease (covid-19) epidemic, france clinical characteristics of covid-19 patients with digestive symptoms in hubei, china: a descriptive, cross-sectional, multicenter study cumulative incidence and diagnosis of sars-cov-2 infection in new york bayesian multinomial logistic normal models through marginally latent matrix-t processes estimating the asymptomatic proportion of coronavirus disease 2019 (covid-19) cases on board the diamond princess cruise ship coronavirus test: what you need to know estimating initial epidemic growth rates incubation period and other epidemiological characteristics of 2019 novel coronavirus infections with right truncation: a statistical analysis of publicly available case data the incubation period of coronavirus disease 2019 (covid-19) from publicly reported confirmed cases: estimation and application presymptomatic transmission of sars-cov-2 -singapore sars-cov-2 viral load in upper respiratory specimens of infected patients temporal dynamics in viral shedding and transmissibility of covid-19 virological assessment of hospitalized patients with covid-2019 competing interests: nh and jds declare that they have no competing interests. adw owns selva analytics llc. data and materials availability: all data associated with this study can be found in the paper or supplementary materials key: cord-263353-4mnsjbib authors: maman, issaka; badziklou, kossi; landoh, essoya d; halatoko, afiwa w; nzussouo, talla n; defang, gabriel n; tamekloe, tsidi a; kennedy, pamela j; thelma, williams; kossi, komlan; issa, zoulkarneiri; kere, abiba b title: implementation of influenza-like illness sentinel surveillance in togo date: 2014-09-20 journal: bmc public health doi: 10.1186/1471-2458-14-981 sha: doc_id: 263353 cord_uid: 4mnsjbib background: the emergence of avian influenza a/h5n1 in 2003 as well as the pandemic influenza a (h1n1) pdm09 highlighted the need to establish influenza sentinel surveillance in togo. the ministry of health decided to introduce influenza to the list of diseases with epidemic potential. by april 2010, togo was actively involved in influenza surveillance. this study aims to describe the implementation of ili surveillance and results obtained from april 2010 to december 2012. methods: two sites were selected based on their accessibility and affordability to patients, their adequate specimen storage capacity and transportation system. patients with ili presenting at sentinel sites were enrolled by trained medical staff based on the world health organization (who) case definitions. oropharyngeal and nasopharyngeal samples were collected and they were tested at the national influenza reference laboratory using a u.s. centers for disease control and prevention (cdc) validated real time rt-pcr protocol. laboratory results and epidemiological data were reported weekly and shared with all sentinel sites, ministry of health, division of epidemiology, who and cdc/namru-3. results: from april 2010 to december 2012, a total of 955 samples were collected with 52% of the study population aged between 0 and 4 years. of the 955 samples, 236 (24.7%) tested positive for influenza viruses; with 136 (14.2%) positive for influenza a and 100 (10.5%) positive for influenza b. the highest influenza positive percentage (30%) was observed in 5–14 years old and patients aged 0–4 and >60 years had the lowest percentage (20%). clinical symptoms such as cough and rhinorrhea were associated more with ili patients who were positive for influenza type a than influenza type b. influenza viruses circulated throughout the year with the positivity rate peaking around the months of january, may and again in october; corresponding respectively to the dry-dusty harmattan season and the long and then the short raining season. the pandemic a (h1n1) pdm09 was the predominantly circulating strain in 2010 while influenza b was the predominantly circulating strain in 2011. the seasonal a/h3n2 was observed throughout 2012 year. conclusions: this study provides information on influenza epidemiology in the capital city of togo. influenza-like-illnesses (ili) is a significant source of morbidity and mortality worldwide [1] . the world health organization (who) estimates that globally influenza accounts for between 3 and 5 million severe cases and 250.000 to 500.000 deaths annually, with most deaths occurring among elderly populations [2] . in temperate regions, ili is reported throughout the year with a marked increase in cases recorded during winter periods [3] . however, in tropical and subtropical regions where viral transmission occurs throughout the year, the data on the burden of influenza-like-illness are limited. nevertheless there is some evidence of a slight increase in cases during the rainy season [4, 5] . the emergence of new highly pathogenic influenza a/ h5n1 viruses in 2003 [6] , their wide circulation in wild and domestic birds and its association with human infections which involves high mortality, has raised global concern about the risk of another influenza pandemic. the emergence of novel human pandemic influenza a (h1n1) in april 2009 [7, 8] and its rapid worldwide spread has motivated the monitoring of influenza and has enhanced preparedness to counter a possible emerging pandemic. in the african region, countries in collaboration with international partners (e.g. who, cdc, namru-3, etc.) put efforts together to establish influenza surveillance capacities as part of the broader strategy for integrated disease surveillance and response (idsr) [9, 10] . while most countries in asia, north america and europe have wellestablished influenza surveillance, few such systems have been established in sub-saharan africa [5, 11] . influenza surveillance helps in understanding the epidemiology and impact of the disease; therefore providing information about seasonality and the groups at high risk of influenza infection. furthermore, the identification and characterization of circulating viruses will help to provide influenza isolates for monitoring changes in viral antigens and the development of vaccines. thus influenza surveillance provides data for pandemic influenza monitoring and planning as well as for decision-making [12] [13] [14] . in togo, the first suspected cases of human avian influenza a/h5n1 were reported between 2007 and 2008 in the maritime region, a few kilometers from the capital city lomé, which has a population of more than 2 millions. between april and december 2009, cases of ili were observed and were suspected to be pandemic influenza given the emergence of the novel human pandemic influenza a (h1n1) pdm09. with no ongoing influenza surveillance, our country was not yet ready to confirm and effectively monitor the severity of the disease. the lack of molecular laboratory technology to detect influenza viruses significantly reduced our ability to manage and control the pandemic. therefore, the ministry of health (moh) in collaboration with the institut national d'hygiène (inh) decided to add influenza to the list of diseases with epidemic potential to be monitored and reported through the idsr program. by april 2010, togo was actively involved in ili surveillance with the support of united state government through the centers for disease control (cdc) and the naval medical research unit-3 (namru-3). this study aims to describe the implementation of ili surveillance and results obtained from april 2010 to december 2012. the ili surveillance system constitutes a collaborative partnership between several togolese institutions within the ministry of health (moh). the departments involved in this surveillance are the division of epidemiology, the national influenza reference laboratory (nil) hosted by the institut national d'hygiène (inh), and the sentinel sites located at the hôpital de bè and military health services in the capital city lomé (figure 1) . a protocol for influenza surveillance was written with the technical support of cdc and namru-3 experts. the ili sentinel surveillance sites were selected based on their accessibility and affordability to patients with low socioeconomic status, the qualifications of medical staff, adequate specimen storage capacity, and an established transportation system to the national influenza reference laboratory (nil). the first site was hôpital de bè, established in april 2010 and located in district n°3. this site was chosen for its geographical location in an area of high population density and high consultation rate. this hospital hosts a pediatric unit and a general medicine ward. the second site established in december 2011, is under the management of the military health services and located in district n°5; its selection was based on the essential role of the armed forces in case of a pandemic and their ability to serve both military and civilian populations. this military site is composed of three units and is attended by military personnel, their families as well as civilians. the two sentinel sites are located in the capital city of togo where approximately 20% of the country's population lives. lomé has two rainy seasons and two dry seasons: the long rainy season (april to june) and the short rainy season (mid-september to october). the long dry season extends from december through march, while the short dry season lasts for two months (july to august). lomé is a coastal city that borders the atlantic ocean to the south, ghana to the west, benin to the east ( figure 1) and is at the crossroad with considerable commercial exchange of goods and movements of population. the who case definition [15] that was used, defined ili as "any person with a sudden onset of fever (≥38°c) and cough or sore throat accompanied or not by general symptoms such as myalgia, prostration, headache or malaise". this definition was used during 2010-2011 period. in 2012, the definition was changed to "any person with a sudden onset of fever (≥38°c) or history of fever and cough or sore throat accompanied or not by general symptoms such as myalgia, prostration, headache or malaise". at both sentinel sites, from monday to friday physicians enrolled the first two outpatients who met the case definition and samples were collected during consultation. the study population included every outpatient, between april 2010 to december 2012, presenting at any of the sentinel sites and meeting the ili case definitions regardless of age or sex and who consented to participate in the surveillance. this population represents a wide cross-section of ethnic and socioeconomic groups. samples collected were nasopharyngeal and oropharyngeal swabs and were placed in the same tube containing a viral transport medium (vtm). they were stored between 2 to 8°c at the bacteriology laboratory of the sentinel site prior to delivery to the nil within 48 hours. before samples were transported, laboratory personnel at the sentinel site conducted quality control checks of information on patients' case report forms. the nil provides the sentinel sites with logistical and material support such as swabs, viral transportation media, cryovials, cool boxes, and ice packs. a quota of 20 samples was targeted from each sites and transported twice a week (tuesday and thursday) to the nil. review meetings with all stakeholders were organized two or three times per year as part of a strategy to in 2007 at sigbéhoué (district des lacs), adétikope (district du golfe) and agodekê (district de zio). in 2008 at agbata (district des lacs). all theses foci were located at few kilometers from the capital city, lomé. improve the surveillance system by identifying strengths and areas of concern during these meetings. socio-demographic (age, sex, date of birth, residential area, travel history) and clinical (date of onset, date of consultation, previous treatment, vaccination status, co-morbidities) data were collected from all patients using a case report form (crf) during consultation. epidemiological data were stored in a single database with laboratory data using a single identification number for each patient. each week, nil provided reports on the distribution of total samples collected, as well as on the number of confirmed influenza cases to the moh, to the sentinel sites, who flunet, cdc, and namru-3. samples collected were analyzed at the national reference influenza laboratory at inh. from every sample, three aliquots were made, two of which were stored at −80°c for external quality control and further analysis (if not subtyped) at namru-3 in cairo, egypt. the other one was kept between 2 to 4°c for rna extraction followed by influenza virus detection by real time rt-pcr within 72 hours after sample reception. for the testing of influenza viruses, rna extraction was performed from 140 μl of naso and/or oro-pharyngeal cells contained in the vtm by using a qiamp viral rna mini kit (qiagen) following the manufacturer's protocol. for detection and typing, it was run on an abi 7300 machine, the real time rt-pcr using the ambion enzyme agpath one-step (ambion, applied biosystems) that amplifies influenza a and b. the u.s. cdc provided the protocol used to detect influenza viruses [16] . in order to determine the quality of the sample, the presence of human ribo-nucleoprotein (rnp) was assessed for each specimen tested. socio-demographic and clinical epidemiological data were entered into a database created using epi-info software version 3.5. data analysis was conducted using spss software version 16.0 (spss inc., chicago, il). student t-test was used for comparison of mean age and the pearson chi-square or fisher exact test to compare laboratory results by age groups and clinical symptoms. the protocol was approved by the moh as part of the monitoring of diseases with epidemic potential and therefore did not require ethical review. verbal consent was obtained from all patients. a total of 955 patients were enrolled in this study. seven hundred and twenty seven (76%) patients were enrolled from the hôpital de bè and 228 (24%) from the military health service site (table 1 ). there was no significant difference in the proportion of females compared to males enrolled in this study (49.9% vs. 50.1%; p = 0.37), and the gender distribution at the two sites was similar. most of patients (65%) were under 15 years of age, while less than 6% were 45 years or older. patients who presented at the hôpital de bè were significantly older (mean age = 17.1 years) compared to those who were seen at the military health service (mean age = 10.6 years with 71% of patients aged less than 5 years) (p = 0.0001). approximately 2% of the patients reported having received influenza vaccination within the last year. of the samples collected for ili surveillance, 236 (24.7%) tested positive for influenza viruses. of these, 136 (14.2%) tested positive for influenza a virus and 100 (10.5%) for influenza b virus ( table 2 table 2 ). the proportion of influenza positive cases varied between different age groups with a higher proportion of influenza a detected in the 15-29 year-old group (20%) than other age groups (p = 0.01; table 3 ). significantly, the pandemic influenza a (h1n1) pdm09 was more often detected in patients aged 5-14 (p = 0.003) and 15-29 (p = 0.03) years than in other age groups. seasonal a/h3n2 was predominant in patients aged 30-44 years (15%; p = 0.0003) and was the only influenza a subtype detected among patients who were 60 years or older. ili was observed throughout every year with irregular peak activity occurring twice annually during the months of; may and november in 2010; may and october in 2011; april/august, october in 2012 ( figure 2) . however, the number of patients/samples enrolled was not consistent. influenza a virus was detected predominantly in 2010 and correlated with the ili peak. the first cases of pandemic influenza a (h1n1) pdm09 were only confirmed in may. during the ili peak, the influenza positivity rate was 26% in may and 30% in november with the pandemic influenza strain, the most subtype detected. the pandemic virus remained predominant between october 2010 and april 2011 (figure 2 ). during the ili peaks in 2011, the influenza positive rate ranged from 20% to 60% with the predominance of influenza b virus activity in may representing 93% of all virus detected (28/30). the second peak was correlated to the seasonal influenza a/h3n2 activity in october with 80% of viruses detected (8/10). from october 2011, there was a co-circulation of influenza type a and type b with low activity of pandemic strain until september 2012 while the seasonal influenza a/h3n2 was detected throughout the year 2012. fever (85%), cough (87%), and rhinorrhea (76%) were the major symptoms for all age groups although sore throat (38%) and headaches (14%) were also recorded ( table 4 ). ili patients who tested positive for influenza were more likely to present with cough (p = 0.004) and headaches (p = 0.03) but were less likely to present with difficulty breathing (p = 0.02) compared to those who tested negative for the influenza virus. among symptoms, only cough was more common in patients testing positive for influenza a than those who tested positive for influenza b (p = 0.003). rhinorrhea was more common in patients with seasonal a/h3n2 than in those with pandemic influenza a (h1n1) pdm09 (p = 0.0004). due to the lack of ili surveillance in togo, there was no information about the epidemiology of ili or influenza viruses until 2010. the first samples collected were processed in may 2010 and the presence of pandemic influenza a(h1n1) pdm09 virus was confirmed in togo one year after the novel pandemic influenza occurred in mexico (april 2009). this is the first report that describes the epidemiology of influenza in togo using data from the ili sentinel surveillance system. during the two and half year period of ili sentinel surveillance, influenza viruses were detected in 236 (25%) of 955 samples. the average percentage positive in this study was higher than the positivity rate observed in 15 other african countries between 2006 and 2010 [17] . however, during the same timeframe, other countries in the temperate climate region: madagascar (40%), morocco [18] . there are several reasons to explain the difference in the percentage positive observed between countries including the temporal distribution of these viruses, the sample collection method, the number of samples collected and the geographical distribution of sentinel sites. most of our data includes post pandemic influenza a (h1n1) pdm09; this is a different picture compared to other african countries (2006) (2007) (2008) (2009) (2010) and to the south american region. the sample collection method was different from one country to another. in our study, we used two swabs (one oro-pharyngeal and one nasopharyngeal) for each enrolled patient and put both swabs in the same cryotube, thereby increasing the viral load and enhancing rt-pcr detection while in some other countries samples were collected either with nasopharyngeal [19] or oro-pharyngeal swab only [18, 20] . in addition, the number of samples tested in most of the other countries was quite high compared to our sample numbers and they were collected from many sites ranging from only 3 to as many as 22. we only used two sentinel sites as our catchment area. the influenza positivity rate varied by year with the highest rate obtained in 2010. since the number of samples collected during this year was very low (87 samples) than the two subsequent years, the rate could be influenced. nevertheless, our percentage positive was similar to that of ghana and rwanda in africa [17] and that of taiwan [21] , but at different periods of time. table 3 distribution of influenza viruses confirmed and ili patients influenza a was predominant in 2010 with pandemic influenza a (h1n1) pdm 09 in our study; this observation was similar to that of other countries in west africa [17] . however, in 2010, the situation was different in other subregions with predominance of influenza b in central/south and north africa [17, 22] and seasonal influenza a/h3n2 in east africa [17] . this difference could be explained by the fact that circulation of pandemic influenza a (h1n1) pdm09 was delayed in west africa and occurred one year after it was predominantly circulating in other african subregions [17, 23] . while two years is not sufficient time for an adequate description of the seasonality of influenza virus transmission, we did observe trends in the lomé commune region. the influenza b virus showed a peak activity during the rainy seasons (may and october) and the pandemic influenza a (h1n1) pdm09 was more frequent during the long dry season while the seasonal a/h3n2 was detected across both seasons. although the seasonality of influenza viruses in african countries is not yet clear, we observed that our trends were similar with the influenza peaks, which have often been associated with the rainy season activity in other tropical countries [24] [25] [26] . in our study, influenza cases were highest (30%) in the 5-14 year age group but also high among other age groups, with lowest percent positive (20.3%) among 0-4 and > 60 years (20.0%). this distribution is consistent with the observation in the study conducted in 15 countries of africa during 2006 to 2010 and in peru [17, 18, 20] in which young children and adults were shown to have the highest influenza viral disease. contrary to our study, a study from venezuela [27] showed higher detection rates in 0-4 year olds. the percentage positive of influenza a was significantly higher in ili patients in the 15-29 age groups. therefore, we found that pandemic influenza a (h1n1) pdm09 was detected significantly among 5-29 years old. this finding is consistent with other studies in the african region [17, 22, 28] that have found that pandemic influenza a (h1n1) pdm09 is most commonly identified in school-age children and young adults. while pandemic influenza a (h1n1) pdm09 appeared more often in older children, seasonal influenza a/h3n2 appeared more likely to infect adults in the 30-44 year-age category. our finding was similar to the observation from a study conducted in peru [18] , where the author found that the seasonal influenza a/h3n2 virus was detected with adults of 45 to 59 years. in conclusion, our results are consistent with studies from africa and south american regions which observed that seasonal influenza a/h3n2 affected a wide range of age groups with predominantly 30 to 60 years old while the pandemic influenza a (h1n1) pdm09 and influenza b virus infections occurred more frequently among older children and young adults. we observed that clinical symptoms were associated with influenza viruses. the influenza type a was more frequently detected than type b in patients presenting with cough and rhinorrhea. this result is consistent with the observation of a study from venezuela [27] . in contrast with this study were pandemic influenza a (h1n1) pdm09 was associated with ili patients with cough, our study showed that ili patients with rhinorrhea were associated with seasonal a/h3n2. our study had some limitations. our data were collected only from 2 sites in an urban area in the capital city of togo and could not be generalized to the population. the percent influenza positivity and age distribution of positive cases were influenced by the low number of samples collected which may be attributable to the non availability of a physician to collect nasal and orpharyngeal swabs. the low number of samples may have also contributed to the high positivity rate. in addition, this low proportion may not be representative to better describe the distribution of influenza cases in the age groups. physician time limitations were due to the time consumption and their workload (number of patients viewed in consultation at the outpatients' department). children were over-represented in this study thus introducing a bias, as the number of adults was not comparable to children under 5 years old. some possible reasons to explain this bias include the fact that the military health service has three units but only the family health care center was functional when added as a site in december 2011. because this unit is a pediatric health center, the high number of children enrolled from this site can account for the observed figures. at the hôpital de bè site, we observed that many patients, mostly adults were not enrolled as ili patients due to the lack of recorded fever (≥38°c), suggesting that we should be considering history of fever as one of the enrollment criteria for ili. this study focused exclusively on outpatients thus limiting our ability to examine the severity of the influenza viruses in hospitalized cases. since our influenza surveillance system had challenges in collecting samples of severe acute respiratory infection (sari) and the lack of data on hospitalizations with patient follow-up we excluded discussions on sari from this study. to improve our influenza surveillance system, it will be necessary to expand the system in other regions by including sari surveillance for severe disease to give a complete picture of influenza burden and epidemiology in our country. these data provided information on the epidemiology of influenza in the lomé commune region in the capital city of togo. some efforts are needed to allow better understanding of influenza burden and epidemiology by expanding sentinel sites in other regions and including sari surveillance. future studies will also be focused on identifying the etiologic agents for the 75% of ili cases that were negative for influenza viruses. retrospective analyses of these stored samples will be necessary to identify other respiratory viruses circulating, including respiratory syncytial virus (rsv), coronaviruses, human metapneumovirus (hmpv) and rhinoviruses. who: the global burden of disease: 2004 update who: influenza fact sheet available the global impact of influenza on morbidity and mortality seasonality of influenza in the tropics: a distinct pattern in northeastern brazil epidemiology and seasonality of respiratory tract virus infections in the tropics who: affected areas with confirmed cases of h5n1 avian influenza who: pandemic (h1n1) epidemiology and factors associated with fatal cases preparedness for highly pathogenic avian influenza pandemic in africa who regional office for africa: integrated disease surveillance in the african region: a regional strategy for communicable diseases influenza in africa: uncovering the epidemiology realities and enigmas of human viral influenza: pathogenesis, epidemiology and control who: recommended composition of influenza virus vaccines for use in the 2013 southern hemisphere influenza season manual for the laboratory diagnosis and virological surveillance of influenza who: cdc protocol of real time rt-pcr for detection and characterization influenza 2009 a (h1n1) pdm virus influenza surveillance in 15 countries in africa influenza-like illness sentinel surveillance in peru influenza surveillance among children with pneumonia admitted to a district hospital in coastal kenya challenges of establishing routine influenza sentinel surveillance in ethiopia nationwide surveillance of influenza during the pandemic (2009-10) and post-pandemic (2010-11) periods in taiwan sentinel surveillance for influenza-like illness, severe acute respiratory illness and laboratory-confirmed influenza in kinshasa, democratic republic of congo pandemic influenza a virus subtype h1n1 circulation in west africa epidemiological and virological influenza survey in dakar, senegal: 1996-1998 influenza in outpatient ili case-patients in national hospital-based surveillance seasonality of influenza in brazil: a traveling wave from the amazon to the subtropics sentinel surveillance of influenza-like illness in two hospitals in influenza viruses in nigeria, results from the first 17 months of a national influenza sentinel surveillance system implementation of influenza-like illness sentinel surveillance in togo we would like to thank all sentinel staff at the hôpital de bè and the military health services for their essential role in the ili sentinel surveillance. we are grateful to the ministry of health and the division of epidemiology for their support and coordination. we would like also express our sincere thanks to the team of the national influenza reference laboratory for their efforts in collecting samples, clinical data and detection by rt-pcr of influenza viruses. we wish to thank mr koffi akolly, field epidemiologist for designing the map of the figure 1 . the influenza sentinel surveillance was successful established with the technical and financial support of cdc and namru-3. we are grateful also to cdc reviewers for their precious analyses and revision of this paper prior it's submission for publication. the authors have declared that no competing interests exist.authors' contributions im contributed to the study design, statistical analyses of data and wrote the paper. kb and edl contributed to the study design, interpreted analysis and review the manuscript. awh and tat contributed in the study design and review the manuscript. tnn, gnd, wt and pjk provide technical advice for study protocol, methodology, revised critically the manuscript for important scientific content and have given final approval for the version to be published. zi and kk were involved in literature review and revising the manuscript. abk contributed to the facilitation of the project, participated in its design, coordination and review the paper. all authors read and approved the final manuscript. key: cord-325794-lir8ht2i authors: kinar, y.; lanyado, a.; shoshan, a.; yesharim, r.; domany, t.; shalev, v.; chodcik, g. title: predicting individual risk for covid19 complications using emr data date: 2020-06-05 journal: nan doi: 10.1101/2020.06.03.20121574 sha: doc_id: 325794 cord_uid: lir8ht2i background: the global pandemic of covid-19 has challenged healthcare organizations and caused numerous deaths and hospitalizations worldwide. the need for data-based decision support tools for many aspects of controlling and treating the disease is evident but has been hampered by the scarcity of real-world reliable data. here we describe two approaches: a. the use of an existing emr-based model for predicting complications due to influenza combined with available epidemiological data to create a model that identifies individuals at high risk to develop complications due to covid-19 and b. a preliminary model that is trained using existing real world covid-19 data. methods: we have utilized the computerized data of maccabi healthcare services a 2.3 million member state-mandated health organization in israel. the age and sex matched matrix used for training the xgboost ili-based model included, circa 690,000 rows and 900 features. the available dataset for covid-based model included a total 2137 sars-cov-2 positive individuals who were either not hospitalized (n=1658), or hospitalized and marked as mild (n=332), or as having moderate (n=83) or severe (n=64) complications. findings: the auc of our models and the priors on the 2137 covid-19 patients for predicting moderate and severe complications as cases and all other as controls, the auc for the ili-based model was 0.852[0.824-0.879] for the covid19-based model 0.872[0.847-0.879].. interpretation: these models can effectively identify patients at high-risk for complication, thus allowing optimization of resources and more focused follow up and early triage these patients if once symptoms worsen. we have search pubmed for coronavirus[mesh major topic] and the following mesh terms: risk score, predictive analytics, algorithm, predictive analytics. only few studies were found on predictive analytics for developing covid19 complications using real-world data. many of the relevant works were based on self-reported information and are therefore difficult to implement at large scale and without patient or physician participation. we have described two models for assessing risk of covid-19 complications and mortality, based on emr data. one model was derived by combining a machine-learning model for influenza-complications with epidemiological data for age and sex dependent mortality rates due to covid-19. the other was directly derived from initial covid-19 complications data. the developed models may effectively identify patients at high-risk for developing covid19 complications. implementing such models into operational data systems may support covid-19 care workflows and assist in triaging patients. since january 2020, the covid-19 pandemic has become a global emergency. healthcare organizations and governments, worldwide, are strained due to shortage of resources and the need to make timely decisions based on very little reliable data. these decisions includewho to test, how to treat positive cases, how to manage social distancing and reach-out to population at risk, contact tracing, and more. many of these decisions could benefit from decision support tools based on emr and additional data sources, such as geospatial information. unfortunately, accurate data-driven tools are still difficult to develop due to the limited availability of covid-19 patients' data with historical emr records. many of the relevant works 1-3 describe risk factors and the tools already developed 4, 5 are based on self-reported information and are therefore difficult to implement at large scale and without patient or physician participation. here, we describe two approaches and tools to assess the individual risk of developing covid-19 complications based on medical records: a model developed by combining a machinelearning approach for influenza-like illness (ili) to be used as a proxy model for covid-19 and a second model using data on covid-19 patients. the models were trained using data from maccabi health service (mhs)a large israeli hmo with a central emr database containing longitudinal data for 2 million active individuals each year between 2010 and 2018. the data included full emr information -demographics (e.g. age and sex), behavioral info (smoking status), vital signs, lab test results, diagnoses and procedures (using the international classification of diseases 9 th version ), medication prescriptions and purchases, and hospital admissions (dates and departments only). all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . since the number of in mhs members who are positive for sars-cov-2 is relatively low, and the data available is biased due to the current limitations of tests and challenges of data collection and curation, we have therefore chosen to test two complimentary approaches. first, we use a proxy model that we derived for identifying patients with high risk of developing complications due to influenza and apply some required adjustments. although influenza and covid-19 are clearly very different diseases 6 , it is already apparent that both diseases have common risk factors for developing complications. however, the initial epidemiological data for covid-19 [china cfr, nyc cfr] already show some major differences between the two diseasesprimarily in the effect of increased age on the risk of complications (which seems much stronger for covid-19) and the much higher risk among men for covid-19 complications and mortality, a trend less evident in influenza (another difference is seasonalitywhich is clear for influenza and less evident for covid-19). following these differences, we modified the ilibased model and forced it to ignore age and sex as risk factors, and then used bayesian correction to add these risk factors using external priors. for the training covid-19ased model, we used information on sars-cov-2 positive individuals aged 19 or above within the mhs population, as well as information regarding hospitalization and in-hospital complications. as an initial prior we used the information based on covid-19 mortality available from china [https://www.worldometers.info/coronavirus/coronavirus-age-sex-demographics/] as proxy for complications probabilities (appendix table 1). fatality rate by sex is given in appendix table 2. due to the over-representation of women among the elderly, we had to replace the 1:1.65 ratio of female-to-male risk with a higher 1:2 ratio per age group, as shown in appendix table 3. a detailed description of our approach to developing a model based on emr data is given elsewhere 7, 8 . for training the ili-based model, a training set of all mhs members at september 1 st of every calendar year who were not vaccinated during the following flu-season. we marked them as cases if they were diagnosed with ili followed by complications (death, hospitalization in internal ward, or severe illness, e.g. pneumoniasee appendix for list of icd-9 codes) within 3 months, and controls if otherwise. bins were matched for age (5yegendar groups) and sex. in addition, we matched for calendar year to avoid biases due to change in collection, registration, or healthcare policy over the period. given the matched set, we generated a large matrix of features per each sample (a sample corresponds to an individual per each relevant year) and applied a process of univariant age-andsex conditioned feature selection on this matrix, and then trained and recalibrated xgboost model 9 using isotonic regression. to combine the prediction of the calibrated model with age and sex priors for complications, we used the following formula where odds is the overall odds of the model's predictions. see the appendix for derivation of the formula. we used the definitions of the israel ministry of health for covid19 complications: moderate (defined as pneumonia, with one of the following: respiratory rate above 30 breaths per minute, all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint respiratory distress, or oxygen saturation below 90%) or severe (pneumonia accompanied by sepsis, shock, ards or death). we then created a vector that of features per each individual, including risk factors and underlying conditions (see appendix). we used xgboost on the features matrix to learn a covid-19 complications predictor based on these features. given that real world data on covid-19 are currently limited, it is difficult to evaluate the performance of our models. we report here several methods we have used to estimate the value of the models. 1. for the covid19-based model, we report the performance (auc) in predicting influenza complications as an initial indication of the value of the models 2. we examined the excess risk of underlying health conditions, compared to information from the cdc [https://www.cdc.gov/mmwr/volumes/69/wr/mm6913e2.htm#f1_down]. for cdc information, we took the proportion of individuals admitted to icu given various comorbidities; for our models, we used the mean prediction over individuals with the corresponding icd-9 codes. evaluating performance of the model on initial covid-19 complications records. for the model directly derived on covid-19 data, we used cross-validation for performance evaluation. lift was evaluated by calculating the average prediction over the population with the underlying conditions, and comparing to the average prediction over a reference population all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint the age and sex matched matrix used for training the xgboost model included, after feature selection, about 690,000 rows and 900 features (compared to about 790,000 rows and 1584 features for the non-matched model). the top 10 important features are given in an appendix. the available dataset included a total 2137 sars-cov-2 positive individuals who were either not hospitalized (n=1658), or hospitalized and marked as mild (n=332), or as having moderate (n=83) or severe (n=64) complications. individuals who were hospitalized but not assigned severity level were excluded. all individuals were linked to their mhs medical record in order to generate the features matrix the auc of the full (non-matched) model for predicting influenza-complication was 0.744, the matched model auc was 0.726. after adjusting for the age and sex priors, the auc for predicting influenza-complications deteriorates to 0.688. in comparing our results to cdc data (table 1), we can see that, although the prevalence of various conditions is quite different between mhs and cdc data, lifts seem similar and correlated, both for the ili-based and the covid19-based models, with the exception of pregnancy. we defined three groups according to the model's predictionhigh risk (top 10%), intermediate (next 15%) and low (bottom 75%). the confusion matrix of our grouping and the mhs status is given in table 2. the overall distribution of sars-cov-2 positives over the three groups is as expected for random infection (ili-based model: 74% for the low risk, 14% for the intermediate risk, and 1% for the high risk; all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint 72%, 15% and 13% respectively for the covid19-based model). the hospitalized population is enriched in high and intermediate risk groups (56% of 479 for the ili-based model, 59% for the covid19-based model). this sensitivity is even higher for the moderate (78% of 83 for both models) and severe cases (84% and 88% of 64). we also compared the performance of our score to using only the priorsthe sensitivity of the severe and moderate cases in the top 25% (high all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint we have described two approaches for assessing risk of covid-19 complications and mortality, based on emr data. one model was derived by combining a machine-learning model for influenza-complications with epidemiological data for age and sex dependent mortality rates due to covid-19. the other was directly derived from initial covid-19 complications data. such models have many potential applications during the covid-19 epidemics, prioritization of tests and antibody testing, follow-up on patients with the disease, decision on hospitalization, reach-out for population at risk when social-distancing and restrictions are gradually lifted, and in possible future outbreaks of the disease, and, hopefully in the near future, prioritization of vaccination. both approaches have many weaknesses, due to the speedy and urgent manner of their derivations. performance evaluation is indicative, at best, of the true performance of the models. a better model will surely be derived once more reliable covid-19 real world data will be available. however, we believe that currently such models can be of great use for health systems and public health entities coping with pandemic. although performance of the covid19-based model seems better than the ili-based model, it is reasonable to suspect due to the small size of the dataset that the latter model is too specific to the mhs and less generalizable compared to the ili-based model. the auc of the ili-based model on the subset of sars-cov-2 positives is the same as the priors only. however, we note a couple of points -first, the significant difference in performance when considering all population, as well as some manual curation of the dataset suggest a possible bias toward older individuals in the definition of covid-19 complications and sars-cov2 positives. second, even though the auc is similar, the ili-based model can all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint identify younger populations at risk of complications, which the priors-only model, of course, cannot. in comparing the two models it is interesting to note that the effect of bmi on the risk for covid-19 complications seems much higher than the risk for influenza complications. this suggests future work of further adjusting the more robust ili-based -model by inserting exterior priors for bmi as well. in comparing our results to cdc data similar lifts both for the ili-based and the covid19-based models, with the exception of pregnancy that was associated with low risk in our model, compared to slightly elevated risk in the cdc data (though based on very few cases). this might be due differences in age and sex distribution between the us and israel populations pregnancy. for all underlying conditions, the covid19-based model showed lower lift compared to the ili-based model despite the model's inherent weaknesses, and due to the clear and urgent needs, the ili-based model was integrated at mhs to support two covid-19 care workflows. first use is triaging testing. with limited resources available for outpatient testing, there is a need to prioritize testing to those individuals at highest risk of complications and mortality from covid-19. the second use is for outpatient virtual management and triage. mhs established a virtual covid-19 management center, occupied by primary care physicians and nurses. the medical staff are the first to contact confirmed covid-19 patients, question them and decide on the appropriate treatment facility based on their symptoms and overall medical assessment. patients can be hospitalized, sent to a special covid-19 care facility, or stay at home. nurses are then following-up on those patients at homecare to continuously assess their condition. a flag was added to patients estimated to be at high-risk, thus allowing optimization of resources and more all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint focused virtual follow up and also helping clinicians to triage these patients if their symptoms worsen. all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. there was no funding for the study. all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. we learn a model for flu complications on an age-matched training set, and evaluate the following probabilities -= ( = | + ) and = ( = | + ì�ì�ì�ì�ì�ì�ì�ì�ì�ì�ì�ì�ì�ì�ì�ì� ); all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . and get the formula in the paper. all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint all rights reserved. no reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted june 5, 2020. . https://doi.org/10.1101/2020.06.03.20121574 doi: medrxiv preprint clinical characteristics of coronavirus disease 2019 in china risk factors associated with acute respiratory distress syndrome and death in patients with coronavirus disease hematologic, biochemical and immune biomarker abnormalities associated with severe illness and mortality in coronavirus disease 2019 (covid-19): a meta-analysis prediction models for diagnosis and prognosis of covid-19 infection: systematic review and critical appraisal clinical course and risk factors for mortality of adult inpatients with covid-19 in wuhan, china: a retrospective cohort study prediction of progression from pre-diabetes to diabetes: development and validation of a machine learning model key: cord-312493-wbhji81g authors: tay, ee laine; grant, kristina; kirk, martyn; mounts, anthony; kelly, heath title: exploring a proposed who method to determine thresholds for seasonal influenza surveillance date: 2013-10-11 journal: plos one doi: 10.1371/journal.pone.0077244 sha: doc_id: 312493 cord_uid: wbhji81g introduction: health authorities find thresholds useful to gauge the start and severity of influenza seasons. we explored a method for deriving thresholds proposed in an influenza surveillance manual published by the world health organization (who). methods: for 2002-2011, we analysed two routine influenza-like-illness (ili) datasets, general practice sentinel surveillance and a locum medical service sentinel surveillance, plus laboratory data and hospital admissions for influenza. for each sentinel dataset, we created two composite variables from the product of weekly ili data and the relevant laboratory data, indicating the proportion of tested specimens that were positive. for all datasets, including the composite datasets, we aligned data on the median week of peak influenza or ili activity and assigned three threshold levels: seasonal threshold, determined by inspection; and two intensity thresholds termed average and alert thresholds, determined by calculations of means, medians, confidence intervals (ci) and percentiles. from the thresholds, we compared the seasonal onset, end and intensity across all datasets from 2002-2011. correlation between datasets was assessed using the mean correlation coefficient. results: the median week of peak activity was week 34 for all datasets, except hospital data (week 35). means and medians were comparable and the 90% upper cis were similar to the 95(th) percentiles. comparison of thresholds revealed variations in defining the start of a season but good agreement in describing the end and intensity of influenza seasons, except in hospital admissions data after the pandemic year of 2009. the composite variables improved the agreements between the ili and other datasets. datasets were well correlated, with mean correlation coefficients of >0.75 for a range of combinations. conclusions: thresholds for influenza surveillance are easily derived from historical surveillance and laboratory data using the approach proposed by who. use of composite variables is helpful for describing influenza season characteristics. influenza infection remains a significant public health problem, resulting in considerable global morbidity and mortality [1] [2] [3] . in temperate regions of australia, seasonal influenza outbreaks usually occur between late autumn and early spring and are associated with an increase in disease burden and utilisation of health service [3, 4] . due to differences in circulating viruses, population immunity and environmental factors, the onset, duration and severity of a season may differ from year to year [2, 5] . ongoing monitoring of influenza is therefore needed to determine the onset and severity of seasons and to monitor changes in disease trends. surveillance which involves laboratory testing can add to data on virus characteristics. influenza thresholds have been developed to indicate a level of disease activity that would signal the start or end of a season or provide an alert to an unusually severe or atypical season. the onset of a season may stimulate diagnosis, enhance case detection, promote awareness of the need for patient cohorting or isolation in hospitals, remind people about vaccination and encourage early prescription of anti-viral medication, especially in vulnerable populations [6, 7] . in the setting of a particularly severe season or pandemic, thresholds may inform the appropriate allocation of resources [8] . many influenza surveillance systems around the world have incorporated the use of thresholds. these include australia, new zealand, europe and united states (us) [9] [10] [11] [12] . methods using a variety of surveillance systems have been developed to establish thresholds for influenza activity. the methods vary in their complexity and can use either short-term or longer historical data to create time-varying or fixed thresholds. there is currently no gold standard or consensus for calculating thresholds. the simplest method uses visual inspection of historical data to create a fixed threshold used throughout the year [13] . other methods include regression models [14] [15] [16] [17] , time series methods [15] , calculation of means and medians [18] [19] [20] and adaptation of industrial control processes such as shewhart charts [21] , cumulative sum (cusum) [15, 18, 22] and the exponentially weighted moving average [23] . the us centre for disease control and prevention (cdc) calculates the baseline for their influenzalike-illness (ili) surveillance by adding two standard deviations to the mean percentage of ili visits during non-influenza weeks for the previous three seasons, with non-influenza weeks defined as periods with less than 2% of the year's total positive specimens for influenza for ≥ 2 consecutive weeks [24] . in victoria, the general practice sentinel surveillance (gpss) for ili has historically relied on thresholds determined by inspection [13] . in 2012 a novel but simple method for defining thresholds was proposed by the world health organization (who) as part of the development of global standards for influenza surveillance [7] . the proposed method aligns several years of historical data on the median week of peak activity and assigns thresholds based on means and standard deviations of aligned data. to our knowledge, this method has not yet been field tested. the aim of this study was to explore the feasibility of the who method for the calculation of influenza thresholds using a range of existing surveillance and laboratory data sources in one surveillance system. we used these data sources to compare the onset, duration and intensity of influenza seasons. victoria is a state with a population of over 5 million people located in the south-eastern part of australia. it has a temperate climate with annual seasons of influenza occurring occur between late autumn (may) and early spring (october). it has a well-established influenza surveillance system that monitors influenza activity using syndromic surveillance of ili presentations to sentinel general practitioners (gp) and a medical locum service; laboratory-confirmed influenza; hospital admissions for influenza; and more recently google flu trends and the influenza complications alert network (flucan), which monitors hospitalised patients from sentinel australian hospitals, including four victorian hospitals [9] . four independent surveillance data sources were used: (i) the victorian gpss, (ii) sentinel data from the melbourne medical deputising service (mmds), (iii) routine laboratoryconfirmed influenza (lab data) from the victorian infectious diseases reference laboratory (vidrl) and the (iv) victoria admitted episode dataset (vaed) for hospital admissions. the gpss is an annual surveillance system for ili and laboratory confirmed influenza that was established in 1993, with laboratory support added in 1998 [25] . surveillance extends from week 18 to 44 each year during the influenza season. the number of participating general practitioners (gp) has varied from 40 to 100 since the scheme's establishment. the ili definition used is based on the nationally agreed case definition of cough, fever (measured or reported) and fatigue [13] . approximately 48% of ili patients seen by sentinel gps were swabbed and of these, influenza virus was detected from an average of 34% (18%-47%) of the swabbed ili patients tested from 2003 to 2011 [9, 26] . an alternative source of community sentinel ili surveillance is the mmds, an out-of-hours medical locum service that covers an approximate 45km radius from central melbourne. gps from the deputising service consult with patients in their own home or aged care facility. the diagnosis made by the attending doctor is recorded electronically and de-identified summary data are available on a password protected website within 24 hours. ili data are extracted weekly based on a previously developed search algorithm [27] . there is no laboratory support for the mmds surveillance which has nonetheless been shown to provide equivalent information to surveillance data from sentinel general practitioners [28] . lab data from vidrl consist of laboratory detections of influenza viruses from all routine respiratory samples sent to vidrl, tested using an in-house respiratory multiplex reverse transcriptase polymerase chain reaction (rt-pcr) test that identifies influenza viruses, adenovirus, picornavirus, respiratory syncytial virus, parainfluenza virus, coronavirus and human metapneumovirus [29] . many of the samples are referred from major adult teaching hospitals in victoria [30] . the vaed is a hospitalisation dataset on all patients admitted to public and private acute care hospitals in the state of victoria. the clinical information coded for each episode of care is based on the international classification of diseases and related health problems, tenth revision, australian modification (icd-10-am). we extracted records containing influenza codes j09-11 in primary or secondary diagnostic fields. in addition, a further two composite variables were created from the product of ili and lab data. these were the gpss composite = proportion of ili cases in gpss x proportion of laboratory samples tests positive for influenza in gpss exploring a who method for influenza thresholds plos one | www.plosone.org mmds composite = proportion of ili cases in mmds x proportion of routine laboratory tests positive for influenza in lab data. while the gpss proportion and laboratory testing are part of the same system, the mmds is independent of routine clinical tests referred to vidrl. however, both mmds and routine vidrl clinical testing focus on older age groups and were matched based on age profiles [30, 31] . the metrics used for threshold calculations were the weekly gpss ili proportion per 1000; the weekly mmds ili per 1000; the proportion of laboratory test positive for influenza using the total number of tests for influenza as the denominator (lab test positive influenza); the weekly proportion of influenza admissions in the population using the mid-year estimated resident population in victoria as the denominator and expressed per 100,000 population [32] ; and the product of weekly ili and lab data per 1000. we preferred the use of test positive influenza to count data to compensate for changes in testing behaviour over time [33] . depending on availability, data were extracted from 2002 to 2011 for gpss and mmds, 2003 to 2011 for lab data and 2005 to 2011 for vaed. for mmds, lab data and vaed, data were complete for all 52 weeks but only weeks 18 to 44 were available for gpss. for vaed, only aggregated data containing admissions of more than five counts were provided to protect privacy and confidentiality of individuals. we defined the three levels of influenza threshold based on the terminology used in the who manual and the existing victorian surveillance thresholds that had been adapted from the united kingdom: seasonal, average and alert [13, 34] . seasonal threshold defines the start and end of an influenza season. the two intensity thresholds, termed average and alert thresholds, describe relative seasonal intensity. the who manual did not prescribe a specific method for determining the seasonal threshold, that is, the start of the season. we used the method of inspection of the complete data for the six datasets to determine the seasonal threshold. for each dataset this was done independently by four of the co-authors (et, kg, am, hk) and differences were resolved by discussion. for the four datasets that provided data for the whole year (mmds, lab data, vaed and mmds composite), we calculated the 95% confidence interval (ci) of the metrics used for each dataset for the period defined as out-of-season, that is, the values below the seasonal threshold, that had been determined by inspection. we also explored the seasonal threshold using the 95th percentile of out-of-season data without assuming data were normally distributed. we then compared the seasonal threshold set by inspection with the 95% ci and 95th percentiles of the average out-of-season values. average and alert thresholds were calculated for each dataset using a variation of the who protocol [7] (figure 1 ). we first determined the median week of peak occurrence using historical data, excluding the pandemic year of 2009 which was atypical from both surveillance and testing perspectives [33] . we then aligned the transmission peaks around the median week of peak occurrence ( figure 1a ) and calculated the weekly mean and standard deviations for each week centred on the median week of peak occurrence ( figure 1b and c) . the who protocol suggests the use of the normal distribution to assign thresholds based on the mean and standard deviation of the aligned data for weekly counts. however, we believed data were unlikely to be normally distributed for all years and tested this by inspection and formally using the shapiro-wilks test for normality for gpss, mmds, lab data and the vaed for each year during season [35] . in addition to the mean and standard deviations, we explored the thresholds using the median and 90th and 95th percentiles. the average threshold was determined by a comparison of the peak weekly mean and median, while the alert threshold was determined by a comparison of the peak weekly upper 90% and 95% ci upper limits with the 90th and 95th percentiles. we also performed log transformation of all datasets and calculated the corresponding geometric mean and 90/95 ci upper limit. once the seasonal thresholds were assigned, we determined the start and end of each season independently for all datasets, each defined as the two consecutive weeks where the seasonal threshold was crossed. we used the average and alert thresholds to categorise the influenza seasons, based on the threshold range of peak seasonal activity, specifically seasonal-average, average-alert or alert. comparisons were made for the onset, duration and intensity of a season across all datasets from 2005-2011 based on data availability. we also compared how all seasons compared to the average season created using aligned historical data described above. finally, to determine how correlated the six datasets were, we calculated the correlation coefficient for each year from 2005 to 2011 for a combination of datasets, from which a mean correlation coefficient and its corresponding 95% confidence limits was derived. data were analysed using microsoft® office excel 2003 and stata version 10.0 (stata corp., college station, tx, usa). the study was approved as a quality assurance project by the melbourne health office of research. gpss and mmds data in this study were collected, used and reported under the legislative authorization of the victorian public health and wellbeing act 2008 and public health and wellbeing regulations 2009. during the study period, the highest number of ili or influenza cases annually ranged from 56 to 208 per week for the gpss; 33 to 164 per week for the mmds; 24 to 135 per week for test positive influenza; and 28 to 204 per week for influenza admissions. testing for normality of the weekly count data for each year suggested no seasonal surveillance data had a classical normal distribution graphically (data not shown). data were consistent with a normal distribution by formal testing for 5/10 seasons in the gpss, for 5/10 seasons in the mmds, for 2/9 seasons in lab data and for 2/6 seasons in the vaed. when data were log-transformed, the number of seasons with normal distribution increased (8/10, 8/10, 5/9 and 3/6 respectively). the threshold parameters from the adapted who method are summarised in table 1 . the median week of peak occurrence for all datasets was week 34 except for the mmds composite and vaed. the values assigned for seasonal thresholds by inspection were similar to the 95% ci upper limit and 95 th percentile of the average out-of-season values for all datasets. for the average threshold, we found the peak mean values for the gpss, test positive influenza and vaed were similar to the median but the peak mean was higher for the mmds and mmds composite due to the high call out proportion in 2003. we therefore use the peak mean to define (table 1) . to set the alert thresholds, the peak 90% ci upper limit was used as we found the parameter to be similar to the peak 95 th percentile across all six datasets (table 1 and figure 2 ). the geometric means and 90% ci upper limits from log-transformed data also produced similar parameters (data not shown). using the gpss dataset as an example, figure 3 compares how an annual season compares against the average season calculated using ten years of historical data. onset, end and duration of influenza season across all datasets. for the seven years where data were available for all datasets, gpss assigned seasons tended to start much earlier for most years compared to other datasets. the use of the gpss composite suggested a later start to the season. season onset according to the vaed generally lagged behind other datasets for most years and was variable for the mmds ( table 2) . for most pre-pandemic years, there was generally good agreement for defining the end of a season across datasets. however, a divergent trend, not reflected in other datasets, was noted in the vaed from 2009 onwards ( table 2) . the vaed assigned seasons ended much later, resulting in a longer assigned seasonal duration. category of influenza season. there was agreement in describing the intensity of influenza seasons in 3/7 years prior to 2009 (table 2 and figure 4 ). during the pandemic year, the season intensity varied according to different data sources and from 2009 onwards, peak seasonal influenza activity was between the seasonal and average thresholds (or seasonalaverage) for all datasets except the vaed. datasets were found to be well correlated, with mean correlation coefficients of >0.75 for a range of combinations (table 3) . correlations between the vaed and other datasets improved once the vaed was aligned with other datasets to correspond to the one week lag in median week of peak occurrence. thresholds for influenza surveillance were easily derived using a simple method proposed by the who. the method was adapted to a non-parametric approach that produced similar findings to the suggested protocol based on the normal distribution. log transformation of the data produced comparable findings to both approaches. comparison of thresholds derived from different datasets revealed variations in defining the start of a season but relatively good agreement in describing the end and intensity of influenza seasons, except in the hospital data after the pandemic year. as the who protocol does not prescribe a method for defining the seasonal threshold, we used the simplest method of visual inspection but showed that the levels were consistent with variation in out-of-season virus circulation. numerous other approaches exist, based on more complex statistical techniques but many of these approaches usually require a pre-determined threshold to be nominated [15, 21, 23] , again often by inspection. in the setting of the average and alert thresholds for our datasets, we used the peak mean values to set the intensity thresholds as per the who protocol after observing the data agreement between the peak weekly means and medians and 90% ci upper limit and 95 percentiles. however, the median and 90 or 95 percentiles may be a more appropriate option when data are not normally distributed and no transformation has been performed. in practice, another point of consideration for the setting of the alert threshold may be a level of influenza activity that corresponds to an increased demand on the health care system [7] . this would be dependent on local health care systems. we also incorporated a two week consecutive rule into the definition of the onset and intensity of a season to reduce the number of false positive signals. based on the seasonal thresholds, we found inconsistencies in defining the start and end of a season across the datasets. given the variations in timeliness of influenza reporting [31] , we would expect the onset of ili surveillance to precede laboratory confirmed influenza and hospital admissions, and that both the ili surveillance systems might coincide with one another. by incorporating a laboratory component to the ili measure, the use of the composite variable appeared to improve the specificity and agreement between vaed and ili surveillance data. the finding is consistent with emerging literature that these composite variables may be a better proxy indicator of influenza incidence than either ili or lab data alone [36, 37] . the use of composite variables in surveillance warrants further investigations. in comparing the intensities of an influenza season, there was good agreement across all datasets, except for the vaed after 2009. the number of hospital admissions coded for influenza has increased both in and out of the influenza season, with the duration of the season prolonged due to the late end signal. these changes were not reflected in the ili or lab datasets. while there may be a number of possible explanations, such as changes in testing behaviours or disease coding, a recent study investigating the increase in out-ofseason influenza in australia suggests a genuine increase in influenza activity, combined with increased testing that occurred following the pandemic [38] . this may reflect an increased awareness of influenza in hospitalised patients among health professionals after the pandemic. additionally, surveillance data at vidrl indicate that approximately 40% of patients with an influenza-like illness were swabbed prior to 2009 [26] but this rose to 70% during the 2009 pandemic [39] and has since remained at about this proportion [9] . the measurement of the intensity a season was based on the peak of influenza activity and whether or not the thresholds were exceeded for two weeks. this reflects only a single dimension of measurement and does not take into account how long the influenza activity remained within a particular category or the rate of increase in the number of cases. for example, a short-term acute rise in influenza cases that marginally exceed the alert threshold does not correspond to a gradual or persistent elevated level of activity that may represent a higher disease burden. finally, when we compared the current parameters to the previous threshold based on seven years of historical ili data from 1994-2000 in victoria, the baseline threshold for the gpss was lower at 2 per 1000 cases compared with the revised threshold of 4 and the alert threshold was higher at 35 per 1000 cases compared with the revised threshold of 24 [13] . these differences indicate the need of regular review of surveillance-derived thresholds. in conclusion, this study has shown that the proposed who threshold protocol is simple to implement and could be easily adapted for any influenza surveillance system with adequate historical data. however, the study was based in a region with a temperate climate, and its application in the tropics would require further work. further exploration of the proposed who method in another temperate region would be of interest. while thresholds are useful as a warning system, they should always be interpreted with other available information. influenza (seasonal) factsheet epidemiology of influenza the annual impact of seasonal influenza in the us: measuring disease burden and costs global influenza seasonality: reconciling patterns across temperate and tropical regions how to deal with influenza: worthwhile surveillance system is in action world health organization (2012) who global surveillance standards for influenza. geneva: global influenza programme, surveillance and monitoring team, world health organization resource allocation during an influenza pandemic virological surveillance: influenza weekly updates centers cdcfor disease control and prevention (2013) flu activity & surveillance: weekly us influenza surveillance report weekly influenza surveillance overview in: control establishing thresholds for influenza surveillance in victoria can syndromic thresholds provide early warning of national influenza outbreaks? methods for monitoring influenza surveillance data a routine tool for detection and assessment of epidemics of influenza-like syndromes in france a statistical algorithm for the early detection of outbreaks of infectious disease epidemic features affecting the performance of outbreak detection algorithms influenza surveillance in europe: establishing epidemic thresholds by the moving epidemic method modelling influenza epidemic -can we detect the beginning and predict the intensity and duration? detection of epidemics in their early stage through infectious disease surveillance do cusums have a role in routine communicable disease surveillance? detecting the start of an influenza outbreak using exponentially weighted moving average charts overview of influenza surveillance in the united states laboratory-supported influenza surveillance in victorian sentinel general practices estimation of influenza vaccine effectiveness from routine surveillance data a medical locum service as a site for sentinel influenza surveillance influenza-like illness surveillance using a deputising medical service corresponds to surveillance from sentinel general practices laboratory diagnosis and surveillance of human respiratory viruses by pcr in h1n1 swine origin influenza infection in the united states and europe in 2009 may be similar to h1n1 seasonal influenza infection in two australian states in a comparison of data sources for the surveillance of seasonal and pandemic influenza in victoria influenza surveillance in australia: we need to do more than count the use of thresholds to describe levels of influenza activity an analysis of variance test for normality (complete samples) improving the estimation of influenza-related mortality over a seasonal baseline predicting the epidemic sizes of influenza a/h1n1, a/h3n2, and b: a statistical method the significance of increased influenza notifications during spring and summer of 2010-11 in australia pandemic influenza h1n1 2009 infection in victoria, australia: no evidence for harm or benefit following receipt of seasonal influenza vaccine in 2009 we gratefully acknowledge the ongoing support of general practitioners and their practice staff participating in the general practice sentinel surveillance system and the continued involvement of the melbourne medical deputising service in influenza surveillance in victoria. we thank the victorian government department of health for provision of hospital admissions data and staff at the viral identification laboratory at vidrl for provision of laboratory data. we acknowledge the world health organization global influenza program for the threshold methodology. sentinel surveillance in victoria is supported by the victorian government department of health. ee laine tay is a post-graduate scholar in the master of philosophy in applied epidemiology program, the field epidemiology training program in australia. conceived and designed the experiments: et kg hk am. analyzed the data: et kg mk hk. contributed reagents/ materials/analysis tools: am. wrote the manuscript: et hk. provided critical revisions to the article: et kg mk am hk. key: cord-305473-w30hsr4m authors: jiang, lili; lee, vernon jian ming; cui, lin; lin, raymond; tan, chyi lin; tan, linda wei lin; lim, wei-yen; leo, yee-sin; low, louie; hibberd, martin; chen, mark i-cheng title: detection of viral respiratory pathogens in mild and severe acute respiratory infections in singapore date: 2017-02-20 journal: sci rep doi: 10.1038/srep42963 sha: doc_id: 305473 cord_uid: w30hsr4m to investigate the performance of laboratory methods and clinical case definitions in detecting the viral pathogens for acute respiratory infections (aris) from a prospective community cohort and hospital inpatients, nasopharyngeal swabs from cohort members reporting aris (community-ari) and inpatients admitted with aris (inpatient-ari) were tested by singleplex real time-polymerase chain reaction (srt-pcr), multiplex rt-pcr (mrt-pcr) and pathogen-chip system (pathchip) between april 2012 and december 2013. community-ari and inpatient-ari was also combined with mild and severe cases of influenza from a historical prospective study as mild-ari and severe-ari respectively to evaluate the performance of clinical case definitions. we analysed 130 community-ari and 140 inpatient-ari episodes (5 inpatient-ari excluded because multiple pathogens were detected), involving 138 and 207 samples respectively. detection by pcr declined with days post-onset for influenza virus; decrease was faster for community-ari than for inpatient-ari. no such patterns were observed for non-influenza respiratory virus infections. pathchip added substantially to viruses detected for community-ari only. clinical case definitions discriminated influenza from other mild-ari but performed poorly for severe-ari and for older participants. rational strategies for diagnosis and surveillance of influenza and other respiratory virus must acknowledge the differences between aris presenting in community and hospital settings. influenza and other respiratory viruses such as respiratory syncytial virus, rhinovirus, parainfluenza virus, adenovirus and human metapneumovirus are common causes of respiratory infections 1 . while often manifesting as a mild illness, these viruses can result in serious complications 2,3 , hospitalisations and deaths 4 . identifying the viral aetiology of respiratory infections has applications in both clinical management and surveillance. early diagnosis may allow timely initiation of appropriate treatment 5, 6 , and where warranted, rapid confirmation of an outbreak can lend itself to control and mitigation efforts for influenza 7, 8 . and while specific therapeutic or preventive measures for most viral respiratory agents (other than influenza) are lacking, diagnosis can still help in ruling out other causes of respiratory illness and facilitate implementation of appropriate infection control measures in healthcare settings 9 . moreover, with increasing concerns about the spread of new and dangerous viral respiratory pathogens such as severe acute respiratory syndrome (sars) and middle east respiratory syndrome (mers) viruses, we need to better appreciate how we can optimally combine clinical information with routine as well more complex and expensive laboratory testing technologies to diagnose and conduct surveillance for unusual pathogens 10 . current approaches to diagnosis and surveillance rely heavily on clinical case definitions and a variety of laboratory assays. however, overlapping symptoms makes existing clinical case definitions for respiratory infections inadequate for diagnosis of specific infections 11, 12 . reverse transcriptase-polymerase chain reaction (rt-pcr), which detects viral nucleic acid by use of amplification techniques, is now considered as the gold standard assay for detection of respiratory viruses, and also has the advantage of short turn-around times as compared to scientific reports | 7:42963 | doi: 10 .1038/srep42963 methods based on virus culture and isolation 13 . moreover, multiplex rt-pcr (mrt-pcr) assays, which allow for rapid detection of multiple types of known viral agents, are now widely available 14 . studies have evaluated the performance of pcr-based assays for diagnosing respiratory viruses [15] [16] [17] , but were mostly restricted to either outpatients or inpatients, or to at-risk groups such as the elderly and young children. moreover, large-scale pathogen detection technologies, like the genome institute of singapore (gis) pathchip, have recently been developed 18 , and there is also a need to rationalize how these could be integrated with routinely available assays. our study had two key objectives. firstly, we investigated the performance of the singleplex rt-pcr (srt-pcr) and mrt-pcr, as well as explored the use of the gis pathchip for the detection of viral respiratory pathogens for mild and severe acute respiratory illness (ari). secondly, we evaluated how well clinical case definitions might differentiate influenza from other causes for mild and severe aris. we collected samples and clinical data from both a community cohort and adult inpatients to address the first objective while we combined this with data from mild and severe influenza cases prospectively recruited during the influenza a(h1n1)pdm09 pandemic of 2009 to support the second objective. of 507 participants from the community enrolled into a cohort, 99 reported having ari episodes. we also enrolled 142 inpatients from tan tock seng hospital (ttsh), and retrieved 128 influenza cases from a previous study (historical-flu). community participants with ari episodes did not differ significantly from the underlying cohort (table 1) . compared to community participants with ari, inpatients were older, more likely to be male and to have chronic medical conditions. similarly, historical-flu participants with medical indications were older, and more likely to have chronic medical conditions as compared to those who were admitted for public health indications. historical-flu participants with medical indications were fairly similar to the inpatients although they were slightly younger and less likely to have copd and heart disease. one hundred and thirty episodes were reported from 99 community participants (community-ari) with 76, 16, 6 and 1 individuals reporting 1, 2, 3 and 4 episodes respectively. for the inpatients, there were 145 episodes with only 1 individual having 2 and another having 3 episodes (inpatient-ari). viruses detected by srt-pcr and mrt-pcr in community-ari and inpatient-ari episodes. on using srt-pcr with mrt-pcr, 53.8% of the community-ari episodes were diagnosed as panel virus positive (i.e. positive for at least one of the viruses on the mrt-pcr panel used), including 10 (7.7%) influenza virus positive episodes (8a(h3n2), 1a(h1n1)pdm09 and 1 influenza b) and 60 (46.2%) non-influenza panel virus positive episodes. two viruses were more common than influenza in community-ari, with 37 (28.5%) episodes positive for rhinoviruses and 13 (10.0%) for coronaviruses (table 2 ). there were no co-infections detected by mrt-pcr in either community-ari episodes or influenza negative inpatient-ari. however, 5 influenza positive inpatient-ari episodes were positive for more than one pathogen, including 4 dual-pathogen (influenza a + influenza b [2] , influenza a + rhinovirus, and influenza b + respiratory syncytial virus) of the 138 and 207 samples from community-ari and inpatient-ari episodes, 18 and 113 respectively were for influenza positive episodes (where multiple samples were collected for each episode). the time between episode onset and the first sample for inpatient-ari was longer than for community-ari (median of 5 days vs 1 day respectively, p < 0.001); however, no such difference was observed between influenza versus other ari episodes either for community-ari or inpatient-ari (p = 0.387 and p = 0.481 respectively). we compared the detection of influenza and non-influenza panel viruses by days post-onset. mrt-pcr detected influenza in only 70.8% of the samples that were positive for influenza by srt-pcr. for influenza positive episodes, the proportion positive by either assay decreased as days post-onset increased for both the community and inpatient samples (fig. 1a ,b). the srt-pcr ct values also increased with days post-onset in both groups, but the ct value for community samples increased more sharply than for the inpatients. the gee model suggested that inpatient-ari had a higher starting ct value than community-ari (p = 0.002, table 3 ). ct value increased as days post-onset increased (p < 0.001), but the increase in ct value with every additional day post-onset for inpatient-ari was lesser than that for community-ari (p < 0.001 for interaction term between participant type and days post episode onset). for non-influenza panel virus episodes, changes in the proportion of positive samples showed no consistent pattern by days post-onset in both community-ari and inpatient-ari (fig. 1c ,d). amongst 131(no samples were sent for pathchip analysis for 4 pcr negative inpatient-ari episodes) pcr negative episodes, pathchip did not detect any viruses in 98 episodes. in 6 episodes, multiple viruses (both respiratory and non-respiratory) were detected; these were excluded from further analysis. 10 episodes were positive to mrt-pcr panel virus; all were from community-ari, with 8 positive for rhinovirus, 1 for influenza a and another 1 for metapneumovirus. 13 more episodes (11 community-ari, 2 inpatient-ari) were positive for viruses including coxsackievirus (1), human enterovirus(9), human parainfluenza virus 4 (1) and influenza c virus (2) where ari is a recognised presentation 19 . 4 episodes had only non-ari viruses detected (human herpesvirus 5, human papillomavirus, molluscum contagiosum virus and human t-lymphotropic virus 2). the pathchip results increased the proportion of community-ari episodes positive for respiratory viral pathogens from 53.8% to 70.0%, and from 63.8% to 78.7% for episodes meeting febrile respiratory illness (fri) criteria (i.e. ari with self-reported fever, regardless of body temperature measurement). due to the small number of influenza positive episodes in community-ari, we combined community-ari and inpatient-ari episodes with influenza episodes identified from a previous study (described further in supplementary material). clinical features of historical-flu admitted for public health indications were fairly similar to influenza cases identified amongst community-ari, which justified grouping them together as mild-ari (i.e. community-ari + historical-flu, public health indications); likewise historical-flu with medical indications for admission was fairly similar to table 2 . non-influenza mono-infection episodes in community-ari and inpatient-ari episodesvirus identified. a as proportion of all 130 community-ari episodes. b as proportion of 60 community-ari episodes positive for non-influenza viruses. c for inpatient-ari, this excludes 5 episodes where there was a co-infection between influenza and non-influenza viruses on mrt-pcr. d as proportion of 24 inpatient-ari monoinfection episodes positive for non-influenza viruses. e includes hcov 229e/nl63 and hcov oc43. influenza from inpatient-ari, and were designated severe ari (i.e. inpatient-ari + historical-flu, medical indications, also see supplementary table s1 ). none of the respiratory symptoms assessed were significantly more common in influenza, and for mild-ari, sore throat and runny nose were significantly less common in those testing positive for influenza (table 4 ). however, for both mild and severe ari, fri and influenza-like illness (ili) table 3 . outcome of the gee model a evaluating the factors on the dynamics of ct values. a generalized estimating equation model which included participant type (community cohort or inpatient), days post episode onset, age, gender, comorbidities (comparing those who reported any comorbidities with those who didn't report any comorbidities, with the comorbidities included being diabetes, asthma, copd, heart disease, cancer and other significant conditions), as well as the interaction term between participant type and days post episode onset. case definitions as well as temperature cut-off points showed significant discriminatory value for influenza over non-influenza episodes or any panel viruses positive over viruses negative episodes. figure 2a shows that in mild-ari, while only 7.7% was influenza positive by srt-pcr, this rose to 19.3%, 35.2% and 32.4% for fri, ili-u (us centers for disease control ili definition) and ili-w (world health organisation ili definition) respectively, with lr+ > 5 for ili case definitions and temperature cut-off points ≥ 37.8 °c. while using ili-w increased the proportion positive for any mrt-pcr panel viruses to 73.0%, this represented a relatively small improvement over the 53.8% positive in all mild-ari, with lr+ of only 2.3 (fig. 2b) . for severe-ari, use of more specific case definitions also improved the likelihood of being positive for influenza and viral respiratory infection, but the lr+ only ranged from 1 to 2.5. notably, for severe-ari, ili-w performed best, with 22.1% and 41.6% of such episodes positive for influenza and any panel viruses respectively (fig. 2c,d) . we performed a sensitivity analysis excluding the historical-flu cases and the results were essentially the same other than for the wider confidence intervals (due to the reduced sample size, see supplementary fig. s1 ). there were insufficient numbers of mild-ari with age ≥ 60 years for age-stratified analysis, but influenza was significantly more likely than non-influenza episodes to meet fri, ili-u and ili-w criteria for both mild-ari and severe-ari in those aged < 60 years (table 5 ). however, case definitions had poorer discriminatory value in severe-ari and older ages. for instance, in those aged < 60 years, severe influenza was significantly more likely than mild influenza to meet ili-w criteria (68.6% vs 48.6% respectively, p = 0.018), but non-influenza causes of severe-ari were also more likely to have febrile presentations, with 32.6% meeting ili-w criteria (vs 8.9% for non-influenza mild-ari, p < 0.001). in addition, comparing severe influenza between age groups shows that episodes in older individuals were also significantly less likely than to meet ili criteria (e.g. for ili-u, 77.1% for severe-ari in those aged < 60 years versus only 51.3% for those aged ≥ 60 years, p = 0.010). our study concurrently assessed the role of routine laboratory diagnostics, and usefulness of the novel pathchip platform as well as ili case definitions in identifying respiratory virus infection in a community cohort and hospital inpatients from a broad range of age groups (6 to 81, and 20 to 89 years respectively), to reflect what may be encountered in either community or primary care (mild-ari) as well as tertiary care settings (severe-ari) in a tropical environment with less distinct seasonal patterns. our study clarifies the role of singleplex and multiplex rt-pcr in the respective populations. we observed imperfect sensitivity of mrt-pcr for influenza (70.8%) as compared to srt-pcr, but better performance was reported in two outpatient studies (91-96%) 15, 20 , possibly related to the longer time between onset and sample collection for inpatients in our study, as the recovery of samples positive for influenza and sensitivity of mrt-pcr for influenza was dependent on the time between onset and sample collection. post episode onset, the proportion positive for influenza decreased (ct value increased) faster in community-ari as compared to inpatient-ari, but no such pattern was observed for non-influenza panel viruses; likewise, others have found slower viral clearance in inpatients compared to outpatients for influenza a(h1n1)pdm09 21 , and a lack of association between days post-onset and ct values for adenovirus, human metapneumovirus, parainfluenza virus 1-3 22 . pcr-based assays may thus retain greater value for detecting influenza amongst inpatients that present late than in the community where episodes presenting more than 7 days post-onset would likely be influenza negative. our study also explored the application of newer platforms like the pathchip for detecting respiratory virus infections. in contrast to multiplexed pcr-based technologies which rely on a set of primers that target a finite number of specific pathogens, the pathchip is designed to detect a wide array of pathogens simultaneously by detecting signatures in the pathogen genome sequences. while costly, our results suggests that it has the potential to complement existing technologies as a diagnostic tool. simões et al previously evaluated the diagnostic value of the pathchip using paediatric nasal wash samples and reported variable sensitivity ranging from 30.8% to 95.0% and reasonable specificity from 88.4% to 100% 18 . in our study, the pathchip was a valuable addition in community-ari where it substantially increased the proportion positive for respiratory etiological agents, so that only about 20% of febrile episodes did not have a viral aetiology identified. notably, the vast majority of additional viruses detected were those known to cause ari. it may thus help in ruling out more sinister causes of ari if added to mrt-pcr for surveillance in community settings; it may also help detect unexpected but dangerous infections such as sars and mers-cov following further validation on actual patient samples with such infections. however, due to the high cost of the assay (about four times mrt-pcr) and the imperfect sensitivity for some pathogens (as low as 30.8% for adenovirus), we opted to test only the rt-pcr negative samples with the pathchip. as such, we are unable to comment on its sensitivity for samples from adults, which may be inferior to the previously published results which used samples from paediatric subjects, who are known to have higher viral loads for some viruses 23, 24 . however, given its current cost, variable sensitivity and slower turn-around time as compared to mrt-pcr (about 20 hours for pathchip), we believe this technology is most appropriately used in the way we designed our study, which is to detect additional respiratory viral pathogen positive episodes as an adjunct to more widely available mrt-pcr panels designed to cover the most common pathogens from community-based samples. finally, our study highlights how the same clinical case definitions, which have been used in both community 11 and inpatient settings 25 , may actually perform differently in the two settings for distinguishing influenza from other respiratory infections, identifying influenza cases, and increasing the yield of diagnostic assays. we found that, while respiratory symptoms themselves are poor in distinguishing influenza from other ari causes, clinical case definitions using fever or temperature cut-off points demonstrated good discriminatory value in our mild-ari episodes, which should reflect what can be anticipated in primary care settings. however, for ari severe enough to require hospitalisation, the case definitions had poorer discriminatory value. this was partly because, while a good majority of older individuals (87.2%) with severe influenza had self-reported fever (as in fri), only about half met the temperature criteria used for ili case definitions ( table 5 ). the use of such high temperature cut-off points would hence substantially reduce sensitivity for detecting influenza in the elderly. also, non-influenza episodes in ari severe enough to require hospitalization are more likely to also have a higher temperature than non-influenza episodes in mild ari; this further reduces the ability of ili case definitions to distinguish influenza from other respiratory causes in the inpatient setting. the choice of an appropriate case definition hence depends on the objectives and setting of the application. if the aim is to diagnose as many influenza cases as possible (e.g. for identifying patients for antiviral treatment or outbreak management), ili case definitions have inadequate sensitivity, particularly in older age groups. however, the discriminatory value of ili criteria finds application when aiming to reduce background noise during syndromic surveillance, either in community type settings or even a healthcare worker population 26 . the ili case definitions are also useful for optimizing the yield of influenza viruses from samples tested, and particularly efficient in settings where milder ari presentations are encountered (fig. 2a) ; and if the aim is to identify all types of respiratory viruses for surveillance, ili criteria would also modestly improve the probability of obtaining a positive result as compared to sampling all ari in both mild as well as severe presentations (fig. 2b,d) . a key limitation we acknowledge is the less than ideal number of influenza infections identified through our community cohort, which led us to supplement our concurrent community and inpatient studies with data from our older prospective study of influenza cases. while not ideal, supplementary table s1 suggests that influenza cases from this historical dataset which were classified as mild were reasonably similar to those from the community cohort, and likewise those classified as severe were similar to those from the later inpatient study. a sensitivity analysis without the influenza cases from our historical dataset gave a similar result. however, even with these additional influenza cases, the numbers remained inadequate to assess the performance of the case definitions for mild influenza in older age groups, and we are also unable to assess performance in paediatric subjects. secondly, the panel of viruses tested on the multiplex pcr assay was not comprehensive; while this was supplemented by the pathchip, there is no means of ascertaining what proportion of the remaining ari episodes are truly not due to an infectious viral aetiology. we also recognise that our study population, in particular the community cohort, may not be representative of the general community, which was further complicated by fluctuations in the rate at which participants notified us of ari episodes, as well as intra-seasonal variations within each year and sporadic outbreaks of particular respiratory pathogens 27 . scientific reports | 7:42963 | doi: 10.1038/srep42963 the performance of various technologies and case definitions for viral respiratory pathogens presenting with ari differs substantially between community and hospital-based settings. pcr-based assays may still be relevant for detecting influenza amongst inpatients that present late but less so for community ari episodes with delayed presentations. the pathchip may add value for respiratory virus detection in samples negative by multiplex rt-pcr, but only for community-ari. finally, us-cdc and who ili case definitions had similar performance for mild ari, but performed less adequately amongst presentations severe enough to require hospitalisation, including in older individuals. rational strategies for diagnosis and surveillance of influenza and other respiratory viruses must acknowledge the differences between these two populations. study population. we recruited a community cohort and inpatients from ttsh between april 2012 and december 2013. the community cohort was recruited by contacting community-dwelling adults who participated in pre-existing prospective cohort studies conducted by the national university of singapore (nus) saw swee hock school of public health. consenting individuals were then enrolled through home visits, where we also opportunistically recruited other members of the household, including children. at enrolment, participants contributed demographic and health information through a baseline interview at enrolment, and then followed-up for up to 1.5 years, with instructions to notify the study team within 24 hours on developing ari symptoms (community-ari). once notified, research staff would obtain nasopharyngeal swabs and symptom data from the participant on the next working day. for inpatients, on each working day, we would screen through the list of adults admitted to ttsh within the last 72hrs who had undergone routine diagnostic testing (by srt-pcr) for influenza. we then enrolled and collected nasopharyngeal swabs from consenting inpatients fulfilling the same ari criteria (inpatient-ari) used for the community cohort. to better characterise influenza, we intentionally oversampled influenza to obtain a ratio of approximately 1 influenza positive to 2 influenza negative patients. additional swabs were taken on days 3 to 5 and days 7 to 9 post-onset (community-ari) or post-enrolment (inpatient-ari) for influenza positive episodes where possible. in addition, we retrieved historical influenza data from a prospective study of admissions to ttsh testing positive for influenza between may 2009 and september 2009 (historical-flu). this included ari patients with medical indications who were admitted alongside ari cases referred to ttsh for public health indications (clinically suspected to have influenza a(h1n1)pdm09 based on epidemiological risk factors like travel and contact history). ethics approvals were obtained from nus institutional review board and the national healthcare group (singapore) domain specific review board, in accordance with relevant guidelines and regulations. a written informed consent was signed by each study participant. laboratory analysis. three assays were performed. srt-pcr and mrt-pcr assays were conducted at the nphl on all available samples while pathchip assays were performed at gis using rt-pcr negative samples. srt-pcr. this assay was only used to detect influenza virus types a and b at a higher sensitivity than might be achieved using multiplexed pcr assays, and to further subtype influenza a positive samples. the protocols had been adopted from the studies by spackman et al. 28 and krafft et al. 29 and, respectively, with cycle threshold (ct) values documented for positive samples (samples with a ct value less than 40 were considered as positive). influenza a positive specimens were then subtyped with subtype specific primers and probes that targeted at haemagglutinin (ha) gene of a(h3n2)3, and ha and nucleoprotein (np) genes of a(h1n1)pdm09 30 pathchip. this assay was used to detect additional viruses beyond what was covered by the primers within the mrt-pcr panel used. for each sample, the cdna was amplified from extracted nucleic acid. the novel platform from the genome institute of singapore (gis) then automatically detects which pathogens' recognition signatures are present, based on a proprietary algorithm constructed based on genetic sequences of viruses clinically relevant to humans (downloaded from the ncbi taxonomy data-base, http://www.ncbi.nlm.nih.gov/taxonomy/ taxonomyhome.html/) 18 to evaluate how the gis-pathchip might add to detection above routinely available mrt-pcr assays, viruses were grouped into panel viruses (i.e. virus group represented in seegene rv12), non-panel viruses for which ari is a common presentation 19 and non-ari viruses. in our analysis on the combined performance of clinical case definitions and laboratory assays, we grouped all community-ari episodes (none of which required hospitalisation) with historical-flu admitted for public health indications; these were designated as "mild-ari" since these would ordinarily not be sufficiently severe as to require hospitalisation. inpatient-ari was grouped with historical-flu cases admitted for medical indications ("severe-ari"). this increased the number of influenza cases available for evaluating the discriminatory value of clinical parameters and case definitions, which were: • ari: episode with acute onset with any key respiratory symptoms including cough, shortness of breath, sore throat, or runny nose. • febrile respiratory illness (fri): ari with self-reported fever, regardless of body temperature (t) measurement. • influenza-like illness defined by centers for disease control and prevention of the united states of america (ili-u): fever ≥ 37.8 °c together with cough and ⁄or sore throat in the absence of a known cause other than influenza 31 . • influenza-like illness defined by world health organization (ili-w): fever of ≥ 38 °c plus cough with onset within the last 10 days 32 . the proportion of episodes fulfilling different criteria was compared by viral pathogen status (influenza viruses, non-influenza panel viruses and negative), and we then investigated the discriminatory performance by calculating the positive likelihood ratio (lr+ ) and its 95% confidence intervals (ci) for positive detection of influenza and any viral infection. this was also done to investigate the effect of age (age < 60 vs ≥ 60 years). estimates of lr+ requires assumptions on the anticipated prevalence of influenza. for community-ari, this was assumed to be what was obtained in ari samples from the community cohort. for inpatient-ari, since influenza positive inpatients were intentionally oversampled (crude_flu), the proportion positive for influenza was estimated based on the proportion positive for influenza from routinely ordered tests of ari admissions to ttsh (adjusted_flu), which was then applied to derive the adjusted (adjusted_non-flu) from the crude proportion (crude_non-flu) positive for non-influenza viruses detected in mrt-pcr using the following formula: adjusted non flu crude non flu (1 adjusted flu)/(1 crude flu) all analyses were conducted using r version 3.0.1 (r foundation for statistical computing, vienna, austria). respiratory viral threats severe complications in influenza-like illnesses impact of influenza and other community-acquired viruses the global burden of respiratory disease early administration of oral oseltamivir increases the benefits of influenza treatment treat or test first? decision analysis of empirical antiviral treatment of influenza virus infection versus treatment based on rapid test results summer influenza outbreak in a home for the elderly: application of preventive measures oseltamivir ring prophylaxis for containment of 2009 h1n1 influenza outbreaks influenza in the acute hospital setting better tests, better care: improved diagnostics for infectious diseases performance of case definitions for influenza surveillance clinical symptoms cannot predict influenza infection during the 2013 influenza season in bavaria molecular diagnosis of respiratory virus infections advances in diagnosis of respiratory virus infections influenza and other respiratory virus infections in outpatients with medically attended acute respiratory infection during the 2011-12 influenza season comparison of multiplex pcr assays and conventional techniques for the diagnostic of respiratory virus infections in children admitted to hospital with an acute respiratory illness performance of diagnostic tests to detect respiratory viruses in older adults pathogen chip for respiratory tract infections detection of influenza virus infection using two pcr methods viral load dynamics in adult patients with a(h1n1)pdm09 influenza association of the ct values of real-time pcr of viral upper respiratory tract infection with clinical severity, kenya correlation of pandemic (h1n1) 2009 viral load with disease severity and prolonged viral shedding in children virus shedding after human rhinovirus infection in children, adults and patients with hypogammaglobulinaemia comparison of severely ill patients with influenza a(h1n1)pdm09 infection during the pandemic and post-pandemic periods in singapore use of healthcare worker sickness absenteeism surveillance as a potential early warning system for influenza epidemics in acute care hospitals seasonal trends of viral respiratory tract infections in the tropics development of a real-time reverse transcriptase pcr assay for type a influenza virus and the avian h5 and h7 hemagglutinin subtypes evaluation of pcr testing of ethanol-fixed nasal swab specimens as an augmented surveillance strategy for influenza virus and adenovirus identification who|who information for molecular diagnosis of influenza virus-update overview of influenza surveillance in the united states |seasonal influenza (flu)| cdc. available at who|who surveillance case definitions for ili and sari. who available at we would like to acknowledge funding from the national medical research council of singapore (ppg10-09). supplementary information accompanies this paper at http://www.nature.com/srep the authors declare no competing financial interests. key: cord-289017-vwye3pk9 authors: comach, guillermo; teneza-mora, nimfa; kochel, tadeusz j.; espino, carlos; sierra, gloria; camacho, daria e.; laguna-torres, v. alberto; garcia, josefina; chauca, gloria; gamero, maria e.; sovero, merly; bordones, slave; villalobos, iris; melchor, angel; halsey, eric s. title: sentinel surveillance of influenza-like illness in two hospitals in maracay, venezuela: 2006–2010 date: 2012-09-11 journal: plos one doi: 10.1371/journal.pone.0044511 sha: doc_id: 289017 cord_uid: vwye3pk9 background: limited information exists on the epidemiology of acute febrile respiratory illnesses in tropical south american countries such as venezuela. the objective of the present study was to examine the epidemiology of influenza-like illness (ili) in two hospitals in maracay, venezuela. methodology/principal findings: we performed a prospective surveillance study of persons with ili who presented for care at two hospitals in maracay, venezuela, from october 2006 to december 2010. a respiratory specimen and clinical information were obtained from each participant. viral isolation and identification with immunofluorescent antibodies and molecular methods were employed to detect respiratory viruses such as adenovirus, influenza a and b, parainfluenza, and respiratory sincytial virus, among others. there were 916 participants in the study (median age: 17 years; range: 1 month – 86 years). viruses were identified in 143 (15.6%) subjects, and one participant was found to have a co-infection with more than one virus. influenza viruses, including pandemic h1n1 2009, were the most frequently detected pathogens, accounting for 67.4% (97/144) of the viruses detected. adenovirus (15/144), parainfluenza virus (13/144), and respiratory syncytial virus (11/144) were also important causes of ili in this study. pandemic h1n1 2009 virus became the most commonly isolated influenza virus during its initial appearance in 2009. two waves of the pandemic were observed: the first which peaked in august 2009 and the second higher than the preceding that peaked in october 2009. in 2010, influenza a/h3n2 re-emerged as the most predominant respiratory virus detected. conclusions/significance: influenza viruses were the most commonly detected viral organisms among patients with acute febrile respiratory illnesses presenting at two hospitals in maracay, venezuela. pandemic h1n1 2009 influenza virus did not completely replace other circulating influenza viruses during its initial appearance in 2009. seasonal influenza a/h3n2 was the most common influenza virus in the post-pandemic phase. acute respiratory infection (ari) remains a leading cause of global burden of disease, and is the second most common cause of illness worldwide, with an annual global incidence exceeding 400 million [1] [2] [3] . a prerequisite of public health planning to reduce global disease burden from ari is to examine data on its epidemiology in order to better define environmental factors as well as target populations for preventive interventions [4] . respiratory viruses are predominant causes of aris, and the epidemiology of acute viral respiratory illnesses in developed countries with temperate climates has been well-characterized [5] [6] [7] . in countries such as the united states, children have been shown to carry a large burden of viral respiratory diseases [5] . recent prospective studies, which utilized more sensitive methods for detecting respiratory viruses such as multiplex polymerase chain reaction (pcr), have similarly demonstrated that the highest rates of viral respiratory infection occur among children and the frequency of infection tends to decrease with age due to increasing acquired immunity [8] . respiratory syncytial virus (rsv), influenza virus, parainfluenza virus, and rhinovirus have long been identified as common causes of ari [9] . recent improvements in molecular detection techniques have allowed the identification of multiple new respiratory viruses such as human metapneumovirus (hmpv), human bocavirus (hbov) and human coronavirus nl63 [8] . while the body of literature describing the epidemiology of acute viral respiratory diseases in developed countries has rapidly expanded, knowledge of the distribution of these diseases in regions such as tropical south america remains limited. influenza viruses are among the most impactful acute respiratory pathogens in terms of morbidity and mortality. despite developed public health intervention programs, the estimated annual average number of influenza-related hospitalizations in the united states exceeds 200,000, and 36,000 deaths are attributable to influenza infections yearly [10, 11] . information on the contribution of influenza viruses to the global burden of disease due to acute respiratory illness is incomplete. data on the epidemiology of influenza viruses in developed countries are derived from multiple sources to include laboratory-based surveillance, sentinel surveillance, as well as hospitalization and outpatient records. in developing countries, where resources are sparse, sentinel surveillance methods may be more readily accessible and more cost-effective than laboratory-based or population-based surveillance for determining the viral etiology of influenza-like illness (ili) in these regions. better identification of the viral causes of ili will enable clinicians in resource-limited settings to appropriately treat and manage patients; more importantly, it will allow public health officials to formulate more effective prevention and control strategies, including monitoring of influenza vaccine efficacy in their communities [12] . studies on the epidemiology of ili in the tropical south and central american countries of peru, brazil, ecuador, nicaragua, honduras, and el salvador have been published [13] [14] [15] [16] . a prospective study of adults with ili in sao paulo, brazil, revealed that while influenza viruses were the predominant cause of ili, rhinoviruses and other respiratory viruses were detected in 19.6% and 13.7% of subjects, respectively [14] . this observation illustrated that a significant proportion of patients who are clinically diagnosed with influenza virus infection may have symptoms indistinguishable from other respiratory viruses. in a prospective study of ili in ecuador, the regional distribution of influenza virus infections varied; a higher detection rate of influenza a occurred in quito, located in the highlands where the level of absolute humidity is lower, whereas influenza a detection rate was lower in the coastal city of guayaquil, which has a more humid tropical climate [15] . additionally, an expanded sentinel surveillance of ili conducted at 31 health centers and hospitals located in 13 peruvian cities showed that the distribution of ili-causing viruses varied by region [13] . these studies illustrate the importance of conducting baseline and continuing ili surveillance in other countries of central and south america because ili can be caused by pathogens other than influenza viruses. the distribution of respiratory infections may vary within the region or within a particular country, depending on climate and topography. in venezuela, ari is the primary cause of weekly notifiable diseases registered by the national epidemiological surveillance system (ness) [17] . nonetheless, very low numbers of respiratory samples were collected (16,664 from 2006 to 2010) and even lower numbers (5, 167) were confirmed by the national system for virological surveillance of ari [17] [18] [19] [20] [21] . furthermore, the epidemiology, clinical characteristics, and viral causes of ili have been poorly characterized in venezuela; to our knowledge, only one longitudinal study, carried out during a limited period of time (february 2005-july 2006) and with a low number of patients (n = 102) with ari, has been published [22] . the objective of this paper is to describe the epidemiology of ili using data from a sentinel surveillance system at two major hospitals in maracay, venezuela. a total of 916 subjects participated in this study. six hundred and thirty seven (69.5%) were recruited at the hospital central de maracay (hcm). there was a slightly larger proportion of females compared with males who enrolled in the study (55.2% vs 44.8%), and the gender distributions at the two sites were similar. subject ages ranged from 1 month to 86 years with a mean age of 19.2 years (s.d. = 14.3 years) and a median of 17 years. a significant percentage of the subjects (17.6%) were under 5 years of age, while less than 1.0% were 60 years or older. the 15-29 year old group was the most predominant group comprising 32.8% of the study population, followed by the 5-14 year old age group (27.2%). the subjects who presented to hcm were significantly older (mean age = 21.2 years; s.d. = 13.2 years) compared with those who were seen at the hospital del instituto venezolano de los seguros sociales jose maria carabano tosta (ivss-jmct) (mean age = 14.4 years; s.d. = 15.6 years). a large proportion (37.8%) of the subjects at hcm belonged to the 15-29 age group, while nearly half of the patients (42.3%) seen at ivss-jmct were younger than 5 years old. approximately 5% of the subjects reported having received influenza vaccination within the last year. a small percentage of patients (2.0%) received antibiotic treatment for their acute respiratory illness prior to enrollment in the study. no hospitalized cases with ili were recruited in this study; only outpatient subjects participated. demographic information is summarized in table 1 . among the 916 patient samples obtained from the sentinel surveillance, at least one viral organism was detected by viral isolation and/or reverse transcriptase-polymerase chain reaction (rt-pcr) in 143 (15.6%) subjects ( table 2 ). only one participant with a co-infection was observed. influenza viruses were the most frequently detected organisms in this passive surveillance, consisting of 67.4% (97/144) of all the viruses detected by viral isolation and/or rt-pcr. of the 916 samples collected, 78 (8.5%) were positive for influenza a viruses and 19 (2.1%) for influenza b viruses by viral isolation and/or rt-pcr. in addition, noninfluenza viruses were detected only by virus isolation; they were: adenovirus (1.6%), parainfluenza virus (1.4%), rsv (1.2%), and other viruses (0.8% ), which included hbov (0.1%), enterovirus (0.2%), hmpv (0.1%), rhinovirus (0.1%), and herpes simplex virus (hsv, 0.2%). the only co-infection was observed in a 22 month old toddler in whom adenovirus and hsv were isolated. the virus etiology and detection rates varied with age ( table 2 ). the most commonly detected viral pathogens in the 0-4 year old group were influenza a (11.1%) and adenoviruses (5.6%). influenza a and b were the most commonly detected viruses in the school-age group (5-14 year old: 6.8% and 4.0%, respectively) and the late adolescent and young adult group (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) year-old: 9.7% and 1.7%, respectively). influenza a was the only detectable virus among patients who were 60 years or older. of the 78 influenza a cases, h3 subtype was detected in 33, and was observed in all the age groups but in subjects aged 45-59 year old ( table 2) . six cases had the h1 subtype (non-pandemic), and 18 influenza a viruses were isolated but not subtyped by one-step rt-pcr; of the latter, 11 can be classified as seasonal because they were isolated from patients samples collected before the start of the h1n1 pandemic outbreak of 2009. twenty-one cases of pandemic h1n1 2009 influenza virus (ph1n1) infections were identified in this sentinel surveillance; thirteen occurred in the 15-29 year old group. there were no cases of ph1n1 infections among adults 45 years old or higher. genetic analysis based on partial hemagglutinin gene sequences (approximately 900 bp) of 27 influenza isolates is shown in figure 1 (see material and methods for genbank accession numbers). this analysis shows that the circulating seasonal a/h1n1 strains in venezuela ( figure 1a ) were similar to the ones previously described in central and south america [15, 16] and all of these samples can be grouped with the a/ brisbane/56/07 like 2008-2009 genotype. the ph1n1 samples were part of only one cluster similar to previously reported strains in latin america [23] . influenza a/h3n2 samples possessed more genetic variability ( figure 1b) ; isolates from 2007 revealed two genotypes, a/brisbane/10/07-like and a/ california/7/04-like, while the more recent isolates were closer to the a/perth/16/09-like genotype. finally, in figure 1c , the genetic analysis for influenza b isolates disclosed the presence of two different genotypes, b/florida4/06-like and b/malaysia/ 2506/07-like, in agreement to what was previously found in the region [13, 16] . adenovirus was detected more frequently from december to april. rsv infection occurred more frequently from june to november. parainfluenza viruses were detectable throughout the year but without distinct seasonality. the passive surveillance illustrated the impact of ph1n1 on the distribution of respiratory viruses associated with ili in 2009 ( figure 3 ). during the ili peaks from 2006-2010, the percentage of ili cases attributable to influenza viruses ranged from 10% to 44%. the first cases of ph1n1 in this surveillance system were detected in july 2009, three months after the swine origin influenza outbreak began in mexico [24] . during the first wave in july 2009, the monthly virus detection rate rose to 33%, and 57% (4/7) of all the viruses detected were of the pandemic strain. the first wave simulated a typical ili peak activity similar to those observed in other years. however, during the larger second wave in october 2009, the virus positivity rate exceeded 40%, and the pandemic strain was identified in greater than 73% (11/15) of the viruses detected. figure 3 demonstrates that ph1n1 did not completely replace seasonal influenza a viruses. the peak pandemic activity in october 2009 was followed by a rapid decline in the rate of pandemic strain detection one month later. meanwhile, seasonal influenza a viruses remained in circulation throughout the pandemic period comprising 27% of all the influenza a viruses detected in october 2009. in 2010, 20% of influenza a specimens obtained via oropharyngeal swabs were randomly selected and tested for the pandemic strain via rt-pcr, and no cases of ph1n1 were detected ( figure 3 ). however, seasonal influenza a/h3n2 continued to be detected by our surveillance system with monthly positivity rates ranging from 16% to 30%. the clinical features of subjects in our surveillance are summarized in table 3 . compared with those whose respiratory secretions tested negative, subjects in whom virus was identified were more likely to have sore throat, headache, pharyngeal congestion, and ear pain. there were no significant differences in the symptoms of individuals who had seasonal influenza a when compared with those who suffered from ph1n1 influenza, except that a higher proportion of the latter subjects had expectoration, myalgias, and lymphadenopathy. the symptoms of influenza a (either seasonal or pandemic) and influenza b were clinically indistinguishable. when compared with patients who had ili due to other viruses, a higher percentage of those with confirmed influenza virus infection experienced sore throat, myalgias, and headache. there is scarce information about the epidemiology of acute febrile respiratory illness in venezuela and, to our knowledge, only one longitudinal study has been published [22] . this investigation, however, was limited to few number of subjects (n = 102) recruited during a short period of time (17 months) and did not describe the transmission seasonality of the viral infections. thus, our study is the first to fully describe the epidemiology and viral etiology of ili in venezuela and provides baseline levels of ili activity in a typical highly-populated urban city. our study demonstrated that influenza viruses are a main cause of ili at hcm and ivss in maracay in agreement with findings reported by venezuelas ness [17] [18] [19] [20] [21] . nevertheless, the rate of confirmed influenza virus infections found in our surveillance (10.6%, 97/916; tables 2 and 3) during the study period was lower than the one (27.7%) reported by the ness for the same period [17] [18] [19] [20] [21] . these differences in detection rates may be attributable to different strategies for capturing ari patients, especially those with influenza, used by the ness and by this study protocol (flores e, director of epidemiology, corposalud 2011, personal communication). the ness randomly selected a sample of ari cases (including those with influenza) with emphasis on severe hospitalized cases, whereas in our protocol we recruited ili subjects in an outpatient setting where the majority had symptoms that were not severe. on the other hand, the percentage of influenza viruses (not including ph1n1) detected in our study during a similar period of time, but in different years accounted for the significant differences found in both studies: a) the collection, preservation and further processing of respiratory samples, and b) the type of cells and ifa reagents used for virus isolation and identification. the proportion of influenza cases was significantly different when comparing non-pandemic and pandemic periods. before the h1n1 2009 pandemic, the ness [21] detected influenza virus in 64.7% of subjects in whom a virus was isolated; a similar proportion to the 55% (data not shown) found in our study. during the height of the pandemic (from july through dec 2009), the ness confirmed 97% of all ari cases with a virus as having either influenza a or b virus compared with 87% observed in our study. the predominance of influenza viruses as etiological agents of ili in maracay, venezuela, is consistent with observations of surveillance studies in other tropical central and south american countries [13, 15, 16] . while the overall virus detection rate (143/ 916, 15.6%; table 2 ) was lower compared with other studies, the positive rates for influenza a (8.5%) and influenza b (2.1%) in this surveillance were comparable to those observed in a similar study in el salvador, honduras, and nicaragua [16] . our findings were also consistent with a study in indonesia, which identified influenza a or b in 11% of all respiratory samples using viral isolation and rt-pcr [25] . our findings were consistent with those from a prospective study of outpatient children in northern taiwan, in which influenza a or b were isolated in 12.2% of subjects with ili, using madin-darby canine kidney (mdck) cell cultures and hemagglutinin inhibition assay for antigen detection [26] . however, our detection rate for all respiratory viruses, as well as influenza a and b viruses, was generally lower compared to other studies conducted in south america. in an expanded sentinel surveillance study in peru, which utilized a similar methodology for viral detection as our study, the virus positivity rate was 42.1%, and influenza a and b rates were 25.1% and 9.7%, respectively [13] . in a prospective surveillance study of ili in two ecuadorian cities using similar methods, at least one virus was detected in 35% of the participants; influenza a was detected by pcr in 21.6% while 6.4% tested positive for influenza b [15] . the lower virus detection rates found in our study is unclear because we used the same operative protocol described in the mentioned studies [13, 15] . nevertheless, the different skills to collect respiratory tract specimens from ili patients, by the health personnel employed in their studies and ours may have accounted for the lower virus detection rates found in our study. while influenza viruses were observed to be the most prevalent viral pathogen, adenovirus, parainfluenza virus, and rsv were important causes of ili at the two hospitals (table 2 ). in venezuela, the ness reported rsv as the most frequently detected non-influenza respiratory virus followed by parainfluenza virus and rhinovirus [17] [18] [19] [20] [21] . the study performed in zulia, venezuela, also reported rsv as the most common detected virus, followed by adenovirus, parainfluenza virus and influenza viruses [22] . as previously mentioned, the differences in the detection rates may be attributed to the different procedures used by the ness, other venezuelan researchers. [22] and us. these viral pathogens were also frequently detected in other surveillance studies in central and south america [14] [15] [16] . the highest detection rate of respiratory viruses was observed in the 0-4 year old group (29.2%; table 2 ). the rate of viral positivity generally decreased with age as acquired immunity increased in older subjects. a slight increase can be seen between the 30-44 year old group (14.1%) and 60 years or older group (14.3%). subjects who were 60 years old or older comprised less than 1% of the study population, and thus, the prevalence in this age group may be unreliable. contrary to our study, the other study from. venezuela had higher detection rates in both the 0-6 and $41 year old groups (48.2% and 57.1%, respectively) [22] . on the other hand, our findings are consistent with observations in the tecumseh study, a community-based surveillance in michigan, in which children were shown to have the highest rate of viral respiratory diseases [5, 6] . the tecumseh study further illustrated the variable impact of influenza viruses among the different age groups depending on the influenza virus subtype. for example, influenza a/h3n2 affected a wide range of age groups while influenza a/h1n1 and influenza b virus infections occurred more frequently among older children and young adults [5, 6] . our findings were similar being seasonal influenza a/h1n1 viruses and influenza b detected primarily in children and young adults, and seasonal influenza a/h3n2 found in all age groups. as expected, ph1n1 was the most common influenza a subtype identified among the subjects with ili in 2009 ( figure 3 ). in 2009, 37.5% (figure 3 ) of ili cases were due to ph1n1; this detection rate was lower than the 52.2% detection rate reported by the venezuela's ness [21] . nevertheless, this finding was similarly demonstrated in respiratory illness surveillance networks in other tropical and temperate south american countries such as guatemala [27] , peru [28] , argentina [29] , and brazil [30] . in 2010, non-pandemic influenza viruses continued to circulate in venezuela, and ph1n1 was not detected in our surveillance study, suggesting that ph1n1 did not displace seasonal influenza a viruses. during the same year, 1.6% of the ari cases reported to venezuela's ness were due to ph1n1 [21] . infection with ph1n1 may have stimulated immunity among the residents of this community during its initial arrival in 2009. this observation suggests that the circulating ph1n1 in 2010 may not have significantly mutated relative to the strain in 2009, so that antibodies stimulated by the natural infection during its initial arrival may have still been highly efficient in protecting the influenza-like illness in maracay, venezuela plos one | www.plosone.org community from another wave of ph1n1 outbreak. in 2010, the infection rate of influenza a/h3 was higher than the rates observed during previous years, further illustrating the lack of cross protection between ph1n1 and influenza a/h3. our study shows that hmpv and hbov were not commonly associated with ili, based on the low detection rates observed in our surveillance. in a study from ecuador which utilized methods similar to those employed in our study, a low detection rate for hbov (0.2%) was reported [15] . the latter study, as well as another from central america which used methods similar to ours, reported rare detection of hmpv (,0.2%) [15, 16] . in contrast, a prospective study of ili among brazilian adults, which utilized viral isolation and rt-pcr testing on respiratory samples, detected rhinoviruses in 19.6% of patients [14] . although rhinoviruses are typically associated with milder illness, they can contribute to the misdiagnosis of influenza based on clinical case definition alone. a cohort study of vietnamese children hospitalized for acute febrile respiratory illness, which applied multiplex-pcr assays on respiratory samples, revealed slightly higher prevalence rates of hbov (2%) and hmpv (5%) infections [31] . it is important to note that our method of identification (culture on three cell lines), compared to molecular diagnostic methods, substantially lacked sensitivity for detecting rhinoviruses, hmpvs, and hbovs. our study shows that patients with ph1n1 infections were more likely to have myalgias, productive cough with expectoration, and lymphadenopathy than with those infected with seasonal influenza a virus (table 3 ). in contrast, clinical manifestations in guatemalan subjects hospitalized for ph1n1 and seasonal influenza a infections did not significantly differ [27] . our study had limitations worth noting. data collection at only two hospitals in an urban area limits our ability to generalize our findings to the population. the low virus detection rates may be attributable to variations in the skills of the health staff employed to collect the respiratory specimens. the sampling method may have a significant effect on the proportion of respiratory viruses identified, and nasopharyngeal washes may yield higher sensitivity over nasopharyngeal or oropharyngeal swabs [32] . study participants were exclusively seen in the outpatient setting, thus limiting our ability to examine the impact of respiratory viruses in hospitalized cases. a large proportion of ili cases were not associated with any pathogen, and the impact of bacteria on this clinical syndrome cannot be determined from this study, since the respiratory samples were not cultured for bacteria. two methods of detection were used for identification of influenza (pcr and culture) whereas only one method of detection was used for the other viruses (culture). twenty-two percent (9/40) of the respiratory samples which were positive for influenza viruses by rt-pcr were negative by viral isolation illustrating that viral detection by culture underestimated the true prevalence. viral culture may not be the ideal way of isolating organisms such as rsv, hmpv, hbov and rhinoviruses leading to significant underestimation of their detection rates [33] . despite these limitations, our study contributes information on the distribution and etiology of ili at two hospitals in maracay, venezuela. this knowledge can serve as a baseline for future, more expansive population-based surveillance studies of influenza and other respiratory viruses in this region. maracay is located in the central northern region of venezuela (10u 159 n, 67u 399 w) , approximately 27 miles from the caribbean coast (figure 4) . the climate is tropical with two seasons: dry (december-april) and rainy (may-november). the monthly averages for temperature, relative humidity and rainfall are 25.1uc (range = 23.4uc-27.6uc), 75% (range = 66%-82%) and 56 mm (range: 0 mm-187 mm), respectively. maracay is comprised of two main urban municipalities, named girardot and mario briceñ o iragorry, and had an estimated total population of 677,359 in 2010. the hcm and the ivss-jmct are located approximately 4 miles apart. both are major county hospitals and referral health centers with adult and pediatric departments including emergency services and 241 (ivss-jmct) to 433 (hcm) hospitalization beds. they provide services to people in maracay as well as those from the adjacent city of aragua and three neighboring states. the study population included every outpatient with ili, regardless of age, who sought attention at the two study sites between october 2006 and december 2010, and agreed to participate in the study. at each site, trained medical personnel were responsible for properly identifying and classifying patients with ili. each person with ili was asked to enroll in the study. a person was defined as having ili if he or she had a sudden onset of fever ($38uc) and either cough or sore throat for less than five days in duration, with or without general symptoms such as myalgias, prostration, headache, or malaise [13] . data on gender, age, previous treatments, medical attention before enrollment, influenza vaccination status, and days of work/ school lost at the time of acute illness were collected utilizing a case report form (crf) from all participants who met the case definition criteria. temporal distribution of the results were recorded by month during the study period, taking into account the number of ili cases identified and the number of confirmed cases of influenza a and b in each study site. monthly reports of enrolled ili participants and laboratory results were sent to the venezuelan ministry of health. regular personnel training in protocol procedures were conducted as part of the strategy to improve sampling, storage,and shipping procedures. this protocol was approved as less than minimal risk research by the naval medical research center (nmrc), silver spring, maryland. institutional review board (irb; protocol nmrcd.2002.0019) authorization was given to perform the study using an information sheet approved and stamped by the irb. as this was part of clinical care and routine surveillance benefiting the ministry of health, verbal consent was obtained from all participants. this method of consent was accepted by the nmrc irb as well as the venezuelan institutions involved. a verbal consent was approved by both irbs following cioms and 45cfr46 (the common rule), 1) it was a minimal risk study, and 2) local health installations would not require a written consent for the procedures required in this study. additionally, this document included all the information that a written consent would require; a copy was provided to each study subject; and study personnel responsible for administering the process of informed consent were trained at each site on human research protection issues and were certified through the collaborative institutional training initiative (citi). finally, the frequent monitoring visits conducted by the study team found no problems in the process nor complaints from study participants. sample collection. nasal (807 of 916, 88.1%) or oropharyngeal (109 of 916, 11.9%) swabs were obtained from each subject for viral isolation and identification. the swabs were placed in viral transport media and stored at -70uc until they were delivered on dry ice to namru-6 in lima, perú, for laboratory analysis. duplicate swabs were processed and analyzed at the laboratorio regional de diagnostico e investigacion del dengue y otras enfermedades virales/instituto de investigaciones biomedicas de la universidad de carabobo (lardidev/ biomed-uc) in maracay, venezuela, following the same operative protocols used in namru-6. virus isolation and identification. nine-hundred and sixteen nasal or oropharyngeal swabs were processed for viral isolation and identification following the procedure described by laguna-torres et al. [13] . briefly, patient specimens were inoculated onto four cell lines: madin-darby canine kidney (mdck; atcch number ccl-34), african green monkey kidney (vero76; atcch numbercrl-1587) and veroe6 atcch number crl-1586), and rhesus monkey kidney (llc-mk2 atcch number ccl-7). upon the appearance of cytopathic effect or after ten days of culture (or thirteen days in the case of vero cells), the cells were spotted onto microscope slides. cell suspensions were dried and fixed in chilled acetone for 15 minutes. virus isolates were identified using direct fluorescence antibody (dfa) assays. the respiratory virus screening and identification kit (d3 dfa respiratory virus diagnostic hybrids; athens, oh) was utilized for the identification of adenoviruses, influenza a virus, influenza b virus, parainfluenza viruses (types 1, 2, and 3), and rsv. the d3 dfa herpes simplex virus (hsv) identification kit and the d3 ifa enterovirus id kit (diagnostic hybrids; athens, oh) were utilized for the identification of hsv (both hsv-1 and hsv-2) and enteroviruses, respectively. for isolation of hmpv, we used vero e6 and llc-mk2 cell lines. for detection of hmpv antigens by direct fluorescence assay, we used an anti-hmpv mouse monoclonal antibody from diagnostic hybrid (athens, oh). all assays were performed following the manufacturers' instructions. hbov was identified using the methods described by salmon-mulanovich, et al [34] . since duplicate swab samples were analyzed in different laboratories (lardidev/biomed-uc and namru-6) with the same diagnostic kit and standard operating procedures, a virus isolation result was considered positive if the specific virus was isolated and identified at either site. a subset of 254 specimens (222 nasals and 32 oropharyngeals) was analyzed by one-step rt-pcr and/or real time rt-pcr in order to sub-type influenza a/h1n1 (including ph1n1), influenza a/h3n2, and influenza b viruses. one-step rt-pcr and/or real time rt-pcr were not used to identify noninfluenza viruses because the specific primers and protocols were available only for sub-typing influenza viruses; thus, only virus isolation was used to detect and identify non-influenza viruses. of the 254 specimens, 48 (18.9%) were randomly selected from repository samples collected before the 2009 pandemic and tested for influenza a sub-typing by one-step rt-pcr and/or real time rt-pcr. at the onset of the 2009 pandemic, 150 of 254 (59.1%) respiratory samples were tested for ph1n1 by real time rt-pcr at the request of the venezuelan ministry of health. after the peak of the second wave of the pandemic, the percentage of respiratory samples tested by real time rt-pcr was reduced to 22% (56 of 254). one-step rt-pcr was performed according the procedure and influenza primers described below. real time rt-pcr was these protocols are available from cdc upon request. for the purpose of this study, an ili case with a confirmed viral respiratory infection was one in which viral isolation and/or rt-pcr identified a virus. rna extraction and one-step rt-pcr. viral rna extraction was performed from the supernatant of infected mdck cells using a qiaamp viral rna kit (qiagen; valencia, ca) following the manufacturer's protocol. the onestep rt-pcr was performed following a procedure described previously [13] with primers that amplified the hemagglutinin (ha) gene of influenza a and influenza b viruses using the superscript iii one-step rt-pcr system kit (invitrogen; san diego, ca). the following primers were used for the amplification of h1 influenza a viruses: h1f-6 (59-aagcaggggaaaa-taaaa-39) and h1r-1193 (59-gtaatcccgttaatggca-39); for h3 influenza a viruses: h3f-7 (59-actat-cattgctttgagc-39) and h3r-1184 (59-atggctgctt-gagtgctt-39); for influenza b viruses: bhaf-36 (59-gaagg-caataattgtact-39) and bhar-1140 (59-accagcaatagctccgaa-39). five ml of the extracted rna was added to 20 ml of master mix containing the enzyme mixture (superscript iii rt/platinum taq), 2x reaction mixture (containing 0.4 mm of each dntp and 3.2 mm of mg 2 so4) and 20 mm of each primer. cycling conditions included a reverse transcription step at 50uc for 30 minutes and a denaturation step at 94uc for 2 minutes. cycling conditions of the pcr were 40 cycles of 94uc for 15 seconds, 52uc for 30 seconds, and 68uc for 75 seconds, followed by a final incubation step at 68uc for 5 minutes. dna sequencing and phylogenetic analysis. for confirmation of serotype and genotype of the circulating influenza viruses, 27 samples were sequenced. as part of the respiratory surveillance protocol in venezuela, these 27 viruses were randomly selected from approximately 10% of the positive samples. only the ha gene region of influenza viruses was analyzed routinely for genotyping the one-rt-pcr products amplified with the primers described before were purified using centri-sep columns (princeton separation; englishtown, nj) and sequenced using the bigdye terminator v. 3.1 cycle sequencing kit (applied biosystems; foster city, ca) following the manufacturers' instructions. sequences were analyzed and edited using the sequencer 4.8 software (applied biosystems; foster city, ca). phylogenetic trees were constructed by the neighbor-joining method and bootstrap analysis to determine the best-fitting tree for the gene using mega software (version 4). the statistical significance of the tree topology was tested by bootstrapping (1,000 replicas). pairwise distances between and within the genotypes at the nucleotide level were calculated with kimura 2 parameters and with poisson correction at the amino acid level with mega software. genbank accession numbers are listed in table s1 . information on the crfs was entered into a database created in microsoft office access 2003. the chi square and fisher exact tests were used to compare means and associations using spss software version 10.0 (spss inc.; chicago, il) and r version 2.8.0 (r development core team; vienna, austria). projections of global mortality and burden of disease from 2002 to 2030 acute respiratory infections are the leading cause of death in children in developing countries the global burden of disease occurrence of respiratory virus: time, place and person viral respiratory infections in the community: epidemiology, agents, and interventions the tecumseh study of respiratory illness epidemiology of viral respiratory infections community epidemiology of human metapneumovirus, human coronavirus nl63, and other respiratory viruses in healthy preschool-aged children using parent-collected specimens viral pneumonia influenza-associated hospitalizations in the united states mortality associated with influenza and respiratory syncytial virus in the united states cdc (2010) prevention and control of influenza with vaccines: recommendations of the advisory committee on immunization practices influenza-like illness sentinel surveillance in peru acute respiratory infection and influenza-like illness viral etiologies in brazilian adults sentinel surveillance of influenza-like illness in two cities of the tropical country of ecuador influenza and other respiratory viruses in three central american countries boletin epidemiologico semanal semana epidemiologica 52 available boletin epidemiologico semanal semana epidemiologica 52 boletin epidemiologico semanal semana epidemiologica 52 boletin epidemiológico semanal semana epidemiologica 52 boletin epidemiologico semanal semana epidemiologica 52 etilogía viral de las infecciones respiratorias agudas genetic analysis of influenza a/h1n1 of swine origin virus (soiv) circulating in central and south america outbreak of swine origin influenza a (h1n1) virus infection -mexico influenza surveillance in indonesia surveillance of respiratory viral infections among pediatric outpatients in northern taiwan a comparison of the epidemiology and clinical presentation of seasonal influenza a and 2009 pandemic influenza a(h1n1) in guatemala pandemic influenza in a southern hemisphere setting: the experience in peru from pandemic (h1n1) 2009 cases epidemiology of human infection with the novel virus influenza a(h1n1) in the hospital das clinicas, sao paolo, brazil viral pathogens associated with acute respiratory infections in central vietnamese children respiratory viruses in adults with community-acquired pneumonia role of cell culture for virus detection in the age of technology frequency of human bocavirus (hbov) infection among children with febrile respiratory symptoms in argentina we would like to express our gratitude to all personnel working at the two sentinel health centers (hcm and ivss-hjmct) in venezuela for supporting this surveillance study. we thank mr. eduardo guerra, the informatics superior technician of lardidev/biomed-uc, for his irreplaceable technical work in venezuela. we also thank the professional staff of the virology department of namru-6 for invaluable laboratory and technical support in the execution of the study.disclaimer: the views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the department of the navy, department of defense, nor the u.s. government. the corresponding author had full access to all data in the study and final responsibility for the decision to submit this publication.additional key: cord-321704-jozrgcq3 authors: tan, xin quan; zhao, xiahong; lee, vernon j; loh, jin phang; tan, boon huan; koh, wee hong victor; ng, sock hoon; chen, mark i-cheng; cook, alex richard title: respiratory viral pathogens among singapore military servicemen 2009 – 2012: epidemiology and clinical characteristics date: 2014-04-15 journal: bmc infect dis doi: 10.1186/1471-2334-14-204 sha: doc_id: 321704 cord_uid: jozrgcq3 background: few studies have comprehensively described tropical respiratory disease surveillance in military populations. there is also a lack of studies comparing clinical characteristics of the non-influenza pathogens with influenza and amongst themselves. methods: from may 2009 through october 2012, 7733 consenting cases of febrile respiratory illness (fri) (temperature [greater than or equal to]37.5degreesc with cough or sorethroat) and controls in the singapore military had clinical data and nasal washes collected prospectively. nasal washes underwent multiplex pcr, and the analysis was limited to viral mono-infections. results: 49% of cases tested positive for at least one virus, of whom 10% had multiple infections. 53% of the fri cases fulfilled the definition of influenza-like illness (ili), of whom 52% were positive for at least one virus. the most frequent etiologies for mono-infections among fri cases were influenza a(h1n1)pdm09 (13%), influenza b (13%) and coxsackevirus (9%). the sensitivity, specificity, positive predictive value and negative predictive value of ili for influenza among fri cases were 72%, 48%, 40% and 69% respectively. on logistic regression, there were marked differences in the prevalence of different symptoms and signs between viruses with fever more prevalent amongst influenza and adenovirus infections than other viruses. conclusion: there are multiple viral etiologies for fri and ili with differing clinical symptoms in the singapore military. influenza and coxsackevirus were the most common etiology for fri, while influenza and adenoviruses displayed the most febrile symptoms. further studies should explore these differences and possible interventions. influenza-like illness (ili) is often used for influenza surveillance [1] , as influenza is a disease of global interest with 5% of adults developing symptomatic disease annually and with case fatalities of 3.5% in susceptible populations [2] . while influenza surveillance remains a priority, ili can also be caused by a wide range of viral pathogens that present with a spectrum of respiratory symptoms [3] [4] [5] [6] . in the tropics, viral respiratory pathogens have been reported to exhibit different seasonality and transmission characteristics compared to temperate climates [2, [7] [8] [9] . this necessitates a better understanding of their epidemiology to assess the utility and importance of surveillance in these settings. the year-round circulation of respiratory viruses in the tropics may also predispose patients to coinfection with multiple pathogens, with implications for severity of disease [10, 11] and secondary bacteria infection [12, 13] . while there have been studies comparing differences in clinical presentation between influenza and non-influenza cases [14] , few describe the epidemiology and differences in clinical presentation among various non-influenza respiratory viruses. as influenza viruses have accounted for only between 10.1% to 53.0% of all ili cases [15] [16] [17] , it is important to understand the contribution of other respiratory pathogens to overall morbidity and to determine their epidemiological distribution and clinical presentation. to address these issues, this study explores data obtained from a respiratory disease sentinel surveillance system in the singapore military to examine the etiologic viral agents of respiratory illnesses in a tropical environment, to determine the viruses that circulate post-influenza vaccination, and to compare the differences in clinical presentation. singapore is a city-state in tropical south-east asia with a population of 5.3 million people (mid-year 2012). the singapore military is based on national service in which all male citizens and liable permanent residents serve for two years after high school. servicemen typically live in barracks-style accommodation on weekdays and return home on weekends. the singapore military started a sentinel respiratory disease surveillance program in 5 major camps (including a recruit training camp) on 11 may 2009, tracking febrile respiratory illness (fri) cases (temperature ≥37.5°c with cough or sore throat). the definition of fri contrasts with influenza-like illness (ili, defined as fever ≥38.0°c with cough or sore throat) to broaden the capture of other febrile cases that also result in absenteeism while limiting cases to those with fever as an indicator of severity. this allows for detection of a larger number of respiratory pathogens. patients who visited the primary healthcare clinics in the camps between 11 may 2009 and 31 october 2012 during regular consultation hours who met the fri criteria were recruited. healthcare workers obtained written informed consent, administered a questionnaire, obtained clinical specimens and performed a clinical examination on partcipants. repeat consultations were excluded if the healthcare worker determined that the patient had not recovered from the first illness episode. we also obtained samples from controls (those without respiratory symptoms or acute infections), who were recruited across the year at between 5 to 10 persons per week. informed consent, the baseline questionnaire, and clinical specimens were obtained. from december 2009, all recruits were administered with the influenza a(h1n1)pdm09 (flu-a(h1n1)pdm09) vaccine. the trivalent seasonal influenza vaccination was first introduced to recruits in december 2010, followed henceforth by all other personnel in november 2011. nasal washes from each side of the nose were taken from consenting participants by trained medical staff, placed in viral transport media and refrigerated. the samples were transported to the laboratory on ice for etiological testing within 24 hours. laboratory analysis was performed in an iso15189accreditated laboratory for molecular diagnostics which regularly takes part in external proficiency programs such as qcmd eqa programs. detailed laboratory methods were previously described [14] . we used the multiplex pcr strategy based on the resplex assays described below, and performed additional singleplex pcr assays to determine the influenza subtype. total nucleic acids were extracted from each specimen using the dna minikit (qiagen, inc, valencia, ca, usa) according to manufacturer's instructions. a total of 20 μl of extract were tested with resplex i and ii (version 2.0, qiagen, inc., valencia, ca, usa) [18] for respiratory micro-organisms on the liquichip 200 workstation, according to manufacturer's instructions. the resplex i and ii (version 2.0) assays are multiplex pcr assays coupled with bead array detection technology and can simultaneously detect and subtype 18 different pathogens including influenza a (flu-a) and influenza b (flu-b). specimens that were resplex ii positive for flu-a were further subtyped with real-time pcr for h1 or h3 (singapore ministry of health), or for flu-a(h1n1)pdm09. briefly, 5 μl of total genetic extracts were tested using an in-house developed assay based on the one-step superscriptiii/platinum taq kit (invitrogen, carlsbad, ca, usa) following manufacturer's instructions on the lightcycler machine from roche or the applied biosystems real-time pcr machine (7500). the analysis was limited to viral mono-infections amongst cases to discern clinical presentations and symptom complexes associated with each pathogen. we excluded viruses with fewer than 20 cases (0.6% of the total), as the number was too small to have a reasonable sample sizethese were coronavirus hku1 (cov-hku1), parainfluenza 1 (hpiv-1), hpiv-2, hpiv-4, influenza a(h1n1) (the prepandemic strain), respiratory syncytial virus a (rsv a), rsv b, cov and bocavirus (bv). this left 14 viruses for the subsequent analyses. the main aim was to compare the differences in clinical expressions, including individual symptoms (or signs), pairs of symptoms, and overall symptom load between patients with different viral infections. we counted the clinical symptoms/signs and calculated the corresponding empirical proportions with 95% confidence intervals (cis) to evaluate the overall symptom load. logistic regression analysis was used to investigate the differences in symptom expressions for each pair. differences were identified at a significance level of 0.05. to assess the presence of paired symptoms/signs for all viruses, we conducted binomial tests to compare the joint proportions of symptom pairs occurring together to the expected proportions assuming independence of symptoms. the ratio of the observed proportion of symptom pairs relative to the product of the marginal proportion of each symptom is defined as the excess probability ratio which measures effect size. multivariate logistic regression analysis was performed to compare the risk of having an individual symptom/sign among viral mono-infections by assigning a categorical variable for all viruses as the primary predictor. potential confounding was addressed by adjusting the model for age, smoking status, asthma and heart disease. non-significant variables were dropped at a significance level of 0.05 to obtain the final model. statistical analyses were performed using the r statistical software (version 3.0.0) [19] . ethics approval was given by the singapore military's joint medical committee for research, and the national university of singapore's ethics review committee. the basic demographic data are described in table 1 . participants were mostly young male adults, with other characteristics largely similar. however, there were significantly less recruits amongst controls than amongst other groups. the temporal distribution of cases is described in figure 1 . no obvious overall seasonal pattern can be observed. the peak in june and july 2009 corresponds to the flu-a(h1n1)pdm09 pandemic [16] . as this peak tailed off, we observed an increase in flu-b cases (starting feb-mar 2010). subsequently, as the flu-b cases fell, adenovirus e (adv-e) cases started to increase. coxsackie/echovirus (cv) and rhinovirus (rv) infections were consistently present in the earlier periods but appeared to tail off by 2012, corresponding to the rise in fri cases due to other viruses. the etiologies of selected infections are illustrated in table 2 . at least one virus was detected in 3794 of the 7733 fri cases (49.1%). in 376 (9.0%) of the 3794 cases, more than one virus was detected and these were excluded. 4120 (53.3%) of the fri cases fulfilled the definition of ili; 2146 (52.1%) of these ili cases were positive for at least 1 virus. of the 3430 fri mono-infection cases, 2128 (62.0%) were viral. 1259 (59.2%) of these 2128 cases met the definition of ili. we examined the proportion of ili cases among those with viral mono-infections (table 3) . influenza viruses accounted for only 40% of ili. among fri cases, more than 60% of patients with influenza and adenovirus infections presented with ili. however, several other viral infections led to high rates of ili, including cv, human metapneumovirus (hmpv) and covs. the sensivity, specifity, positive predictive value (ppv) and negative predictive value (npv) of ili for influenza was 72.2%, 48.1%, 40.1% and 69.3% respectively. on from the multivariate analysis ( figure 2 ), compared to most other viruses, a flu-a(h1n1)pdm09 and flu-a (h3n2) less commonly resulted in sorethroat. running nose was more common in enterovirus (ev) and rv cases and less common in adv. flu-b was more likely than a majority of the other viruses to cause dry cough while ev and adv-e were less likely to cause dry cough. cov-oc43 and hmpv were more likely to cause cough with phlegm than the most other viruses. influenza viruses and adenoviruses were more likely to cause fever ≥38.0°c. in figure 3 , we explored the associations (and dissociations) between different clinical symptoms and signs across all viruses. some are expected, such as association of fever ≥37.8°c and fever ≥38.0°c and dissociation of dry cough and cough with phlegm. fever ≥38.0°c was also associated with systematic complaints, such as chills, bodyache, headache and eye pain. sorethroat was associated with an injected pharynx. our study shows the different viral etiologies of ili and compares the clinical characteristics of different viral etiologies in a tropical setting. this data series only spanned three years, and initial observations showed no clear seasonal variation compared to temperate regions, similar to previous reports of overall tropical respiratory disease patterns [7, 9, 20] . the initial peak corresponded to the flu-a(h1n1)pdm09 pandemic [21] , with the subsequent lower incidence in the recruit population likely due to vaccination with the pandemic vaccine a year before annual seasonal vaccination was started across all personnel [22] . the number of fri cases remained fairly consistent throughout the study period (except the pandemic). however, prevalence of pathogens varied throughout, with some negative correlation observed between the virusese.g. a drop in flu-a followed by a rise in flu-b activity, and a drop in flu-b cases followed by a rise in adenovirus activity. correlation of viral activity have previously been reported -wang et al [23] reported negative association between rv and adv rates, while bellei et al [15] and razanajatovo et al [24] described concomitant rise of influenza and rv, and influenza and adv activity respectively. in addition, kasper et al reported that ili rates remained constant despite varying prevalences of influenza [25] . this supports our findings that multiple agents are capable of causing ili, and a decrease in the prevalence of one virus was replaced by an increase in prevalence of another. further studies across a longer time period are necessary, especially for vaccine effectiveness evaluation. 49.1% of fri and 52.1% of ili cases were positive for a virus, similar to the 44.5% and 61.8% reported by studies targeting similar panel of organisms [3, 6, 15, 17, 26, 27] . the remaining fri cases may be due to non-viral agents, agents beyond the ability of the test, non-infectious causes, and possible sampling errors. viruses most commonly detected in ili cases were flu-a, flu-b, and cv in that order. influenza was also the top etiologic agent for ili in some studies [6, 15, 17, 24, 28] although other pathogens have been identified to be most prevalent in different settings, such as rsv and hmpv in france [29] , influenza and rsv in the usa [30] , influenza and rv in central america [16] and china [6] and in italy, influenza and adv [31] . the range of pathogens indicates a need to perform local continual surveillance since prevailing pathogens differ across different populations, geographic regions and climates. the high incidence of cv warrants further studyhand, foot and mouth disease is endemic to singapore and cv is frequently identified in pediatric samples [32] , and it is possible that cv circulates at high levels in adults also. previous studies have identified a co-infection rate of 11.0 to 47.0% [6, 11, 24, 33, 34] for viruses, higher than the 9.9% found in this study despite some studies using a less extensive diagnostic panel in a similar age group. possible reasons include our highly influenza-vaccinated population or a warmer climate with higher relative humidity resulting in lower virus circulation [35] and a study population that did not include children (studies reported higher co-infection rates amongst pediatric patients [17, 24] ). in both univariate and multivariate analysis, adv and influenza viruses were more likely to cause fever (≥38.0°c). this finding has been demonstrated in other studies [14, 15, [36] [37] [38] [39] . fever also tends to be associated with other systemic complaints such as eye pain, bodyache and headache; this may be due to cytokine mediated systematic inflammatory response [40] and could indicate more severe disease. bellei et al's [15] study in brazil also found that ev, rv and cov were least likely to cause fever, and that rv and ev cases were the most likely to present with rhinorrhea. cough with sputum in flu-a(h3n2) and flu-a were less prevalent than in other viruses, in contrast to reports in other settings [14, [37] [38] [39] . we found that flu-b cases were more likely to report dry cough, similar to other studies [14, [37] [38] [39] . the heterogeneity of results across different studies highlight the difficulty of using clinical symptoms in determining the etiology of ili. we also detected a small proportion of asymptomatic individuals who tested positive for the various viruses ( table 2 ). these could represent carriage without infection or subclinical/asymptomatic infections during periods of virus circulation. although ili is widely used to identify influenza, the traditional definition would have picked up only 69% of influenza infections (except untyped flu-a) that were identified as fri. adv infections also frequently fulfilled the ili definition (89.7% of adv-b and 65.0% of adv-e), as did substantial fractions of cv, hmpv, and cov. only 40.1% of viral mono-infections that met the ili definition were due to influenza. we found a fairly high sensitivity (72.2%) of ili for influenza, but a low specificity (48.1%) in keeping with sensitivities of 55.4-86.8% and specifities of 39.3-67% reported elsewhere [25, 39, 41, 37] . color cells represent variables that are significant at the 5% level, and the thickness of the cell wall represents the p-value (thin means 0.01 < p < 0.05; medium, 0.001 < p < 0.01; and thick, p < 0.001). the odds ratios are encoded by colors where a red cell indicates an odds ratio > 1; and blue otherwise. for example, for a sore throat, flu-a(unknown), flu-b, cv, rv, adv-e, cov-oc43 and cov-nl63 have more of the sore throat than iflu-a(h1n1)pdm09 indicated by the red cells in the row for flu-a (h1n1)pdm09 and corresponding columns. ppv of ili was low (40.1%) compared to other studies (77.6-79.0%) likely because these studies were conducted during influenza seasons. this is supported by the higher npv (69.3%) compared to other studies (48.9-55.0%) [25, 39, 41, 37] . two tropical studies [14, 25] also report low ppv of 21.0-37% and high npv of 81-90.0%. adv are as likely as influenza viruses to present with fever and tend to be captured by the ili definition. however, they are more likely to present with cough and sorethroat than influenza. it may be possible to differentiate adv and influenza infection based on rhinorrhea and other symptoms. this could be useful in surveillance and clinical management, especially when deliberating whether to start antivirals. early etiologic diagnosis of influenza has been shown to be cost effective [42] with reduced antibiotic use and may reduce complications with early antivirals. it may be possible to combine a clinical diagnostic model with rapid testing to achieve these goals. the analysis was limited to viral mono-infections and future studies should explore co-infections and bacterial infections. this study involved predominantly young adult males, and results may not be generalizable to the overall population, necessitating further studies among various age groups and gender. there were also less recruits amongst controls than amongst other groups, and this would be an important consideration when comparing the two groups in the future. finally, the actual clinical impact of differentiating between various viral etiological agents may be limited, and we could not determine the relative severity of symptoms other than fever. our study highlights the varied etiology for fri and ili in the tropical settinginfluenza and adv and cv were all common. influenza and advs tend to present with higher fever, and vaccination should be considered. the utility of ili for tropical surveillance of influenza needs to be reviewed given the low ppv and high npv compared to temperate regions. the surveillance system has enabled the singapore military to understand the etiologic agents affecting servicemen, hence implementing and evaluating controls measures such as vaccination. figure 3 correlation of symptoms and signs across all viruses. clinical signs or symptoms are listed by average frequency from the most to the least. binomial test is used to assess the discrepancy between the observed proportion of symptom pairs and the expected proportion of symptom pairs which is the product of the two marginal distributions by assuming symptoms develop independently. color cells represent differences that are significant at the 5% level, and the thickness of the cell wall represents the p-value (thin means 0.01 < p < 0.05; medium, 0.001 < p < 0.01; and thick, p < 0.001). the excess probability encoded by colors measures the effect size. if the observed proportion is lower than the expected proportion, the cell will be shaded by blue color, and red color otherwise. mounts aw: strategy to enhance influenza surveillance worldwide acute respiratory illness in the community. frequency of illness and the agents involved viral agents responsible for febrile respiratory illnesses among military recruits training in tropical singapore acute respiratory symptoms in adults in general practice surveillance of 16 respiratory viruses in patients with influenza-like illness in nanjing epidemiology and seasonality of respiratory tract virus infections in the tropics influenza in the tropics seasonal trends of viral respiratory tract infections in the tropics single versus dual respiratory virus infections in hospitalized infants: impact on clinical course of disease and interferon-[gamma] response epidemiology of multiple respiratory viruses in childcare attendees respiratory viral infections in infants: causes, clinical symptoms, virology, and immunology invasive pneumococcal pneumonia and respiratory virus co-infections a clinical diagnostic model for predicting influenza among young adult military personnel with febrile respiratory illness in singapore acute respiratory infection and influenza-like illness viral etiologies in brazilian adults influenza and other respiratory viruses in three central american countries. influenza other respir viruses viral aetiology of influenza-like illness in belgium during the influenza a(h1n1)2009 pandemic evaluation of commercial resplex ii v2.0, multicode-plx, and xtag respiratory viral panels for the diagnosis of respiratory viral infections in adults r: a language and environment for statistical computing prevalence and seasonality of influenza-like illness in children outbreak of pandemic influenza a (h1n1-2009) in singapore effectiveness of pandemic h1n1-2009 vaccination in reducing laboratory confirmed influenza infections among military recruits in tropical singapore broad spectrum respiratory pathogen analysis of throat swabs from military recruits reveals interference between rhinoviruses and adenoviruses viral etiology of influenza-like illnesses in evaluation of an influenza-like illness case definition in the diagnosis of influenza among patients with acute febrile illness in cambodia seasonal variations of 15 respiratory agents illustrated by the application of a multiplex polymerase chain reaction assay rapid multiplex nested pcr for detection of respiratory viruses influenza-like illness sentinel surveillance in peru nationwide surveillance of 18 respiratory viruses in patients with influenza-like illnesses: a pilot feasibility study in the french sentinel network aetiology of influenza-like illness in adults includes parainfluenzavirus type 4 respiratory viruses and influenza-like illness: a survey in the area of rome dual respiratory virus infections respiratory viral infections during the 2009-2010 winter season in central england, uk: incidence and patterns of multiple virus co-infections relationship between humidity and influenza a viability in droplets and implications for influenza's seasonality fever in respiratory virus infections clinical signs and symptoms predicting influenza infection differing symptom patterns in early pandemic vs seasonal influenza infections predicting influenza infections during epidemics with use of a clinical case definition understanding the symptoms of the common cold and influenza influenza-like illness criteria were poorly related to laboratory-confirmed influenza in a sentinel surveillance study clinical and financial benefits of rapid detection of respiratory viruses: an outcomes study submit your next manuscript to biomed central and take full advantage of: • convenient online submission • thorough peer review • no space constraints or color figure charges • immediate publication on acceptance • inclusion in pubmed, cas, scopus and google scholar • research which is freely available for redistribution key: cord-261241-eqf6ame6 authors: van beek, josine; veenhoven, reinier h; bruin, jacob p; van boxtel, renée a j; de lange, marit m a; meijer, adam; sanders, elisabeth a m; rots, nynke y; luytjes, willem title: influenza-like illness incidence is not reduced by influenza vaccination in a cohort of older adults, despite effectively reducing laboratory-confirmed influenza virus infections date: 2017-08-15 journal: j infect dis doi: 10.1093/infdis/jix268 sha: doc_id: 261241 cord_uid: eqf6ame6 background: data on the relative contribution of influenza virus and other respiratory pathogens to respiratory infections in community-dwelling older adults (≥60 years) are needed. methods: a prospective observational cohort study was performed in the netherlands during 2 winters. nasopharyngeal and oropharyngeal swabs were collected during influenza-like illness (ili) episodes and from controls. viruses and bacteria were identified by multiplex ligation–dependent probe amplification assay and conventional bacterial culture. results: the ili incidence in the consecutive seasons was 7.2% and 11.6%, and influenza virus caused 18.9% and 34.2% of ili episodes. potential pathogen were detected in 80% of the ili events with influenza virus, coronaviruses, rhinoviruses, human metapneumovirus, respiratory syncytial virus, parainfluenza viruses, and haemophilus influenzae being the most common. influenza vaccination reduced influenza virus infection by 73% (95% confidence interval [ci], 26%–90%) and 51% (95% ci, 7%–74%) in ili patients. however, ili incidence was similar between vaccinated (7.6% and 10.8%) and nonvaccinated (4.2% and 11.4%) participants in 2011–2012 and 2012–2013, respectively (p > .05). conclusions: influenza virus is a frequent pathogen in older adults with ili. vaccination reduces the number of influenza virus infections but not the overall number of ili episodes: other pathogens fill the gap. we suggest the existence of a pool of individuals with high susceptibility to respiratory infections. clinical trials registration: ntr3386. influenza virus causes seasonal epidemics, resulting in 3-5 million severe cases and 250 000-500 000 deaths globally each year [1] . elderly persons, individuals with certain medical conditions and children aged <2 years have the highest risk for complications. vaccination is an important tool to prevent infection and to reduce morbidity and mortality [1] . in the netherlands, individuals aged ≥60 years are offered the annual influenza vaccination. however, public acceptance of vaccination is moderate [2] . vaccine effectiveness (ve) varies per season and depends on age and health of the recipients, and the antigenic match of vaccine strains with circulating strains [3] [4] [5] . furthermore, there is scientific debate about the methodology of determining ve [5, 6] . these discussions reach the media and influence the general opinion on influenza vaccine benefit. moreover, to the public, flu as caused by influenza virus is the same as influenza-like illness (ili) caused by respiratory pathogens, against which influenza vaccination will not protect. consequently, influenza vaccination is perceived to be ineffective and vaccine uptake is reduced. to counter this trend, data on the relative contribution of influenza virus and other respiratory infections to ili in older community-dwelling adults are lacking and needed [7] [8] [9] [10] . this group is underrepresented in the dutch primary care sentinel surveillance system [10, 11] and by definition absent in the dutch sentinel nursing home surveillance network [12] . this is also the case in many other studies worldwide [7] [8] [9] . the aim of this prospective observational study was to determine the relative contribution of influenza virus and other respiratory pathogens to ili in older adults (aged ≥60 years) in 2 consecutive seasons in the netherlands. in addition, influenza ve was estimated in both seasons. upon ili, we determined the presence of potential pathogens in the pharynx of the participant within 72 hours of symptom onset. as a control, we analyzed samples of the same individuals taken after recovery (8 weeks) and in a subset of asymptomatic controls for the same potential pathogens. this prospective observational study was conducted during 2 consecutive influenza seasons (from december 2011 to april 2012 [2011] [2012] and from october 2012 to may 2013 [2012] [2013] ) in the netherlands. adults aged ≥60 years were recruited through their general practitioner or through the civil registry. for the second season, participants were reinvited and additional participants were recruited through the civil registry. we also started earlier to use the same monitoring period as in the dutch sentinel surveillance system [11] . there were no exclusion criteria for the study. influenza vaccination status was recorded (2009) (2010) (2011) (2012) . participants were part of the study for the entire duration of each season and contacted at the end of season to verify participation. written informed consent was obtained from all participants. all trial-related activities were conducted according to good clinical practice, which includes the provisions of the declaration of helsinki. the study was approved by the acknowledged ethical committee metc noord holland (http://www.trialregister.nl; ntr3386). participants were instructed about ili symptoms according to the dutch pel criteria, defined by fever (≥37.8°c) with at least 1 other symptom of headache, myalgia, sore throat, coughing, rhinitis, or chest pain [13] and to report ili as soon as possible after onset. a research nurse performed a home visit within 72 hours of fever onset. during this acute phase, nasopharyngeal and oropharyngeal swabs were obtained and additional information on demographics and comorbidities was recorded. a second visit (recovery phase) was performed 8 weeks (±1 week) later, during which the same samples were collected. if a new ili episode was reported, participants were visited again. in the second season, a control group of asymptomatic participants, equally distributed over the different age groups and season, was sampled and questioned. nasopharyngeal and oropharyngeal samples were obtained with a sterile swab with a flocked nylon tip and stored separately in 1 ml modified liquid amies transport medium (eswab, copan, brescia, italy). samples were transported at room temperature to the laboratory and processed and stored at -80°c within 8 hours after sampling. dna and rna were isolated from 200 µl of both swabs by easy-mag isolation and eluted in 25 µl of buffer (biomérieux, the netherlands). five microliters was used for the detection of a panel of respiratory viruses and bacteria by a real-time polymerase chain reaction (pcr)-based multiplex ligation-dependent probe amplification (mlpa) assay (respifinder smart 22 kit; pathofinder, the netherlands). all analyses were performed on a roche lightcycler 480. mlpa analysis was performed on both swabs separately. a participant was excluded from the analysis and considered missing if either of the swabs or data for a swab were missing. for mlpa data analysis, a participant was considered positive for a target if at least 1 swab of the participant was positive. influenza virus-positive samples were subtyped by real-time reverse-transcription pcr using the roche lightcycler 480 system with slightly modified protocols as described previously [14, 15] . conventional culture of oropharyngeal swabs was performed for streptococcus pneumoniae and haemophilus influenzae according to standard procedures [16] . discrimination of h. influenzae and haemophilus haemolyticus was performed by matrix-assisted laser desorption/ionization-time of flight (maldi-tof) [17] . participants without available oropharyngeal swabs were excluded from analysis and designated as missing. for analysis of the contribution of influenza virus, a sample size of 200 ili cases was estimated to be required. based on an expected ili incidence of 7.5% and a drop-out rate of 5%, a cohort size of 2100 participants was calculated. based on results of the first season, 2500 participants were included in the second season. pearson χ 2 testing and independent samples t test of the means was applied to analyze participant characteristics with spss 19.0 for windows software. a p value ≤.05 was considered significant. the incidence of different viruses or bacteria was calculated as the percentage of swabs positive with the potential pathogen of the number of ili events during the season. attack rates were calculated as percentage of detected pathogens per number of monitored participants. vaccine effectiveness was determined by test-negative design analysis of ili positive participants in the influenza-active period [18] . the analysis is restricted to the period that influenza virus was circulating in the netherlands for that particular season, defined by the national influenza surveillance weekly reports. participants with <14 days between the date of vaccination and the date of home visit were excluded from the ve analysis, as it is uncertain whether the vaccine already had any effect in this period. second and third ili periods were included in the analysis only if an earlier ili period was influenza virus negative. for ili participants, influenza virus positive was considered "case" and influenza negative "control. " the ve is calculated as (1 -odds ratio [or]) × 100% with 95% confidence interval (ci) and is calculated per influenza virus subtype or lineage. the following factors were regarded as potential confounders: period in the season (early and late season), sex, smoking, comorbidity, and age (natural smoothing spline, 4 degrees of freedom). the association between the potential confounders and influenza virus positivity (any subtype) was analyzed with univariate logistic regression. variables with p < .20 were considered in the multivariable analysis. variables that changed the or by at least 5% are included in the final multivariable logistic regression model for any influenza subtype (backward selection). all analyses were performed with sas version 9.4 software. sensitivity analyses were performed in the ve analyses for other control groups (supplementary materials). in this prospective study, we observed an ili incidence of 7.2% (143/1992) and 11.6% (275/2368) in 2 consecutive seasons (2011-2012 and 2012-2013, respectively) ( figure 1a ; table 1 ). the average age of vaccinated individuals was significantly higher than that of unvaccinated individuals (respectively, 70.4 vs 66.9 years in 2011-2012 and 71.9 vs 67.9 years in 2012-2013; table 2 ). the asymptomatic controls from the second season were older and more often vaccinated compared with the overall cohort ( figure 1b ; table 2 ). furthermore, participants who reported comorbidities were vaccinated significantly more often than participants who did not report comorbidities (table 3) . no significant differences were found between individuals with or without ili with respect to sex, age, and chronic illnesses (tables 1 and 3) . importantly, no differences were found in the incidence of ili episodes between vaccinated and unvaccinated participants (table 1) . in 79.1% and 78.0% of the acute ili samples from the 2 seasons, at least 1 potential pathogen could be identified by mlpa or bacterial culture ( in 16.2% and 17.0% of the samples from the acute phase, >1 potential pathogen was detected, but no specific combinations of viruses and/or bacteria were observed (data not shown). in recovery samples, 8 weeks after acute ili, potential pathogens were detected in 27.0% and 24.8% of cases in 2011-2012 and 2012-2013, respectively. in asymptomatic control samples, similar potential pathogens were observed as in recovery samples (21.5%) ( figure 2c ). in 2011-2012, influenza virus was detected in 18.9% of the acute ili samples, predominantly of the a(h3n2) subtype (96.3%) ( figure 3 ; supplementary table 1a) . influenza virus was not detected in the corresponding recovery samples in this season, suggesting that influenza virus was the actual cause of ili. in 2012-2013, influenza virus was detected in 34.2% of the acute ili samples, and all 4 circulating subtypes were detected: 43.6% a(h3n2), 25.5% a(h1n1)pdm09, 25.5% b/yamagata lineage, and 5.3% b/victoria lineage. in addition, influenza virus was detected at a very low level in both the recovery samples and in samples of asymptomatic controls (1.1% and 0.9%, respectively). we investigated which other viruses and bacteria were detectable during ili episodes. in 60.8% (2011-2012) and 44.7% (2012-2013) of ili samples, potential pathogens other than influenza virus were detected (figure 3 ; supplementary table 1 ). coronaviruses of all 4 common human subtypes (18.2% in 2011-2012 and 11.3% in 2012-2013), human metapneumovirus (hmpv) (20.3% and 3.6%), rhinoviruses (8.4% and 21.1%), respiratory syncytial virus (rsv) (4.9% and 6.5%), and parainfluenza viruses (2.8% and 5.1%) were detected in >5% of the ili samples in at least 1 season. hmpv and rhinovirus varied most between the 2 seasons. all viruses were detectable at low levels in recovery and control samples. rarely, the same virus was observed during both the ili event and the recovery sampling, except for rhinoviruses, which were detected frequently at both visits (8.3% in 2011-2012 and 17.2% in 2012-2013), but as our test only detected rhinovirus in general, we cannot exclude that these are different serotypes. the attack rate of influenza virus was significantly higher in the second season compared to the first season (p < .0001; table 4 ). interestingly, the increased attack rate could be attributed to significant increases in the attack rates of h1n1, the b/victoria-like subtype, and the b/yamagata-like subtype (p < .0001, p = .04, and p < .0001, respectively), whereas the attack rate of the h3n2 subtype was not significantly different between these seasons. for the other viruses, we observed significant increased attack rates in the second season for rhinovirus (p < .0001) and parainfluenza viruses (p = .04), whereas hmpv attack rates were significantly lower in the second season (p = .0004). the other viruses had similar attack rates during the 2 seasons. the only bacterial species detected by conventional culture in a significant number of acute ili cases was h. influenzae (15.4% in 2011-2012 and 11.3% in 2012-2013), frequently as the sole pathogen, while presence of other bacteria such as h. haemolyticus and s. pneumoniae was low. haemophilus influenzae was also detected frequently in recovery and control samples. we evaluated whether influenza vaccination reduced the overall influenza virus infection incidence and whether this influenced the incidence of ili. although influenza virus lost to follow-up n=20 no ili reported n=2240 per protocol no ili reported n=2108 a subject could have multiple ili episodes per season. an ili visit (v1) was considered "out of window" if the sample was taken >72 hours after start of fever. for the recovery visit (v2), the window was 7-9 weeks after ili onset. subjects were considered lost to follow-up if they did not respond to the end of study mailing and had no ili visit (a). after the baseline visit had been performed, a subject could have an ili event. subjects were considered lost to follow-up if they did not respond to the end of study mailing and had no ili visit (b). infection incidence was significantly lower in influenza vaccinated than in unvaccinated individuals (table 5) , the incidence of ili cases was not reduced by vaccination (table 1) . among participants with ili, we observed a high ve of 73% (95% ci, 26%-90%) in 2011-2012 and a moderate ve of 51% (95% ci, 7%-74%) in 2012-2013 against influenza virus during the influenza-active period (table 6 ). furthermore, the ve for the predominant influenza virus subtype a(h3n2) in 2011-2012 was 71% (95% ci, 19%-90%). in 2012-2013, the ve against influenza virus type a(h3n2) was 67% (95% ci, 20%-86%). additional sensitivity analyses for multiple ilis, households with ili, and the presence or absence of other virus infections did not affect this conclusion (supplementary table 2 ). in this study in a cohort of community-dwelling older adults in the netherlands, we show that influenza virus was present in 18.9% and 34.2% of ili cases in 2 consecutive seasons and that influenza vaccination significantly reduced laboratory-confirmed influenza virus infection. in 60.8% and 44.7% of the acute ili cases, potential pathogens other than influenza virus were detected. in addition, these pathogens were more often present during ili than after recovery or in asymptomatic elderly persons. in 20% of the ili cases, no potential pathogen was detected. the incidence of ili cases was expected to decrease by a reduction in influenza virus-caused ili through vaccination; however, this effect was not observed. instead, the incidence of c individuals in the asymptomatic subset were selected to be evenly distributed over the different age groups; therefore, the overall vaccination level was higher in the asymptomatic subset compared to the ili and non-ili groups. ili remained the same between the vaccinated and nonvaccinated individuals. influenza vaccination is offered to individuals 60 years and older in the netherlands, as older adults are at increased risk for morbidity and mortality from influenza virus infections due to an aging immune system and age-related chronic illnesses [19, 20] . however, this policy is based on studies mostly performed in elderly persons in nursing homes, who are substantially older and more frail compared with community-dwelling elderly individuals. studies in community-dwelling older adults are scarce [7] [8] [9] [10] , and only a restricted number of pathogens have been analyzed in these surveillance studies. to fill the gap in knowledge, we recruited participants through general practitioners and the civil registry, resulting in >99% of the participants living in the community. in line with the risk-based vaccination strategy in the western world, older participants and participants with chronic conditions were more likely to be vaccinated [2] . the age distribution in the cohort was similar to that observed for those ≥60 years in the general dutch population; only the oldest age group (>80 years) was underrepresented (table 2) . these elderly persons are often less likely to participate in studies. "healthy user effect" or "frailty selection bias" is a well-known issue in observational influenza vaccine studies, as the very frail elderly persons are difficult to reach [21, 22] . we found that influenza virus is involved in 18.9%-34.2% of ili cases in the 2 seasons studied. in addition, coronaviruses, rhinoviruses, hmpv, rsv, parainfluenza viruses, and h. influenzae were frequently observed as the sole detected pathogen in acute ili cases, whereas they were low to absent in most of the recovery samples. however, rhinoviruses and the bacterium h. influenzae were also commonly detected in asymptomatic controls. the rhinoviruses may be of different subtypes, as we did not type these viruses. viruses can often be detected in individuals without clinical manifestations [7] , but when they are found as the sole agent in the context of disease, they are commonly considered to be the cause of the disease. for bacteria, this is less clear as they are commonly carried in asymptomatic persons but may expand during respiratory viral infection or new acquisition. haemophilus influenzae may induce an enhanced inflammatory state in the presence of other pathogens, as was shown in children [23] . this may lead to ili. however, in most ili cases where we found h. influenzae, we did not find a second pathogen that could also explain the ili symptoms. influenza virus incidence in our cohort broadly matched the incidence reported in the dutch sentinel surveillance system [11] , reported a high ili incidence and mortality in the 2011-2012 season [24, 25] . this underlines a potential difference between the generally healthy community-dwelling elderly persons and that of the generally more frail institutionalized elderly persons, who may overlap in age. it also shows that data acquired in one group do not necessarily apply to the other, including exposure and susceptibility to infection, as well as influenza ve. we show that the incidence of influenza virus infection was reduced in individuals who received influenza vaccination and that ve to laboratory-confirmed influenza virus infection was high to moderate in the 2 seasons. the 95% cis of the data are wide, probably due to the relatively small sample size, a common problem in similar studies. ve estimates are notoriously variable between studies [18], but the ve data from this study are in line with those reported by van der hoek et al [18] . the most striking finding in this study was the similar incidence in ili cases observed between vaccinated and the nonvaccinated individuals. this may be explained by assuming that a pool of people exists that is highly susceptible to respiratory infections. the reduction by vaccination in the number of cases caused by influenza virus infections is offset by a rise in the number of cases caused by infections by other pathogens. cowling et al have described a similar increased risk of noninfluenza respiratory virus infection in influenza-vaccinated children [26] . however, we cannot attribute this effect to a specific pathogen. influenza virus may take preference over other viruses and prevent them from filling the niche, possibly by inducing a prolonged antiviral state, as has been described for other viruses [23, 27] . when vaccination reduces influenza virus infections, the other pathogens can fill the gap. it needs to be confirmed in other studies whether influenza virus vaccination has no effect on the total number of ili cases. it would have important consequences for decisions on implementation of vaccination for community-dwelling older adults when looking at the overall disease burden and cost-effectiveness. however, a limitation of our study is that we did not directly monitor the duration and severity of disease in the ili cases: difference in disease or hospitalization was only registered in post-ili questionnaires, and no significant events were reported. more detailed data on severity and duration of symptoms would allow assessing the relative risk posed by different pathogens, taking into account that influenza virus, unlike most other potential pathogens, can vary in pathogenicity between seasons. in combination with other medical information, it may then be possible to assemble a profile of individuals potentially at risk for ili or worse. such a profile would be of great value to public health professionals. a limitation of this study is the definition of ili used, which differs between the world health organization, the european centre for disease prevention and control, and different countries, although most include fever and cough [28] [29] [30] . in this study we used the dutch pel criteria [13] , which includes fever. it has been described that this specific ili definition can affect the number and type of pathogens detected. falsey et al showed that fever is more frequently associated with influenza virus infection in the elderly persons compared with other respiratory infections [7] . cough is not a prerequisite in the pel criteria, but as >80% of the participants with ili and >90% of the participants with influenza virus-positive ili reported coughing (data not shown), it is unlikely that cases were missed due to this difference in definition. in summary, we show that influenza virus caused between 18.9% and 34.2% of ili cases in community-dwelling older adults aged ≥60 years in 2 influenza seasons in the netherlands, leaving the remainder caused by other pathogens. we also show that influenza vaccination was effective in reducing the incidence of influenza virus infections but did not reduce the ili incidence, which may have important public health and healthcare consequences. our data will also help to better for more details, see supplementary table 3 . abbreviations: ci, confidence interval; ve, vaccine effectiveness. a corrected for the possible confounders age group, comorbidity, sex, and smoking. data were calculated for the influenza-active period in the netherlands as defined by the netherlands institute for health services research [11] . inform the public what to expect from influenza vaccination and how it will not protect against all cases of ili, popularly seen as "flu. " supplementary materials are available at the journal of infectious diseases online. consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. notes influenza (seasonal) social determinants of health and seasonal influenza vaccination in adults ≥65 years: a systematic review of qualitative and quantitative data effectiveness of seasonal influenza vaccine in community-dwelling elderly people: a meta-analysis of test-negative design case-control studies the efficacy of influenza vaccination in elderly individuals. a randomized double-blind placebo-controlled trial vaccines for preventing influenza in the elderly cochrane re-arranged: support for policies to vaccinate elderly people against influenza respiratory syncytial virus and other respiratory viral infections in older adults with moderate to severe influenza-like illness acute viral infections of upper respiratory tract in elderly people living in the community: comparative, prospective, population based study of disease burden influenza vaccine effectiveness among elderly persons living in the community during the 2003-2004 season a case-control study of acute respiratory tract infection in general practice patients in the netherlands nivel influenza surveillance sentinel surveillance network on infectious diseases in nursing homes study group. absence of influenza a(h1n1) during seasonal and pandemic seasons in a sentinel nursing home surveillance network in the netherlands proefonderzoek naar de frequentie en de aetiologie van griepachtige ziekten in de winter 1963-1964 preparing the outbreak assistance laboratory network in the netherlands for the detection of the influenza virus a(h1n1) variant differentiation of influenza b virus lineages yamagata and victoria by real-time pcr effect of 7-valent pneumococcal conjugate vaccine on nasopharyngeal carriage with haemophilus influenzae and moraxella catarrhalis in a randomized controlled trial identification of haemophilus influenzae and haemophilus haemolyticus by matrix-assisted laser desorption ionization-time of flight mass spectrometry letter to the editor: influenza vaccine effectiveness: heterogeneity in estimates for the 2012/13 season frailty, inflammation, and immunity vaccines for the twenty-first century society mortality benefits of influenza vaccination in elderly people: an ongoing controversy effectiveness of influenza vaccine in aging and older adults: comprehensive analysis of the evidence dutch rsv neonatal network. respiratory syncytial virus and recurrent wheeze in healthy preterm infants excess mortality among the elderly in 12 european countries increased risk of noninfluenza respiratory virus infections associated with receipt of inactivated influenza vaccine does viral interference affect spread of influenza? world health organization. ili sari surveillance case definition overview of influenza surveillance in the united states european centre for disease prevention and control. influenza case definitions we gratefully acknowledge all participants for their time and commitment to the study. we thank the study staff at the spaarne hospital and the laboratory staff members at the regional laboratory kennermerland and the centre for infectious disease control at the national institute for public health and the environment.financial support. this work was supported by the dutch ministry of health, welfare and sport.potential conflicts of interest. e. a. s. has received research grants from pfizer and gsk, in addition to fees paid to wilhemina children's hospital/university medical center for advisory boards and participation in independent data monitoring committees for pfizer and gsk. r. v. has received research support from gsk and pfizer for vaccine studies and consulting fees for gsk. all other authors report no potential conflicts of interest. all authors have submitted the icmje form for disclosure of potential conflicts of interest. conflicts that the editors consider relevant to the content of the manuscript have been disclosed. key: cord-275605-mbiojk39 authors: benkouiten, samir; al-tawfiq, jaffar a.; memish, ziad a.; albarrak, ali; gautret, philippe title: clinical respiratory infections and pneumonia during the hajj pilgrimage: a systematic review date: 2018-12-04 journal: travel med infect dis doi: 10.1016/j.tmaid.2018.12.002 sha: doc_id: 275605 cord_uid: mbiojk39 background: the islamic hajj pilgrimage to mecca is one of the world's largest annual mass gatherings. inevitable overcrowding during the pilgrims' stay greatly increases the risk of acquiring and spreading infectious diseases, especially respiratory diseases. method: the medline/pubmed and scopus databases were searched for all relevant papers published prior to february 2018 that evaluated the prevalence of clinical symptoms of respiratory infections, including pneumonia, among hajj pilgrims, as well as their influenza and pneumococcal vaccination status. results: a total of 61 papers were included in the review. both cohortand hospital-based studies provide complementary data, and both are therefore necessary to provide a complete picture of the total burden of respiratory diseases during the hajj. respiratory symptoms have been common among hajj pilgrims over the last 15 years. in cohorts of pilgrims, cough ranged from 1.9% to 91.5%. however, the prevalence rates of the most common symptoms (cough, sore throat, and subjective fever) of influenza-like illness (ili) varied widely across the included studies. these studies have shown variable results, with overall rates of ili ranging from 8% to 78.2%. these differences might result from differences in study design, study period, and rates of vaccination against seasonal influenza that ranged from 1.1% to 100% among study participants. moreover, the definition of ili was inconsistent across studies. in hospitalized hajj pilgrims, the prevalence of pneumonia, that remains a major concern in critically ill patients, ranged from 0.2% to 54.8%. conclusions: large multinational follow-up studies are recommended for clinic-based syndromic surveillance, in conjunction with microbiological surveillance. matched cohorts ensure better comparability across studies. however, study design and data collection procedures should be standardized to facilitate reporting and to achieve comparability between studies. furthermore, the definition of ili, and of most common symptoms used to define respiratory infections (e.g., upper respiratory tract infection), need to be precisely defined and consistently used. future studies need to address potential effect of influenza and pneumococcal vaccine in the context of the hajj pilgrimage. ksa for several weeks throughout the month-long hajj season, presenting a major public health and infection control concern, and a challenge both for the saudi authorities, as well as for the national authorities of the countries of origin of the pilgrims. in addition to physical exhaustion, sleep deprivation [3] , and heat stress [4] , inevitable overcrowding, both in housing and ritual sites, especially in mina encampment (this is approximately a 3-kilometer square area where pilgrims are accommodated in air-conditioned semi-permanent tents, some with up to 50-100 people) and inside the sacred mosque in mecca (with up to six pilgrims per square meter) [5] , greatly increases the risk of acquiring and spreading infectious diseases [6] [7] [8] , especially respiratory diseases [9, 10] . to minimize the spread of infections during the pilgrimage or in the pilgrims' home countries upon their return, vaccination and non-pharmaceutical interventions are thus recommended by national and international public health agencies [11, 12] . we carried out a systematic review of cohort and hospital studies that reported the prevalence of clinical symptoms of respiratory infections and pneumonia among pilgrims during the hajj, and both their influenza and pneumococcal vaccination status, with the aim to provide data allowing the investigation of the impact of this large mass-gathering event on public health policies and services and to identify potential targets for preventive measures. this review was performed according to preferred reporting items for systematic reviews and meta-analyses (prisma) guidelines (http://www.prismastatement.org). the medline/pubmed and scopus databases were searched for all relevant papers published prior to february 2018, using the terms: in addition, the saudi epidemiology bulletin (http://seb. drupalgardens.com/) was hand searched for additional papers for inclusion. finally, the reference lists of reviewed articles were searched for additional relevant papers. for inclusion, the article had to meet the following criteria: (1) original study involving hajj pilgrims; (2) detailed description of the study population, including influenza and pneumococcal vaccination status when available; (3) clinical or self-reported respiratory symptoms and diseases. only articles published in english were included for review. we excluded cohort studies with less than 50 participants and case reports. we also excluded studies conducted among selected groups of individuals suffering from respiratory tract infections, due to lacking denominator data. the two authors independently performed the searches, screened titles/abstracts for eligibility, selected papers that appeared to be relevant according to the review's inclusion criteria, and reviewed each of the selected manuscripts in full. the data were extracted from the included papers by one reviewer (sb) and collected in the summary table that was included in the review. the extracted data were checked by the two authors (sb and pg) for accuracy. minor discrepancies were resolved by the authors' discussion. the search strategy initially yielded 391 records, of which 143 were duplicates. twenty-nine additional papers were identified through manual searches. of the 277 papers identified 183 records were excluded after screening the title and abstract. of the 94 full text articles reviewed, 61 were deemed suitable for inclusion in this review influenza-like illness (ili) was defined according to the presence of the triad of cough, subjective fever and sore throat. c ili was defined as subjective (or proven) fever plus one respiratory symptom (e.g. dry or productive cough, runny nose, sore throat, shortness of breath). d ili was defined as subjective (or proven) fever and at least one respiratory symptom such as cough, sore throat and rhinorrhea. e ili was defined as symptoms and signs such as: sudden headache, dry cough, high grade fever, myalgia, coryza, malaise and loss of appetite with an abnormal general appearance. f upper respiratory tract infections (urti) was defined as any person who reported having developed at least one of the constitutional symptoms (fever, headache, myalgia) and one of the local symptoms (running nose, sneezing, throat pain, cough with/or without sputum) after reaching mecca for the hajj or within 2 weeks from return to riyadh. g acute febrile respiratory infection (afri) was defined as the presence of subjective fever plus at least one respiratory symptom (cough, sore throat, runny nose or breathlessness). h two travelers who reported ''bronchitis'' as a symptom were also included. i ili was defined as fever plus sore throat and/or coughing. j common cold was defined as sore throat with coryzal symptoms, and low grade fever. k ili was defined as fever > 38.5°c, myalgia, low back pain, coryzal symptoms and cough. l acute respiratory infection (ari) was defined as one of the constitutional symptoms (fever, headache, myalgia) along with one of the local symptoms (running nose, sneezing, throat pain, cough with/without sputum, difficulty breathing). m ari was defined as any person suffering from at least one of the constitutional symptoms (fever, headache, myalgia) along with one of the local symptoms (runny nose, sneezing, throat pain, cough with/without sputum, difficulty in breathing) developing after reaching makkah for the hajj. n ili was defined as cough and fever > 38°c with or without the coryzal symptoms and myalgia. o ari was defined as any person suffering from at least one of the constitutional symptoms (fever, headache, myalgia) along with one of the local symptoms (runny nose, sneezing, throat pain, cough with/without sputum, difficulty in breathing) developing after reaching mecca for the hajj. p ili was defined as sore throat with either temperature ≥38.8°c or cough. q cough or sore throat or rhinorrhea or muscle ache or headache. according to the inclusion/exclusion criteria. the results of the search strategy are shown in fig. 1 . a total of 45 publications were identified. these studies were conducted among cohorts of pilgrims from the 1999 through the 2015 hajj seasons. the results of these studies are presented in table 1 . various study designs were used, including cross-sectional studies, case-control studies, and prospective cohort studies with follow-up of pilgrims, before, during and after the hajj. participants were from different countries and continents (africa, north america, asia, europe, as well as from australia), with the majority from iran, and they were recruited from different settings, including travel medicine clinics, vaccination centers, hajj travel agencies, international airports and transit zones, mecca's city and mina encampments. their numbers varied widely in these studies, ranging from 106 to 107,074. respiratory symptoms were common during the hajj. overall, the prevalence of cough ranged from 1.9% in domestic and international pilgrims in 1999 [13] to 91.5% in malaysian pilgrims in 2007 [14, 15] ( table 3) . more recent studies, conducted in different populations of pilgrims during the 2011-2014 hajj seasons, reported prevalence of cough ranging from 46.3% to 86.8% [16] [17] [18] [19] [20] [21] [22] [23] [24] . these studies also reported a comparable prevalence of sore throat ranging from 34.7% to 91% among pilgrims [16] [17] [18] [19] [20] [21] [22] [23] . in addition, many of these studies have investigated the epidemiology of respiratory tract infections among pilgrims by estimating the common prevalence of upper respiratory tract infection (urti), acute respiratory infection (ari) or influenza-like illness (ili), which were inconsistently defined across studies by a combination of general symptoms (e.g. cough, sore throat and fever). overall prevalence of ili varied in these studies from 8% to 78.2% [14] [15] [16] [18] [19] [20] [21] [22] [23] ( [47, 48] . however, the ili syndromic case definition used in the 2003-2004 study (ili was defined as cough and fever of more than 38°c with or without the coryzal symptoms and myalgia) [41, 42] was different with that used in the 2004-2008 study (ili was defined as symptoms and signs such as sudden headache, dry cough, high grade fever, myalgia, coryza, malaise and loss of appetite with an abnormal general appearance) [47, 48] . also, it is unclear from the 2005 study [46] if the definition used was consistent with those used in the two previous studies [41, 42, 47, 48] . in a recent large study, conducted among 3364 egyptian pilgrims between 2012 and 2015, the prevalence of ili was 30.4% (ili was defined according to the world health organization definition as the presence of measured fever of ≥38 c°, and cough; with onset within the last 10 days) [45] . other studies of different sizes (from 129 to 468) and design were conducted from 2007 through 2014 among different populations of pilgrims using a common ili definition (the association of cough, sore throat, and subjective fever). these studies have shown variable results, with overall rates of ili ranging from 8% to 78.2% [14] [15] [16] 18, 19, [21] [22] [23] 25, 26, 31, 35, 37] . thus, during the 2013 hajj season, while the highest prevalence of ili was observed among malaysian pilgrims, with a prevalence estimated at 78.2% [25] , a lower prevalence was observed among french pilgrims (47.3%) [18, 21] . coverage of seasonal influenza vaccination among pilgrims was evaluated in many studies, which have yielded varying results, with reported rates of influenza vaccination ranged from 1.1% to 100% [14, 15, 18, [21] [22] [23] 25, 26, 28, 29, [31] [32] [33] [34] 36, 37, [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [53] [54] [55] [56] . a variation over time in influenza vaccination coverage was observed, as exemplified by a rate of 10.5% observed in a survey of pilgrims from riyadh in 2003 [43] , but 94.4% in a similar survey in 2010 [32, 33, 36] . during the 2013 hajj season, influenza vaccination rates also varied according to pilgrims' country of origin [29] , with 20% observed among saudi pilgrims, 80% among qatari pilgrims, and 87% among australian pilgrims, while a study involving french pilgrims interestingly reported that none of them had received the 2013 influenza vaccine before departing for the hajj because the vaccine was not available at this time [18, 21] . the majority of the studies reported influenza vaccination coverage among pilgrims, but only 13 [18, 19, [21] [22] [23] 25, 27, 28, 31, [46] [47] [48] 55] reported their pneumococcal vaccination status, with rates ranging from 1.2% among a multinational cohort of 1676 pilgrims from 13 countries (from africa, asia, usa and europe) in 2013 [28] to 51.2% among a small study of 129 french pilgrims in 2013 [18, 21] . of the 61 publications that were included in this review, 16 specifically addressed ill hajj pilgrims at health care facilities from 1993 through 2014 hajj seasons. medical facilities included primary health care centers (phccs) and different specialized wards in tertiary care hospitals, including ear, nose and throat (ent) departments, intensive care units, emergency units, infectious disease units and unspecified medical units. pilgrim participants were included either as inpatients or outpatients. the results of these studies are summarized in table 2 . overall, the prevalence of upper respiratory tract infections (urti) ranged from 1.4% to 42.1% (table 3 ). this prevalence was 1.4% among 141 pakistani pilgrims who attended the king abdul aziz hospital in medina during the 1992 hajj [57] and 42.1% among 3087 saudi and non-saudi patients (47.5% of them were pilgrims) who attended the ent clinic at al-noor specialist hospital in mecca during the 2009 hajj [58] . pharyngitis was also frequently reported among ill pilgrims. thus, in this study of 3087 pilgrims during the 2009 hajj, the overall prevalence of pharyngitis was 45.7% [58] . more recently, in 2008, the prevalence of pharyngitis in a large cohort of 4136 outpatients patients from 82 nationalities who attended 13 randomly selected mina phccs (94.9% of whom were pilgrims) was found to be 23.7% [59, 60] , and 61% in a study of 1047 saudi and non-saudi patients (2.3% of them were inpatients) [61] . however, in this second study of 1047 patients, only 34.5% were pilgrims. on the contrary, lower prevalence rates of bronchitis were reported during the hajj (1.4%-9.6%) [59] [60] [61] [62] [63] . a recent retrospective cross-sectional multicenter study of 185 turkish inpatients (87.5% were pilgrims) who returned to turkey from the arabian peninsula countries between 2012 and 2014 reported a slightly higher prevalence of acute tracheobronchitis (13.6%) [64] . in addition, in this study, pneumonia was among the most common clinical diagnosis among the hospitalized hajj patients and represented about half of diagnoses [64] . as pneumonia remains a major concern in critically ill patients, most of them reported the prevalence of pneumonia among pilgrims [57, 59, 60, [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] , with reported rates ranging from 0.2% in 2008 in 13 randomly selected mina primary health care centers [59, 60] to 54.8% in 2004 in two icu in mecca [68] (table 3 ). the prevalence of pneumonia was not reported in 3 papers [58, 61, 72] . pneumonia was the second most common admitting diagnosis (22%) in a study of 140 patients admitted to the icus in four hospitals in mina during the 2004 hajj [68] . this result is further confirmed by a recent study of 452 critically ill hajj patients, of over 40 nationalities, admitted to 15 hospitals in 2009 and 2010. in this study, pneumonia was defined as the primary cause of critical illness (27.2%) of all icus admissions during the hajj [65] . also, in another prospective study of pilgrims admitted in two major icus in mecca for the 2004 hajj season, community acquired pneumonia (cap) was the commonest source of sepsis, 54.8% [66] . [57] acute bronchitis: 1.4% a upper respiratory tract infection (urti) was defined as an acute infection that includes tonsillitis, pharyngitis, laryngitis, sinusitis, otitis media, and the common cold. b acute tracheobronchitis was defined as a patient with dry cough and/or low-grade of fever (< 38°c), sub-sternal pain, and fatigue in the absence of opacities on chest x-ray. c acute exacerbation of chronic obstructive pulmonary disease (copd) was defined as an association with increased frequency and severity of coughing and/or shortness of breath and wheezing, increased amount of sputum production, and/or a change in appearance of sputum in a patient with copd. d was not defined. the purpose of this review was to provide syndromic surveillance data that may be useful, in conjunction with microbiological data that will be presented in further papers, for the surveillance of respiratory infections and pneumonia during the hajj. despite the fact that some of the included studies in our review were performed among small numbers of pilgrims and cannot be extrapolated, it is clear from this work that respiratory symptoms have been common among hajj pilgrims over the last 15 years, as evidenced by the high prevalence of cough (over 90%) among malaysian pilgrims during the 2007 hajj [73] . cough is a common symptom among pilgrims [16, 74] and likely results from crowded conditions during the hajj. this close contact among such individuals may increase the risk of the transmission of respiratory pathogens, and therefore may contribute to respiratory disease outbreaks. climatic conditions and air pollution in mecca and surrounding holy sites during the hajj [75] may also play a role. recent follow-up studies thus evidenced a significant acquisition of respiratory viruses, particularly rhinovirus, influenza virus, and coronaviruses other than middle east respiratory syndrome coronavirus (mers-cov), and of bacteria, including streptococcus pneumonia, hemophilus influenza, staphylococcus aureus and klesiella pneumonia by hajj pilgrims upon their return from the hajj [76, 77] . respiratory diseases are the most common diseases observed among pilgrims attending mina primary health care centers [59] and a major cause of hospital admission during the hajj [70] , with pneumonia a leading cause of admission to intensive care units [62, 68] , where they are responsible for about half of the cases of sepsis [66] . unfortunately, while numerous articles on hajj pilgrims were retrieved from our literature search, relatively few recent articles specifically addressed ill pilgrims in the context of hospital settings. the use of cohort studies allows investigators to evaluate the actual incidence of clinical events in hajj pilgrims since it provides a denominator, but may not identify and capture the prevalence of some underlying conditions and of severe forms of respiratory tract infections, which are more likely to be evidenced in hospital patient populations. conversely, hospital studies use data that may be biased, frequently lacking denominator values, and so probably overestimating the occurrence of severe illness. moreover, a hospital-based study will, by definition, not capture some minor illness cases that do not require hospitalization. the prevalence rates of cough, sore throat and subjective fever varied widely across the included studies. these differences may result from differences in study design that may lead to potential biases (for example bias related to the method of data collection, using either selfreport questionnaires or telephone interview), study period (with regards to the seasonality of respiratory viral infections), and rates of vaccination against seasonal influenza among study participants which may widely vary from one study to another, as described in this review. thus, all data regarding the pilgrims, including demographic data, medical history, clinical data and information on vaccination status and compliance with non-pharmaceutical preventive measures, should be carefully collected by using standardized questionnaires. in addition, in the context of syndromic surveillance for respiratory pathogens, data regarding the pilgrim's symptoms should be collected prospectively during face-to-face interviews by trained medical investigators who travel with the pilgrims. one important result of this review is the finding of a lack of consistency ili syndromic case definitions across included studies. thus, in a 2003 study (that did not fulfill the inclusion criteria for this review) [78] , of 1310 malaysian pilgrims who had a clinic visit for upper respiratory tract symptoms at five clinics during the 2000 hajj, with the aim of determining influenza vaccine effectiveness against clinically defined ili, 63% had ili (defined as sore throat in combination with either temperature ≥38°c or cough) and 14% had influenza by the cdc definition (defined as measured fever [≥100°f (37.8°c)] and a cough and/or a sore throat). only one of the studies reported here used the cdc definition of ili or the who definition (an acute respiratory infection with measured fever of ≥38 c°a nd cough, with onset within the last 10 days) [45] . in his paper, rashid et al. demonstrated the low sensitivity of the cdc criteria and proposed therefore the use of the triad of 'cough, sore throat and subjective fever' to clinically define ili at the hajj or other mass gatherings, since this new simple clinical case definition is more specific and sensitive than the cdc definition [79] . this definition was used over the last years by french [16, 18, 19, [21] [22] [23] 31, 37] , malaysian [14, 15, 25] , indian [26] and afghan [35] investigators leading cohort studies among hajj pilgrims, thus allowing more reliable comparisons of findings between studies (table 1) . respiratory diseases are a major concern during the hajj. nonpharmaceutical interventions (e.g., hand hygiene, wearing face masks, social distancing) are known to reduce the spread of respiratory viruses from person to person and are therefore recommended to pilgrims by public health agencies. although hand hygiene compliance is high among pilgrims, face mask use and social distancing remain difficult challenges. data about the effectiveness of these measures for preventing acute respiratory infections at the hajj are limited, and results are contradictory, highlighting the need for future large-scale studies [80] . in addition to non-pharmaceutical interventions, vaccination against influenza is recommended for all hajj pilgrims by the ministry of health of saudi arabia [11, 12] . differences in study design and heterogeneity in the ili definition across studies make it difficult to compare findings from different studies and inhibits the drawing of conclusions regarding the potential effects of this vaccination on related clinical symptoms of influenza disease. however, recent papers by alqahtani et al. and alfelali et al. found the influenza vaccine to be effective, respectively, against both laboratory-confirmed influenza [81] and clinical influenza [82] . as influenza vaccination is generally considered effective in reducing influenza-related infections, the scientific committee for influenza and pneumococcal vaccination guidelines (scipv) thus recommends, in its recent guidelines, an influenza vaccination for all people, especially those at high risk, at least 2 weeks before the hajj [83] . it also recommends, for the next hajj seasons that will take place from june to september, the administration (prior to the hajj) of the southern hemisphere influenza vaccine for pilgrims from the southern hemisphere (where influenza positivity rates are higher during this period). furthermore, as the influenza vaccine is not expected to be available for pilgrims from the northern hemisphere before these next hajj seasons, the scipv also recommends the administration of the southern hemisphere influenza vaccine for those pilgrims from the opposite hemisphere before the hajj [83] . because of the mismatching between circulating and vaccine strains that has frequently occurred since 2003 [84] , alfelali et al. recommends, when the composition of influenza vaccines differs and whenever logistically feasible, taking into consideration the dual vaccination of hajj pilgrims with both the southern and northern hemispheres' vaccines. however, such strategy is impaired by the frequent unavailability of the southern hemisphere influenza vaccine in the northern hemisphere. the issue of influenza vaccine availability to match southern and northern hemispheres was discussed by the saudi ministry of health in consultation with the who and it was recommended to use the available hemisphere strain as long as there is a match in circulating strains [85] . despite the risk of acquisition of s. pneumoniae during the hajj, there is currently no consistent guideline on the use of pneumococcal vaccine for hajj pilgrims across pilgrim countries of origin [86, 87] . thus, and because many of the hajj pilgrims are elderly and have chronic illnesses and underlying risk conditions for which pneumococcal vaccination is recommended [86] , the scipv also recommended, in its 2016 pneumococcal vaccination guidelines, pneumococcal vaccination of the atrisk population at the appropriate time before the hajj, using the 2 types of pneumococcal vaccines that are currently available: the 23valent polysaccharide pneumococcal vaccine (ppsv23) and the 13-valent conjugate vaccine (pcv13) [88] . however, it did not recommend providing a pneumococcal vaccine routinely to healthy persons aged less than 50 years, because of lack of evidence. in addition, it has been well demonstrated that the conjugate vaccine against s. pneumoniae targets the most virulent serotypes associated with invasive pneumococcal diseases (ipd) that are also associated with antibiotic resistance [89] . these arguments reinforce the need for compliance with current recommendations for vaccinating at-risk hajj pilgrims against ipd and influenza [89] . respiratory tract infections, including influenza, continue to be a major concern during the hajj. both cohort-and hospital-based studies provide complementary data and potentially useful information, and both are therefore necessary to provide a complete picture of the total burden of respiratory diseases during this mass gathering. large multinational follow-up studies are thus recommended for clinic-based syndromic surveillance, in conjunction with microbiological surveillance. matched cohorts ensure better comparability across studies, particularly in terms of origin of pilgrims and possible travelling conditions. however, the study design and data collection procedures should be standardized, to facilitate reporting and to achieve comparability between studies. furthermore, the definition of ili, and of most common symptoms used to define respiratory infections (e.g., urti), needs to be precisely defined and consistently used. future studies need to address the potential effects of influenza and pneumococcal vaccine in the context of the hajj pilgrimage. moreover, because of the mismatching between circulating and vaccine strains that has frequently occurred since 2003 [84] , alfelali et al. recommends, when the composition of influenza vaccines differs and whenever logistically feasible, taking into consideration the dual vaccination of hajj pilgrims with both the southern and northern hemispheres' vaccines. however, such strategy is impaired by the frequent unavailability of the southern hemisphere influenza vaccine in the northern hemisphere. despite the risk of acquisition of s. pneumoniae during the hajj, there is currently no consistent guideline on the use of pneumococcal vaccine for hajj pilgrims across pilgrim countries of origin [86, 87] . thus, and because many of the hajj pilgrims are elderly and have chronic illnesses and underlying risk conditions for which pneumococcal vaccination is recommended [86] , the scipv also recommended, in its 2016 pneumococcal vaccination guidelines, pneumococcal vaccination of the at-risk population at the appropriate time before the hajj, using the 2 types of pneumococcal vaccines that are currently available: the 23-valent polysaccharide pneumococcal vaccine (ppsv23) and the 13-valent conjugate vaccine (pcv13) [88] . also, it did not recommend providing a pneumococcal vaccine routinely to healthy persons aged less than 50 years, because of lack of evidence. respiratory tract infections, including influenza, continue to be a major concern during the hajj. both cohort-and hospital-based studies provide complementary data and potentially useful information, and both are therefore necessary to provide a complete picture of the total burden of respiratory diseases during this mass gathering. large multinational follow-up studies are thus recommended for clinic-based syndromic surveillance, in conjunction with microbiological surveillance. matched cohorts ensure better comparability across studies, particularly in terms of origin of pilgrims and possible travelling conditions. however, the study design and data collection procedures should be standardized, to facilitate reporting and to achieve comparability between studies. furthermore, the definition of ili, and of most common symptoms used to define respiratory infections (e.g., urti), needs to be precisely defined and consistently used. future studies need to address the potential effects of influenza and pneumococcal vaccine in the context of the hajj pilgrimage. none. the authors have no conflicts of interest to declare. the general authority for statistics in the kingdom of saudi arabia from hajj services to mass gathering medicine: saudi arabia formalizes a novel discipline hajj: journey of a lifetime social identification moderates the effect of crowd density on safety at the hajj hajj: infectious disease surveillance and control global perspectives for prevention of infectious diseases associated with mass gatherings health risks at the hajj respiratory tract infections during the annual hajj: potential risks and mitigation strategies respiratory tract infection during hajj health conditions for travellers to saudi arabia for the pilgrimage to mecca (hajj) -2015 health conditions for travellers to saudi arabia for the pilgrimage to mecca (hajj) health status of non-organized hajjees (muftaresheen) during 1420h, hajj season the prevalence of acute respiratory symptoms and role of protective measures among malaysian hajj pilgrims the association between pre-morbid conditions and respiratory tract manifestations amongst malaysian hajj pilgrims the inevitable hajj cough: surveillance data in french pilgrims high prevalence of common respiratory viruses and no evidence of middle east respiratory syndrome coronavirus in hajj pilgrims returning to ghana respiratory viruses and bacteria among pilgrims during the acquisition of streptococcus pneumoniae carriage in pilgrims during the 2012 hajj pilot randomised controlled trial to test effectiveness of facemasks in preventing influenza-like illness transmission among australian hajj pilgrims in 2011 lack of mers coronavirus but prevalence of influenza virus circulation of respiratory viruses among pilgrims during the 2012 hajj pilgrimage lack of nasal carriage of novel corona virus (hcov-emc) in french hajj pilgrims returning from the hajj 2012, despite a high rate of respiratory symptoms serological study of bordetella pertussis, mycoplasma pneumonia and chlamydia pneumonia in iranian hajj pilgrims with prolonged cough illnesses: a follow-up study the prevalence and preventive measures of the respiratory illness among malaysian pilgrims in 2013 hajj season influenza like illness (ili): prevalence and preventive practices among indian haj pilgrims of karnataka impact of the hajj on pneumococcal transmission mass gathering and globalization of respiratory pathogens during the 2013 hajj viral respiratory infections among hajj pilgrims in 2013 acute febrile respiratory infection symptoms in australian hajjis at risk of exposure to middle east respiratory syndrome coronavirus relative risk for influenza like illness in french hajj pilgrims compared to non-hajj attending controls during the 2009 influenza pandemic patterns of diseases and preventive measures among domestic hajjis from central patterns of diseases and preventive measures among domestic hajjis from central, saudi arabia protective practices and respiratory illness among us travelers to the occurrence of acute respiratory infection, diarrhea and jaundice among afghan pilgrims pattern of diseases and preventive measures among domestic hajjis from riyadh, 1431 h protective measures against acute respiratory symptoms in french pilgrims participating in the hajj of surveying respiratory infections among iranian hajj pilgrims hajj-associated acute respiratory infection among hajjis from riyadh effect of the use of face mask on hajj related acute respiratory infection among hajjis from riyadh, a health promotion intervention study the comparison of influenza vaccine efficacy on respiratory diseases among iranian pilgrims in the 2003 and 2004 seasons the incidence of influenza like illness and determination of the efficacy of flu vaccine in iranian pilgrims during hajj pilgrimage incidence of hajj-related acute respiratory infection among hajjis from riyadh, 1423 h the incidence of vaccine preventable influenza-like illness and medication use among pakistani pilgrims to the haj in saudi arabia cross-sectional survey and surveillance for influenza viruses and mers-cov among egyptian pilgrims returning from hajj during 2012-2015 comparison of mortality and morbidity rates among iranian pilgrims in hajj treatment and prevention of acute respiratory infections among iranian hajj pilgrims: a 5-year follow up study and review of the literature trend of diseases among s iranian pilgrims during five consecutive years based on a syndromic surveillance system in hajj incidence of hajj-associated febrile cough episodes among french pilgrims: a prospective cohort study on the influence of statin use and risk factors mers-cov but positive influenza viruses in returning hajj pilgrims effect of influenza vaccination on acute respiratory symptoms in malaysian hajj pilgrims time to revisit presumptions on the essentiality of influenza vaccination for hajj pilgrims: a prospective cohort study international pandemic 2009 influenza a (h1n1) infection among 2009 hajj pilgrims from southern iran: a real-time rt-pcr-based study health related experiences among international pilgrims departing through king abdul aziz international airport effects of flu vaccine, solely or accompanied by pneumovax-23 vaccine on clinical consequences of the respiratory diseases among iranian pilgrims in hajj influenza among u.k. pilgrims to hajj medical problems of pakistani pilgrims in saudi arabia impact of ph1n1 influenza a infections on the otolaryngology, head and neck clinic during hajj pattern of diseases among visitors to mina health centers during the hajj season, 1429 h (2008 g) pattern of diseases among visitors to mina health centers during the hajj season road map of an ear, nose, and throat clinic during the 2008 hajj in makkah, saudi arabia causes of hospitalization of pilgrims in the hajj season of the islamic year pattern of surgical and medical diseases among pilgrims attending alnoor hospital makkah infections in travellers returning to turkey from the arabian peninsula: a retrospective cross-sectional multicenter study clinical and temporal patterns of severe pneumonia causing critical illness during hajj severe sepsis and septic shock at the hajj: etiologies and outcomes pattern of diseases among visitors of health care facilities in madinah during hajj season, 1428 h (2007 g) causes of admission to intensive care units in the hajj period of the islamic year 1424 pattern of medical diseases and determinants of prognosis of hospitalization during muslim pilgrimage hajj in a tertiary care hospital. a prospective cohort study pattern of admission to hospitals during muslim pilgrimage (hajj) pattern of medical problems among haj pilgrims admitted to king abdul aziz hospital, madinah al-munawarah epidemiological pattern of diseases and risk behaviors of pilgrims attending mina hospitals, hajj 1427 h (2007 g) preference of treatment facilities among malaysian hajj pilgrims for acute respiratory symptoms hajj: health lessons for mass gatherings air quality in mecca and surrounding holy places in saudi arabia during hajj: initial survey hajj-associated viral respiratory infections: a systematic review circulation of respiratory pathogens at mass gatherings, with special focus on the hajj pilgrimage a case-control study of influenza vaccine effectiveness among malaysian pilgrims attending the haj in saudi arabia influenza and the hajj: defining influenza-like illness clinically non-pharmaceutical interventions for the prevention of respiratory tract infections during hajj pilgrimage vaccinations against respiratory tract infections at hajj hajj research team. changes in the prevalence of influenza-like illness and influenza vaccine uptake among hajj pilgrims: a 10-year retrospective analysis of data the saudi thoracic society guidelines for influenza vaccinations mismatching between circulating strains and vaccine strains of influenza: effect on hajj pilgrims from both hemispheres expected immunizations and health protection for hajj and umrah 2018 -an overview prevention of pneumococcal infections during mass gathering pneumococcal infections at hajj: current knowledge gaps the saudi thoracic society pneumococcal vaccination guidelines-2016 emergence of drug resistant bacteria at the hajj: a systematic review none. supplementary data to this article can be found online at https:// doi.org/10.1016/j.tmaid.2018.12.002. key: cord-318856-f0m3wuyj authors: hoogeveen, martijn j.; van gorp, eric c.m.; hoogeveen, ellen k. title: can pollen explain the seasonality of flu-like illnesses in the netherlands? date: 2020-10-22 journal: sci total environ doi: 10.1016/j.scitotenv.2020.143182 sha: doc_id: 318856 cord_uid: f0m3wuyj current models for flu-like epidemics insufficiently explain multi-cycle seasonality. meteorological factors alone, including the associated behavior, do not predict seasonality, given substantial climate differences between countries that are subject to flu-like epidemics or covid-19. pollen is documented to be allergenic, it plays a role in immuno-activation and defense against respiratory viruses, and seems to create a bio-aerosol that lowers the reproduction number of flu-like viruses. therefore, we hypothesize that pollen may explain the seasonality of flu-like epidemics, including covid-19, in combination with meteorological variables. we have tested the pollen-flu seasonality theory for 2016-2020 flu-like seasons, including covid-19, in the netherlands, with its 17.4 million inhabitants. we combined changes in flu-like incidence per 100k/dutch residents (code: ili) with pollen concentrations and meteorological data. finally, a predictive model was tested using pollen and meteorological threshold values, inversely correlated to flu-like incidence. we found a highly significant inverse correlation of r(224)= -0.41 (p < 0.001) between pollen and changes in flu-like incidence, corrected for the incubation period. the correlation was stronger after taking into account the incubation time. we found that our predictive model has the highest inverse correlation with changes in flu-like incidence of r(222) = -0.48 (p < 0.001) when average thresholds of 610 total pollen grains/m3, 120 allergenic pollen grains/m3, and a solar radiation of 510 j/cm2 are passed. the passing of at least the pollen thresholds, preludes the beginning and end of flu-like seasons. solar radiation is a co-inhibitor of flu-like incidence, while temperature makes no difference. however, higher relative humidity increases with flu-like incidence. we conclude that pollen is a predictor of the inverse seasonality of flu-like epidemics, including covid-19, and that solar radiation is a co-inhibitor, in the netherlands. makes pollen less airborne, and cools the bio-aerosol down. very high humidity levels (rh 98%) are even detrimental to pollen (guarnieri, 2006) . an rh 98% effect on pollen could thus provide an alternative explanation as to why flu-like incidence in tropical countries is higher during the rainy season, and reduced during the rest of the year. we hypothesize that pollen bio-aerosol has an inverse effect on flu-like incidence, including covid-19 (see figure 1 ), whereby pollen is known to be triggered and influenced by meteorological variables, which can then jointly explain the seasonality of flu-like incidence. this indirect explanation of the pollen effect is based on the fact that pollen bio-aerosol and uv light exposure lead to immuno-activation, and sometimes allergic symptoms, which seem to protect against flu-like viruses, or at least severe outcomes from them. the indirect pollen effect is explained by the spread of pollen bio-aerosol under sunny and dry conditions. further, it is unknown how viral bio-aerosol and pollen bio-aerosol interact with each other in the air, and whether anti-viral phytochemicals in pollen could then play a role in an alternative explanation. to further understand the impact of pollen as an environmental factor influencing the life cycle of flulike epidemics, the objective of this study is to determine the correlations of pollen and meteorological variables with (changes in) flu-like incidence and develop and test a discrete predictive model that combines pollen and meteorological co-inhibitors. our main hypothesis, therefore, is that pollen is the missing link, jointly explaining with certain meteorological variables, flu-like seasonality, and that a compound threshold based factorcombining detected flu-inhibitorsis a good unified predictor of such seasonality. regarding covid-19, we have limited ourselves to observing whether or not covid-19 at the tail-end of the 2019/2020 flu-like season is able to break with the flu-like seasonality pattern. to study the relationship between pollen and flu-like incidence in the netherlands, we used the public datasets of elkerliek hospital (elkerliek.nl) about the weekly allergenic, low-level allergenic and total pollen concentrations in the netherlands in grains/m 3 , whereby for 42 types of pollen particles the j o u r n a l p r e -p r o o f journal pre-proof numbers are counted and averaged per day per 1m 3 of air. the common burkard spore trap was used, through which a controlled amount of air was ingested. the applied classification and analysis method conforms to the eaaci (european academy of allergology and clinical immunology) and the ean (european allergy network) standards. allergenic pollen includes nine types of particles that are classified as moderate (corylus, alnus, rumex, plantago and cedrus libani) , strong (betula and artemisia), or very strong allergenic (poaceae and ambrosia). additionally, we included low-level allergenic pollen concentrations in addition to the allergenic ones because we assume that they may also have effects. low-level allergenic pollen includes the other 33 particles that are classified as nonallergenic to low-level allergenic (cupressaceae, ulmus, populus, fraxinus, salix, carpinus, hippophae, fagus, quercus, aesculus, juglans, acer, platanus, pinus, ilex, sambucus, tilia, ligustrum, juncaceae, cyperaceae, ericaceae, rosaceae, asteraceae, ranunculaceae, apiaceae, brassicaceae, urtica, chenopodiaceae, fabaceae, humulus, filipendula, and indet) . total pollen concentration is the sum of the average allergenic and low-level allergenic pollen concentrations. advantages of using the total pollen metric are that there are hardly any 0 values (only 3 out of 266), and we did not need to limit ourselves to just parts of the seasonal cycle, which might introduce subseasonal bias into our research. we also assumed that long-distance pollen transport is accounted for, as foreign pollen will also be counted by a pollen measuring station that works all year round. furthermore, we used the data from the dutch state institute for public health (rivm.nl) gathered by nivel (nivel. nl) about weekly flu-like incidence (who code "ili" -influenza like illnesses) reports at primary medical care level, per 100,000 citizens in the netherlands. primary medical care is the using the number of influenza-like reports per primary care unit divided by the number of patients registered at that unit. this is then averaged for all primary care units and then extrapolated to the complete population. the datasets run from week 1 of 2016 up to week 18 of 2020 (n = 226 data points) to include the recent covid-19 pandemic at the tail-end of the 2019/2020 flu-like season. to underpin the relative importance of covid-19: sars-cov-2 has been detected in the netherlands since week 9, 2020. according to the figures of nivel.nl (2020, see figure 2 ), from week 13 onward sars-cov-2 is the outcome of the (vast) majority of positive tests for patients at primary care level with flu-like complaints, and by week 18 100% of positive tests indicate sars-cov-2 (other tested viruses are five influenza a and b subtypes, rsv, rhinovirus and enterovirus). furthermore, we also included meteorological datasets from the royal dutch meteorological institute (knmi.nl), including average relative humidity/day, average temperature/day and global solar radiation in j/cm 2 per day as an indicator of uv radiation. these datasets were obtained from the knmi's centrally located de bilt weather station. next, we calculated the weekly averages for the same periods that featured in the other datasets. de bilt is traditionally chosen as it provides an approximation of modal meteorological parameters in the netherlands, which is a small country. furthermore, all major population centers in the netherlands, which account for around 70% of the total dutch population, are within a radius of only 60 kilometers from de bilt. we therefore assumed in this study that the measurements from de bilt are sufficiently representative for the meteorological conditions typically experienced by the dutch population. to test allergenic versus low-level allergenic pollen assumptions, against hay fever and pre-covid-prevalence of allergic rhinitis that is more or less similar to that in western europe, being around 23%, and frequently undiagnosed (bauchau & durham, 2004) . furthermore, it can be noted that the prevalence of allergic diseases in general in the netherlands is around 52% (van de ven et al, 2006) . datasets were complete, except that three weekly pollen concentration measurements were missing (1.3%). this was due to a malfunctioning monitoring station during week 26 of 2016, week 21 of 2017 and week 22 of 2019. these missing measurements appeared to be completely random. we imputed missing values to avoid bias and maintain power. we used a four-week surrounding average to estimate the three missing data points and thus avoid breaking lines in visuals. we checked that the missing data has no material impact on the results by comparing these averages with the data of previous years for similar periods, and by observing whether removal from statistical tests had any effect on outcomes and conclusions. regarding the incidence of flu-like symptoms, we calculated the weekly change compared to the previous week (δili=ili t -ili t-1 ). this was to obtain an indication of the flu-like epidemic life cycle progression, whereby a decline is interpreted as ro<1 and an increase as ro>1 (ro is the reproduction number of flu-like viruses). furthermore, to cater, in one time-series metric, for changes in flu-like incidence as well as for an incubation period of up to two weeks, we calculated a three-week moving average (3wma) of changes in flu-like incidence, of which two weeks are forward looking: (δili 3wma = (δili t + δili t+1 + δili t+2 )/3) thus, δili 3wma has on average a one week lag. a general advantage of a moving average is that it reduces statistical noise. it should be noted that whenever we use the term incubation time, we also mean to include reporting delay (estimated to be around 4.5 days). we have not assumed delay effects for meteorological variables or pollen concentrations, so we have not calculated moving averages for other time series. compared with our previous study (hoogeveen, 2020) , there is an overlap in datasets of less than 10%. the datasets are extended by the extension in time, the addition of meteorological datasets and non-allergenic pollen, and the introduction of newly calculated variables, such as total pollen j o u r n a l p r e -p r o o f journal pre-proof concentration, δili 3wma , the compound predictor and the log10 transformations on pollen, ili and the hay fever index. we formulated the following statistical null hypotheses for falsification. h1 0 : there are no inverse correlations for total pollen concentrations with flu-like incidence (corrected for incubation period). h2 0 : there are no inverse correlations between pollen and changes in flu-like incidence (δili or corrected for incubation time: δili 3wma ). h3 0 : there is no predictive significance of a discrete model's compound value, based on thresholds for pollen and meteorological co-inhibitors, related to changes in flu-like incidence (δili 3wma ). to understand the role of meteorological variables, to check whetherin our datasetsmeteorological variables show their well-established effects on pollen as assumed, and to select coinhibitors: h4 0 : meteorological variablessolar radiation, temperature and relative humidityhave no effect on pollen and/or flu-like incidence change (δili 3wma ). low-level allergenic pollen is sometimes known to have a slight allergenic effect. to understand how to interpret adding none-to-low-level allergenic pollen to the total pollen metric, we wanted to verify their effects on the hay fever index: h5 0 : low-level allergenic pollen has no effect on hay fever and (changes in) flu-like incidence. note that with the exception of h5, all hypotheses are related to potential causality: the temporal sequentiality (temporality) of the respective independent variables, and flu-like incidence corrected for incubation period. whenever we refer to temporality, we mean to indicate that the datasets behave as if there is causality, on the understanding that statistics alone cannot prove causality in uncontrolled settings. variables are presented with their means (m) and standard deviations (sd). we calculated correlation coefficients to test the hypotheses and to assess the strength and direction of relationships. as a sensitivity analysis, we also calculated the bootstrapped correlation coefficients. we used the full datasets, to avoid sub-seasonal bias, and by extending the number of years the distortions by incidental and uncontrolled events are supposed to be minimized. however, as a second sensitivity analysis, we removed from the datasets the autumn weeks between 42 and 50, which typically show low pollen concentrations of up to 20 grains/m 3 , which are applied to analyze the main outcome (h2 0 ). further, as a third sensitivity analysis we calculate correlations per individual time lag included in δili 3wma in relation to h2 0 . next, linear regression (f-test) on identified inhibitors and interactions was used descriptively to determine whether the relationship can (statistically) be described as linear, and to determine the equation using estimates and intercept values, and produce probability, significance level, f-value, and the multiple r squared correlation to understand the predictive power of the respective inhibitor. standard deviations and errors, and degrees of freedom (df) were used as input for calculating the 95% probability interval. we have reported in the text the outcome of statistical tests in apa style, adapted to journal requirements. for relationships that appear non-linearlogarithmic or exponential we have used the log10 function to transform the data if that makes the relationship appear linear, before re-applying linear regression. we have also used the log10 transformed datasets for the calculation of correlation coefficients, to correct for skewness. finally, we created a simple, discrete model resulting in one compound value, using selected flu-like inhibitors. this was to determine the optimal average threshold values for these inhibitors, which have between total pollen and flu-like incidence, including the first cycle of the covid-19 pandemic. furthermore, we can reject h5 0 in favor of our assumption that it makes sense to also include low-level allergenic pollen concentrations in our study. low-level allergenic pollen is inversely correlated to flulike incidence (r(221) = -0.37, p < .00001), especially when corrected for the 2 weeks incubation time (r(219) = -0.53, p < .00001). the fact that the correlations become stronger when taking into account incubation time, implies temporality. furthermore, we can also observe from figure 3 that flu-like incidence starts to decline after the first pollen bursts. moreover, flu-like incidence starts to increase sharply after pollen concentrations become very low or close to zero. this is a qualitative indication of temporality. furthermore, we can notice that the first covid-19 cycle behaved according to pollen-flu seasonality, at least does not break with it. when testing the impact on δili, the weekly changes in medical flu-like incidence, the extended dataset till 2020, including covid-19, shows a strong and highly significant inverse correlation with total pollen (r(226) = -0.26, p = 0.000063). therefore, we can falsify the null-hypothesis (h2 0 ) that there is no inverse correlation between the weekly pollen concentrations and weekly changes in flulike incidence (δili), including the period covering the first cycle of the covid-19 pandemic. this inverse correlation therefore provides further support for the alternative hypothesis that the presence of an elevated level of pollen has an inhibiting effect on flu-like incidence, and starts to immediately influence the direction and course of the epidemic life cycle. also, during the covid-19 dominated period of the last 9 weeks, it appears that flu-like incidence behaves according to the expected pollenflu seasonality. this strengthens the idea that covid-19 might itself be seasonal, like all other flu-like pandemics since the end of the 19 th century. also when studying other data from rivm.nl about covid-19 hospitalizations, we cannot conclude that covid-19 breaks through the seasonal barrier. for example, new covid-19 hospitalizations decreased from a peak of 611 on march 27 to just 33 on may 3, the last day of week 18. using the three-week moving average (δili 3wma ) of changes in flu-like incidence, the correlation coefficients become stronger and are again highly significant for total pollen concentration (r(223) = -j o u r n a l p r e -p r o o f journal pre-proof 0.41, p < 0.00001). the bootstrapped correlation coefficient calculation gives a comparable outcome (r(223) = -0.38, p < 0.0001). as a second sensitivity analysis, we used the reduced dataset (minus the weeks of low pollen activity) and again found similar correlations (r(191) = -0.44, p < 0.0001; bootstrapped r(188) = -0.44, p < 0.0001, ci 95% -0.46 to -0.25)). finally, as a third sensitivity analysis, we analyze each time lag included in the δili 3wma calculation separately. per individual time lag there are as well highly significant inverse correlations: as given before r(226) = -0.26, p = 0.000063 in case of no time lag (δili t ); r(225) = -0.22, p = 0.000713 in case of a time lag of one week (δili t+1 ); r(225) = -0.23, p = 0.000552 in case of a time lag of two weeks (δili t+2 ); and the bootstrapped correlations for these are similar. we can thus also reject the null-hypothesis (h2 0 ) that there is no inverse relationship between pollen and changes in flu-like incidence including incubation time (δili 3wma or δili t+1 or δili t+2 ). these correlations (see also figure 4 ) are a further indication of temporality and does not contradict the idea that covid-19 is subject to pollen induced fluseasonality. the fact that the correlation with δili 3wma is stronger than those for each of the included time lags might be an indication of the noise reduction effect of this moving average, and makes thus the compound effect of the three covered time lags more visible. linear regression analysis shows that pollen has a highly significant inhibitory effect on flu-like incidence change (δili 3wma ) of f(1, 222) = 37.1, p < 0.001 (see table 2 , line 1), as a further basis for using total pollen concentration as a predictor. a log10 transformation of pollen to compensate for visual non-linearity leads to a similar outcome: f(1, 219) = 43.87, p < 0.001 (see table 2 , line 4). at least visually, it is a good fit (see figure 4 ). of the meteorological variables, only solar radiation has a highly significant inverse correlation with changes in flu-like incidence (δili 3wma ): (r(224) = -0.25, p = 0.000156). thus, of the meteorological variables, when it comes to solar radiation and relative humidity the nullhypothesis (h4 0 ) can also be rejected, as they seem to effect the flu-like epidemic lifecycle. of these two, only solar radiation is a flu-like inhibitor in line with its positive effect on pollen concentration, its association with immune-activation and the effect that uv has on viruses. a univariate linear regression also shows the highly significant negative correlation for solar radiation on flu-like incidence change (δili 3wma ) (f(1, 222) = 14.43, p < 0.001 (see table 2 , line 2). as the correlation is weak (multiple r-squared = 0.06), we have interpreted solar radiation as a co-inhibitor in relation to pollen; as a stand-alone independent variable its effect is too weak to explain flu-like seasonality. taking into account all these findings, we developed a discrete, compound model in which we included the changes in flu-like incidence (δili 3wma ), a threshold value for solar radiation (k r ), and both pollen threshold values for allergenic (k ap ) and total pollen (k p ). we found that the compound model has the highest inverse correlation (r(222) = -0.48, p < 0.001) for the following threshold values: k r : 510 j/cm 2 , k ap : 120 allergenic pollen grains/m 3 , and k p : 610 total pollen grains/m 3 . the bootstrapped correlation coefficient calculation gives a comparable outcome (r(222) = -0.47, p < 0.0001). in line with the previous outcomes, the inclusion of relative humidity, low-level allergenic pollen or temperature did not improve the correlation strength of this model. furthermore, given that they showed no significant interaction effects with pollen, it was not necessary to take such interactions into consideration in the model. in each of the observed years, the now (re)defined pollen thresholds are passed in week 10 (± 5 weeks), depending on meteorological conditions controlling the pollen calendar and coinciding with reaching flu-like peaks, and again in week 33 (± 2 weeks), marking the start of the new flu-like season. there is a highly significant inverse relationship between our compound threshold-based predictor value with flu-like incidence change (δili 3wma ) of f(1, 222) = 65.59, p < 0.001 and a multiple r-j o u r n a l p r e -p r o o f squared correlation of 0.2281 (see table 2 , line 3). this confirms the usefulness of a discrete, pollen and solar radiation threshold-based model as a predictor of switches in flu-like seasonality, whereby the effect of pollen is stronger than that of solar radiation. as a consequence, we can reject the nullhypothesis (h3 0 ) that this compound pollen/solar radiation value has no predictive significance for flulike seasonality. first of all we will discuss the possible implications of the results for our theoretic model and alternative explanations. next, we will discuss our methods. we found highly significant inverse relationships between pollen and solar radiation and (changes in) flu-like incidence: a higher pollen concentration or an increase in solar radiation in the netherlands is related to a decline in flu-like incidence. this inverse correlation with pollen becomes stronger when the 2019/2020 period is included, which has been increasingly dominated by covid-19 during the last 9 weeks. given that more time will be needed to draw conclusions about whether the spread of covid-19 is seasonal or not, from the data in this study it can only be observed that covid-19 is not breaking with the flu-like seasonality pattern. alternatively, social distancing may have contributed to flattening both the flu-like epidemic and covid-19 pandemic curves at the tail-end of the 2019/2020 flu-like season. the dutch government imposed hygiene measures from march 9, 2020 onward and a mild form of a lockdown, that included social distancing, from march 11. such behavioral policies the highly significant inverse correlation between hay fever and flu-like incidence confirms that allergic rhinitis makes it more difficult for flu-like viruses to propagate. solar radiation, the only meteorological variable that has a co-inhibitive effect on changes in flu-like incidence, has a stimulating effect on aerosol pollen formation and is responsible for melatonininduced immuno-activation. relative humidity reduces pollen aerosol formation, and correlates positively with flu-like incidence. we did not specifically look at precipitation, but it might make sense to explicitly consider this independent variable, given that it reduces pollen dissemination. in our study we showed that temperature, aside from the fact that it influences pollen, has no predictive value for changes in flu-like incidence. therefore, its inverse correlation with flu-like incidence might be interpreted in a number of ways: a) as spurious: the common causal factor is solar radiation, or b) as a stressor that has immediate effects on the functioning of the immune system of already infected persons. when discussing the influence of meteorological variables, we assume that the associated behavioral aspects are covered. these are sometimes summarized as seasonal behavior, but this independent variable might have a cultural dimension that needs to be better understood. we showed that a compound value, based on threshold values for pollen and solar radiation, results in a stronger correlation with the flu-like lifecycle than the individual inhibitors. this model could form an empirical basis for understanding flu-like seasonality, its ro and reliably predicting the start and end of each flu-like cycle. given that behavior, in the form of hygiene and social distancing, is also widely seen as an inhibitor, it might be worthwhile to also include this factor in our compound value. this will probably lead to an even stronger predictor for the evolution of the reproduction number ro of flu-like epidemics, although this might be beyond explaining the seasonality effect itself. for as long as the level of herd immunity (fine et al., 2011) for covid-19 is still below required thresholds for ending pandemics (plans-rubio, 2012) , it might make sense to also include indications of herd immunity levels in the theoretic model. finally, despite air pollution not been seen as an inhibitor of flu-like incidence (coccia, 2020), it still might interact with pollen. a more complete theoretic model, controlling for the (interactions with) air pollution, could give more insight in how to interpret the findings of this or similar studies. in general, statistical research cannot prove causal relationships in uncontrolled environments, even if datasets seem to behave as if there is causality. such statistics, however, can provide indications and identify reliable predictors, help filter out bad ideas, and be the inspiration for testable hypotheses that can be verified in laboratory and other fully controlled experiments. with a predictor we mean that a reliable temporal relationship between two variables is identified, without yet having validated causality, i.e., a bellwether factor. although the datasets seem to be sufficiently representative, there appears to be room for improvement. for example, including the data of more weather stations might help to improve the approximation of the weather conditions the dutch population experiences on average, and help to distinguish patterns per province. furthermore, it might be useful to include wind speeds, given that these constitute a vector for the dispersal of pollen in the netherlands, which has a maritime and temperate climate. additionally, the effects of climate change on pollen maturation (frei & gassner, 2008 ) might also be an important factor. another example of improving the representativeness would be by including more pollen types in the particle counts than are currently covered by the current methodology of the european allergy network. further, reclassification or recalibration of pollen types on a rational scale in terms of allergenicity, let's say 0-100%, would be very useful. for example, if pollen types are identified with a reliable score of 0%, these could be used to differentiate impact of pollen eleven faces of coronavirus disease 2019 herd immunity": a rough guide seasonality in risk of pandemic influenza emergence towards a data-driven characterization of behavioral changes induced by the seasonal flu ripe pollen carbohydrate changes in trachycarpus fortunei: the effect of relative humidity climate change and its impact on birch pollen quantities and the start of the pollen season an example from switzerland for the period 1969-2006 pollen exposure weakens innate defense against respiratory viruses nivel zorgregistraties eerste lijn -surveillance -wekelijks bulletin over symptomen en aandoeningen op basis van gegevens van huisartsen why lungs keep time: circadian rhythms and lung immunity the vaccination coverage required to establish herd immunity against influenza viruses dynamical prediction of flu seasonality driven by ambient temperature: influenza vs. common cold current understanding of the pathophysiology of allergic rhinitis covid-19 transmission in mainland china is associated with temperature and humidity: a time-series analysis humidity and latitude analysis to predict potential spread and seasonality for covid absolute humidity and pandemic versus epidemic influenza the role of temperature and humidity on seasonal influenza in tropical areas: guatemala, el salvador and panama global influenza seasonality: reconciling patterns across temperate and tropical regions potential utility of melatonin in deadly infectious diseases related to the overreaction of innate immune response and destructive inflammation: focus on covid-19 & asfiang p. correlation between weather and covid-19 pandemic in jakarta atopic diseases and related risk factors among dutch adolescents receptor recognition by the novel coronavirus from wuhan: an analysis based on decade-long structural studies of sars coronavirus association between ambient temperature and covid-19 infection in 122 cities from china no association of covid-19 transmission with temperature or uv radiation in chinese cities clinical characteristics of 140 patients infected with sars-cov-2 in wuhan key: cord-028048-0oqv2jom authors: rguig, ahmed; cherkaoui, imad; mccarron, margaret; oumzil, hicham; triki, soumia; elmbarki, houria; bimouhen, abderrahman; el falaki, fatima; regragui, zakia; ihazmad, hassan; nejjari, chakib; youbi, mohammed title: establishing seasonal and alert influenza thresholds in morocco date: 2020-06-29 journal: bmc public health doi: 10.1186/s12889-020-09145-y sha: doc_id: 28048 cord_uid: 0oqv2jom background: several statistical methods of variable complexity have been developed to establish thresholds for influenza activity that may be used to inform public health guidance. we compared the results of two methods and explored how they worked to characterize the 2018 influenza season performance–2018 season. methods: historical data from the 2005/2006 to 2016/2018 influenza season performance seasons were provided by a network of 412 primary health centers in charge of influenza like illness (ili) sentinel surveillance. we used the who averages and the moving epidemic method (mem) to evaluate the proportion of ili visits among all outpatient consultations (ili%) as a proxy for influenza activity. we also used the mem method to evaluate three seasons of composite data (ili% multiplied by percent of ili with laboratory-confirmed influenza) as recommended by who. results: the who method estimated the seasonal ili% threshold at 0.9%. the annual epidemic period began on average at week 46 and lasted an average of 18 weeks. the mem model estimated the epidemic threshold (corresponding to the who seasonal threshold) at 1.5% of ili visits among all outpatient consultations. the annual epidemic period began on week 49 and lasted on average 14 weeks. intensity thresholds were similar using both methods. when using the composite measure, the mem method showed a clearer estimate of the beginning of the influenza epidemic, which was coincident with a sharp increase in confirmed ili cases. conclusions: we found that the threshold methodology presented in the who manual is simple to implement and easy to adopt for use by the moroccan influenza surveillance system. the mem method is more statistically sophisticated and may allow a better detection of the start of seasonal epidemics. incorporation of virologic data into the composite parameter as recommended by who has the potential to increase the accuracy of seasonal threshold estimation. seasonal influenza epidemics result in considerable annual morbidity and mortality, with an estimated 291,243 to 645,832 deaths per year globally [1] . associated with these seasonal epidemics are substantial economic losses due to absenteeism, lost wages and increased utilization of health care services [2] . the influenza-associated respiratory annual mortality rate for people aged 65 and older in morocco has been recently estimated by the us centers for disease control and prevention (us cdc) at 3.7 per 100,000 (95% credible interval of 0. 4-22. 3) [1] . the risk of hospitalization due to influenza is 5 to 10 times greater in high-risk populations in morocco (e.g., the elderly and people with chronic disease) than in the general population [3] . the most effective ways to prevent or mitigate these effects are through vaccination combined with appropriate clinical management of persons infected with influenza. optimal impact of vaccination campaigns is achieved by timing them prior to the beginning of the influenza season to ensure maximum coverage and protection among the population. likewise, a timely signal to healthcare providers that the influenza season is underway helps to guide their patient management decisions and to mitigate the effects of illness in the individual and in the community. local patterns of influenza virus circulation and seasonality may differ geographically, necessitating national estimates of seasonal influenza activity to inform public health guidance. national surveillance data is essential for understanding those patterns and establishing signals for the beginning of the influenza season and epidemic periods. establishing baseline activity, epidemic and alert thresholds is a useful tool to inform recommendations for timely influenza vaccination to lessen the burden of seasonal epidemics [4] . while several statistical methods are commonly used, there is no gold standard for calculating influenza epidemic thresholds. the methods developed to date vary in their complexity and calculate either time-varying or fixed thresholds. the simplest ones use visual inspection of historical data to create a fixed threshold indicating the expected level of activity throughout the year [5, 6] . statistical methods include regression models [7] [8] [9] [10] , time series methods [11] , adaptation of industrial control processes such as shewart charts [12] , cumulative sum (cusum) [13] and rate difference models [14] . methods that involve calculation of means and medians are of medium complexity but are practical as they may be simple to implement. the objective of this study was to evaluate the performance of two methods using means and medians to establish thresholds using data from the moroccan national influenza-like illness (ili) syndromic surveillance system. we compare the results of the world health organization averages method (who method) with the moving epidemics method (mem) which is recommended by both the who and the european centre for disease prevention and control (ecdc). as a complement to the thresholds using syndromic data, we also calculated a threshold using a composite parameter integrating both syndromic and virologic surveillance data. following these direct comparisons of the methodologies, we explored the best method for characterizing the 2017/ 2018 influenza activity. in 2004, the epidemiology department of the ministry of health of morocco launched a year-round public sector syndromic surveillance system for ili comprised of 412 primary health centers, with a catchment population of almost 12 million people. sites report weekly ili activity to the regional and central levels, where health officials aggregate the surveillance data. a case definition similar to the 1999 who ili case definition recommended for public health surveillance, defined as "a sudden onset of fever, a temperature >38°c and cough or sore throat in the absence of another diagnosis" was used from 2004 to 2015 [15, 16] . in 2015, morocco adopted the updated who standard ili case definition [5] developed in 2011 as "an acute respiratory illness with a measured temperature of ≥ 38°c and cough, with onset within the past 10 days" [17] . reporting includes the total number of ili consultations aggregated by gender and age group, as well as total outpatient consultations. the proportion of ili visits among all outpatient consultations is used as a proxy for influenza activity. in 2007, the moroccan national influenza center (nic) began a virologic surveillance system in both ambulatory and hospital sites to complement the syndromic system and provide data on laboratory-confirmed influenza activity [18] . after an interruption in data collection beginning in 2010, virologic surveillance was resumed in 8 sentinel sites in 2014. specimens were collected and characterized between september and june. enrolling patients from both out-and in-patient facilities allowed the integration of epidemiologic and virologic data representing the spectrum of illness from mild (ili) to severe (e.g. severe acute respiratory infection or sari) [17] . we used 11 seasons of syndromic surveillance data (2005/2006 to 2016/2017, excluding the 2009/2010 pandemic year from analysis as influenza activity was not reflective of a typical season); this was described elsewhere [19] . we compared two methodologies for establishing seasonal baseline activity and epidemic thresholds. we also compared the calculated thresholds with the observed weeks for the start and end of the 2017/2018 season. using three seasons of virologic ili surveillance data (2014/2015 to 2016/2017), we used the mem method to make calculations using the composite parameter recommended by who [20] ; this method estimates the proportion of laboratory-confirmed influenza ili consultations among all outpatient consultations, or the product of weekly ili consultations of total outpatient visits and weekly percentage of influenzapositive specimens among respiratory tests. the methods discussed in order to standardize country information on influenza activity, have raised basic concepts summarized in table 1 . the 2012 who global epidemiological surveillance standards for influenza (who manual) [5] included a simple method to establish an average epidemic curve to identify the beginning of the influenza season using national influenza surveillance data. this method characterizes the intensity of influenza activity each year and may be used to describe the seasonality of influenza virus circulation. using ili as a proxy for influenza virologic activity [21, 22] , we used weekly proportion of ili among all outpatient consultations as our indicator of influenza activity. with this method, we were able to produce an average epidemic curve. using data from the average epidemic curve, we used statistical measures of variance to establish an alert threshold. we determined the flat baseline for expected influenza activity throughout the year in order to develop an indicator for the onset of influenza season (seasonal threshold). sustained influenza activity (i.e., three consecutive weeks) above this baseline indicated the start of the influenza season or the epidemic period [5] . in the final step, moderate, high, and extraordinary intensity thresholds were estimated as described in the who pandemic influenza severity assessment manual [20] , (fig. 1) . the moving epidemic method (mem) [23] [24] [25] [26] [27] [28] is an alternative tool developed to help model influenza epidemics also using retrospective national surveillance data. it may be described as a combination ratedifference model that uses cumulative differences in mem software produces an average curve, lower interval, and higher interval. calculate the mean and standard deviation (sd) of the average epidemic curve. for each week, the alert threshold is 1.645 sd above the weekly ili% mean. ili% > 1.645 sd indicates high ili activity or outbreaks and may be used to characterize a severe season. a graph consisting of the alert thresholds for each epidemic week. median weekly ili% over all weeks (i.e., the average epidemic curve is not used). indicates the level of influenza activity that signals the start and end of the annual influenza season(s). for prospective surveillance: upper limit of the 95% onesided confidence interval of the arithmetic mean of the 30 highest pre-epidemic weekly ili% values. parameter value which marks the start of the epidemic period. for prospective surveillance: upper limit of the 95% onesided confidence interval of the arithmetic mean of the 30 highest post-epidemic weekly ili% values. the third of three consecutive weeks with ili% above seasonal threshold. indicates that influenza activity occurs consistently. for retrospective analysis of individual season data: see "length of epidemic period". the third of three consecutive weeks with ili% below seasonal threshold for retrospective analysis of individual season data: see "length of epidemic period". weeks from epidemic start to end. for retrospective analysis of individual season data: mem software uses a "maximum accumulated proportions percentage (map)" algorithm to split the season into three periods: a pre-epidemic, an epidemic, and a post-epidemic period. proportion of total cases that occurred during the epidemic period upper 40% limit of 1-sided ci of mean of all peak values. upper 40% limit of the one-sided confidence interval of the geometric mean of the 30 highest epidemic weekly ili% values. upper 90% limit of 1-sided ci of mean of all peak values. upper 90% limit of the one-sided confidence interval of the geometric mean of the 30 highest epidemic weekly ili% values. upper 97.5% limit of 1-sided ci of mean of all peak values. upper 95% limit of the one-sided confidence interval of the geometric mean of the 30 highest epidemic weekly ili% values. rates to determine epidemic periods and intensity of activity [27, 28] . using the free software r for statistical computing and graphics [25] and its open source user interface rstudio [26] , we uploaded our surveillance data via the mem application [23] , and fit the model using three steps. we first visually compared activity over the 11 seasons in order to compare the timing of peak activity and activity trends across seasons. the mem procedure has three main steps: first, the length, start and the end of the annual epidemics are determined, splitting the season in three periods: a pre-epidemic, an epidemic and a postepidemic period [27, 28] . in the second step, we built the model by using retrospective data from all 11 seasons. the mem app calculated the pre-epidemic threshold that marks the start of the epidemic period (analogous to the seasonal threshold in the who method). in the third step, medium, high, and very high intensity thresholds were estimated ( table 2) . using the app, we produced graphs of each season showing the preepidemic, epidemic and post-epidemic periods (fig. 2 ). in addition, as the assumption that ili activity is reflecting influenza virus circulation has limitations, we created a second seasonal threshold with this methodology using the composite parameter recommended by who for three seasons of virologic ili surveillance (fig. 3) . lastly, we calculated indicators of performance of the app to detect epidemics, using values from the model for sensitivity, specificity, positive predictive value, negative predictive value, percent agreement and the matthew correlation coefficient ( table 3 ). the application allowed us to optimize the model by searching the optimum slope of the map curve to optimize the goodness-of-fit of the model for detecting epidemics. the mem app calculates goodness-of-fit indicators in an iterative process using a cross-validation procedure [27] . true positives (tp) were then defined as values of epidemic period above the threshold, true negatives (tn) as values of the non-epidemic period below the threshold, false positives (fp) as values of the non-epidemic period above the threshold and false negatives (fn) as values of epidemic period below the threshold. the process was repeated for each season in the dataset and all tp, tn, fp and fn were pooled. to measure the performance of the threshold, the following statistics and definitions were used [27] : 1. sensitivity: the number of epidemic weeks above the pre-epidemic threshold and above the postepidemic threshold divided by the number of epidemic weeks (epidemic length). 2. specificity: the number of non-epidemic weeks below the pre-epidemic threshold and below the post-epidemic threshold divided by the number of non-epidemic weeks. 3. positive predictive value (ppv): the number of epidemic weeks above the threshold divided by the number of weeks above the threshold. the ili sentinel surveillance system is a public health activity organized by the ministry of health of morocco. personally identifiable data is excluded from this surveillance system; as a result, no request for authorization from the national ethics committees was required. indeed, the royal dahir n°1-15-110 dated august 4, 2015, promulgating the law n°28-13 relating to the protection of persons participating in biomedical research, provides for special provisions for non-interventional or observational researches as stipulated in its articles 2 and 26. when applying the who method to our 11 years of surveillance data, we estimated that the seasonal threshold was the point at which more than 0.9% of outpatient consultations were due to ili (table 2) . influenza activity crossed this threshold on average at week 43 and the beginning of the epidemic period would be declared after three consecutive weeks of activity above this threshold, on average at week 46. the typical epidemic period lasted 24 weeks, finishing at week 18, when activity was below the seasonal threshold for three consecutive weeks. the average peak activity occurred during week 3. seasons where ili activity regularly crossed the alert threshold may be characterized as severe ( fig. 1 and table 2 ). intensity thresholds were ili% of 2.13, 2.77 and 3.06% for moderate, high and extraordinary intensity thresholds) ( fig. 1 and table 2 ). the mem model produced an estimate that the average annual influenza epidemic period began on week 49, and that the epidemic period lasted on average 14 weeks. the epidemic threshold (corresponding to the who seasonal threshold) was higher, at 1.51% of ili patients among all outpatients. the average peak activity occurred during week 3, consistent with the estimate using the who method. intensity thresholds were of 2.12, 2.81 and 3.19% of ili patients among all outpatients for respectively medium, high and very high intensity thresholds ( fig. 2 and table 2 ). indicators related to the goodness-of-fit of the mem model for detecting the epidemics, using these retrospective data showed that the sensitivity of the mem epidemic threshold was 0.81 whereas the specificity was 0.92. positive predictive value was 0.71 and negative predictive value was 0.95 (table 3) . using three seasons of virologic data, we established a third seasonal baseline based on the composite parameter recommended by who, which integrated both laboratory-confirmed influenza and syndromic ili reporting (fig. 3) . this method allowed us to compare the results of characterizing seasonality using these data types to identify the beginning of the influenza season. applying the mem methodology to our combined data, we determined that the average epidemic began at week 50, average peak activity occurred at week 3 and the average epidemic period lasted 15 weeks. using this method, medium, high and very high intensity thresholds were set at 0.59, 1.5 and 2.05% of laboratory-confirmed ili patients among all outpatients ( fig. 3 and table 2 ). goodness-offit indicators showed a sensitivity of 76%, specificity of 95%, positive predictive value of 80% and negative predictive value of 93% (table 3) . 2018 ili data with the who/2018 ili data with the who thresholds, the curve overlapped the average epidemic curve and activity crossed the seasonal threshold during week 43 of 2017 and was sustained after this time, confirming that this was the start of the epidemic period (fig. 1) . the season peaked during the second week of 2018, 1 week earlier than the average identified by the who methodology (week 3); we observed peak activity of 2.14% of ili patients among all outpatients (fig. 1) . when using the mem method with ili proportions, the epidemic period began at week 47, or the end of november 2017. this finding indicated an early season, beginning 2 weeks before the average epidemic start week of 49. the season peaked at week 2 of 2018 (beginning of january), 1 week before the average peak week determined by mem (week 3), with peak activity above 2% of ili patients among all outpatients. this season was characterized as one of medium intensity (fig. 2) . when considering the composite parameter, the mem method showed that the epidemic period began at week 48, or the end of november 2017, with a sharp increase of the epidemic curve 2 weeks prior to the average start (week 50). the seasonal peak occurred at week 2 of 2018, 1 week before the average peak week (week 3), with peak activity above 1.25% of confirmed ili patients among all outpatients. this season almost reached the threshold for high intensity (fig. 3 ). the occurrence of the 2009 h1n1 pandemic highlighted the need for a robust and standardized method to make timely assessments of the severity of influenza activity that may be used as an indicator of an unusual event. who developed and began implementing a framework on pandemic influenza severity assessment (pisa) [20] in march 2017. member states are encouraged to establish influenza baseline and epidemic alert thresholds from surveillance data and to monitor and describe the severity of each influenza season (seasonal, epidemic or pandemic influenza) using these thresholds. for this purpose, a simple method proposed by the who was used [22, 29, 30] . who is now recommending mem, which is a more sophisticated method of reporting influenza activity adopted by the european centre for disease prevention and control [31] [32] [33] [34] and adopted by several countries from other regions [35, 36] . the analysis using the mem application with 11 seasons of syndromic surveillance data showed clear seasonality to ili activity and visual inspection of graphed data revealed a single seasonal peak per year. the data show seasonal peaks between december and march, varying by year, as described by barakat et al. based on visual analysis [18] , matching trends observed in other northern hemisphere countries [37] . the average seasonal peak in morocco occurs at week 3 (mid-january) using either method. the seasonal threshold established using the method described in the who influenza surveillance guidelines was lower than the epidemic threshold calculated by the mem method when ili proportions are considered (0.9% versus 1.51% of ili patients among all outpatients). the average epidemic start week was estimated to be earlier when using the who method, with an average start at week 46 versus week 49 or 50 by using respectively ili proportions or the composite parameter with the mem method. there is a three-to four-week difference between these 2 methods when describing the typical start to a season; the optimal timing of a seasonal influenza vaccination campaign might vary accordingly. public health officials must weigh the costs and benefits of the optimal campaign period. influenza vaccine administration is ideally timed at least several weeks prior to influenza virus circulation as antibody response is achieved on average 2 weeks post vaccination [38] . the average epidemic period estimated by the who method was longer compared with the mem method (24 weeks vs. 14 or 15 weeks respectively). there are few publications with estimates of the typical duration of an influenza season [37] . according to the available evidence, the duration of the influenza season in the temperate zone of the northern hemisphere, ranges 12-19 weeks in europe [39] . the goodness-of-fit calculations from the mem application indicate that the mem capacity for detecting epidemic activity had a sensitivity of 81% and a specificity of 92% when using ili proportions, implying that it is better for eliminating false signals than it is for detecting a true signal. our finding is similar to that of vega et al., who also found the sensitivity to be significantly lower than the specificity [27] . using cambodian surveillance data, ly et al. [30] also found that the who methodology appeared to have a higher sensitivity for detecting early epidemic activity, but a lower specificity than mem, implying a greater risk of signalling false starts to the season. timely detection of the start of seasonal epidemics may be important to alert health services and to mitigate morbidity, mortality and economic costs by allowing resource allocation and adjusting response measures to face the seasonal overload in the healthcare system. the public health implications for this difference between methodologies are that using the mem method without applying the seasonal threshold established using the who method, there is a risk of missing the beginning of the epidemic period and not providing timely guidance to clinicians to indicate influenza season has begun, and to manage patient treatment accordingly. using the lower who threshold for public health messaging regarding the beginning of the influenza season may pose the risk of a false alert and perhaps overprescribing antiviral medications. from another point of view, using a low seasonal threshold could influence decision-makers to recommend earlier vaccination. as our results showed that the seasonal threshold typically occurs between mid-november and mid-december in morocco, appropriate timing for vaccination could be about 1 month before this date. of note, the us advisory committee on immunization practices (acip) recommends that vaccination should be offered by the end of october, considering the unpredictability of timing of onset of the influenza season and concerns that vaccineinduced immunity might wane over the course of a season [40] . low seasonal thresholds may be crossed multiple times as was the case in our application of the who threshold for several seasons (2005/2006, 2006/2007, 2010/2011, 2011/2012 and 2013/2014 [not shown]), due perhaps to variability in reporting by the surveillance sites. because of this variability, it is possible that declaring the start of the influenza season after two or three sustained weeks of activity above the threshold as recommended by who, is a prudent option for considering influenza transmission as epidemic. the mem methodology, however, calculates the length of the epidemic period during each season separately in order to determine the average length. thus, the epidemic threshold calculated with the mem method could be preferable to that established with the who method. mem was first used in in the who european region to estimate epidemic period and intensity using a minimum of five historical seasons for the calculations and the target season [27] . despite the availability of only 3 years of virologic data in morocco, we followed a who recommendation to use the composite parameter with mem [20] . this allowed a clearer cut estimation of the beginning of the influenza epidemic period, characterized by a sharp increase in influenza-confirmed ili cases. when ili proportions are used, the two methods produce similar values for each intensity threshold considered in the pisa assessment of seasonal transmissibility; who has adopted the mem for this purpose. when comparing the highest weekly activity per season (the seasonal peak) to the intensity thresholds established by who and mem procedures, the 2017/2018 season was of moderate intensity ( figs. 1 and 2) . using the composite parameter, the 2017/2018 seasonal peak nearly reached the high intensity threshold, whereas this curve didnot cross the medium intensity threshold when using only ili proportions. our study has several limitations. first, the assumption that ili activity reflects influenza virus circulation is limited because of possible concurrent circulation of other respiratory viruses (e.g., rsv) [41, 42] . who recommends using a composite parameter defined as the product of the ili or ari proportion and the percentage positive for the transmissibility indicator of the pisa tools [20] . unfortunately, virologic data collected prior to 2014 was not consistently available for the period of our study as virologic surveillance was disrupted between 2010 and 2014. despite this limitation, our laboratory-confirmed data showed something different than the syndromic data as the start of the virologic activity occurs suddenly and is therefore clearly identified. it is obvious that the inclusion of virologic data increases the specificity of seasonal threshold estimation. according to the who guidelines [5] , "a combination of parameters may be preferable. for example, a seasonal threshold could be defined as the week in which the ili rate crosses a certain value and the percentage of specimens testing positive reaches a certain point". given the long life of our surveillance system, our data were limited by changes in data collection practices, inconsistency of reporting by surveillance sites, and variable access to primary health care. these problems are not unique to the morocco ili surveillance system, and we believe they are the nature of routine, sentinel surveillance. another limitation was the adoption of a new case definition in 2015, at which point we also relaunched our surveillance system using a new protocol. these changes may have affected the trends that we observed in ili activity from that year forward. since both methods we used to establish thresholds recommend using a minimum of three to five seasons of data, we would not have enough data to run the models if we used only data from 2015 onward. determining a gold standard for influenza epidemic and intensity thresholds has been a long-standing research question for both international organizations and country-level public health authorities, and there is no consensus on the best method [5, 27, 28, 37, [43] [44] [45] . both the who method and the moving epidemic method translate quantitative trend data into standardized qualitative intensity levels, which permit countries to determine if the current season is atypical or to assess country or regional differences in activity and intensity. both methods identified that the 2006/2007 season was the most active in morocco, excluding the 2009/2010 pandemic season according to non-published observations. both methods are coherent to identify excess activity or high intensity thresholds even though with adequate laboratory data mem with the use of the composite parameter, gives a theoretically better qualitative measure of the level of activity. this comparative study has shown that the threshold methodology presented in the who manual is simple to implement and easy to adopt for use by the influenza surveillance system in morocco or the national surveillance systems of other similar countries. mem is more statistically sophisticated and may provide a more accurate detection of the start of seasonal epidemics in temperate countries with clear seasonal circulation of influenza viruses, especially if virologic data are considered. whichever method is used, analysis of surveillance data will provide information about seasonal thresholds and epidemic curves that may help health care personnel in the clinical management of respiratory illness after the start of influenza season. establishing a seasonal threshold for influenza helps health authorities to identify suitable periods for annual vaccination campaigns and for health practitioners to administer influenza vaccines or prescribe influenza antiviral drugs. computerization of the influenza surveillance system improves timeliness and assessment of the intensity of the influenza epidemic early in its course will guide policymakers in ensuring the appropriate allocation of resources to control seasonal epidemics. estimates of global seasonal influenza-associated respiratory mortality: a modeling study influenza (seasonal) fact sheet 6 epidémiologie de la grippe et facteurs de risque d'infection respiratoire aiguë sévère au maroc influenza seasonality: timing and formulation of vaccines world health organization. global epidemiological surveillance standards for influenza overview of influenza surveillance in the united states epidemic features affecting the performance of outbreak detection algorithms modelling influenza epidemic -can we detect the beginning and predict the intensity and duration? can syndromic thresholds provide early warning of national influenza outbreaks? methods for monitoring influenza surveillance data a routine tool for detection and assessment of epidemics of influenza-like syndromes in france a statistical algorithm for the early detection of outbreaks of infectious disease do cusums have a role in routine communicable disease surveillance? public health rate-difference method proved satisfactory in estimating the influenza burden in primary care visits revision of clinical case definitions: influenza-like illness and severe acute respiratory infection ministerial circular n°54 dated on july 23, 2004 and having for object ministry of health of morocco, directorate of epidemiology and disease control el aouad r. influenza surveillance among outpatients and inpatients in morocco pandemic influenza a virus subtype h1n1 in morocco, 2009-2010: epidemiology, transmissibility, and factors associated with fatal cases world health organization. pandemic influenza severity assessment utilizing syndromic surveillance data for estimating levels of influenza circulation exploring a proposed who method to determine thresholds for seasonal influenza surveillance the moving epidemic method: the web mem application, technical manual v10 the moving epidemic method: guidelines to monitor influenza and other respiratory virus infections epidemics and pandemic the r project for statistical computing. the r foundation integrated development for r: rstudio influenza surveillance in europe: establishing epidemic thresholds by the moving epidemic method. influenza other respir viruses influenza surveillance in europe: influenza surveillance in europe: comparing intensity levels calculated using the moving epidemic method. influenza other respir viruses national influenza surveillance in the philippines from 2006 to 2012: seasonality and circulating strains establishing seasonal and alert influenza thresholds in cambodia using the who method: implications for effective utilization of influenza surveillance in the tropics and subtropics harmonizing influenza primary-care surveillance in the united kingdom: piloting two methods to assess the timing and intensity of the seasonal epidemic across several general practice-based surveillance schemes moving epidemic method (mem) applied to virology data as a novel real time tool to predict peak in seasonal influenza healthcare utilization. the scottish experience of the 2017/18 season to date influenza surveillance: determining the epidemic threshold for influenza by using the moving epidemic method influenza seasons assessment of two complementary influenza surveillance systems: sentinel primary care influenza-like illness versus severe hospitalized laboratory-confirmed influenza using the moving epidemic method establishing thresholds and parameters for pandemic influenza severity assessment evaluating tools to define influenza baseline and threshold values using surveillance data, egypt, season 2016/17 seasonality, timing, and climate drivers of influenza activity worldwide kinetics and humoral antibody response to trivalent inactivated split influenza vaccine in subjects previously vaccinated for the first time influenza activity in europe during eight seasons (1999-2007): an evaluation of the indicators used to measure activity and an assessment of the timing, length and course of peak activity (spread) across europe prevention and control of seasonal influenza with vaccines: recommendations of the advisory committee on immunization practices -united states, 2019-20 influenza season influenza interaction with cocirculating pathogens and its impact on surveillance, pathogenesis, and epidemic profile: a key role for mathematical modelling possible interference between seasonal epidemics of influenza and other respiratory viruses in hong kong establishing thresholds for influenza surveillance in victoria european centre for disease prevention and control. indicators of influenza activity detecting the start of an influenza outbreak using exponentially weighted moving average charts publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations we would like to thank dr. amgad elkholy and dr. mohamed elhakim from infectious hazard management (ihm)/emro at world health organization and dr. henry laurenson-schafer for organizing training sessions on statistical methods for analyzing data provided by influenza surveillance systems as well as pr. abderrahmane maaroufi, former director of epidemiology at the ministry of health of morocco. a pilot study to assess the historical surveillance data of influenza in morocco and to compare the who method authors' contributions ar and ic designed the study. ic performed data analysis, interpretation of results and drafted the manuscript. mmc helped with study design, data analysis, interpretation of results, and drafting of the manuscript. ar and my assisted with study implementation and provided oversight of study personnel. ho, ab, fef, zr and hi assisted with access to and interpretation of laboratory testing results. st and he helped with data collection and study design and implementation. cn read and approved the final manuscript. all authors have read and approved the manuscript. none.availability of data and materials datasets were collected by each participating site including the national influenza center and gathered on a pooled database at the direction of epidemiology and disease control of the ministry of health of morocco. data cannot be publicly shared due to internal regulations of the ministry of health of morocco. the datasets analyzed during the current study could be available from the corresponding author on reasonable request and with special authorization of the ministry of health of morocco. the ili sentinel surveillance system is a public health activity organized by the ministry of health of morocco. personally identifiable data is excluded from this surveillance system; as a result, no request for authorization from the national ethics committees was required. indeed, the royal dahir n°1-15-110 dated august 4, 2015, promulgating the law n°28-13 relating to the protection of persons participating in biomedical research, provides for special provisions for non-interventional or observational researches as stipulated in its articles 2 and 26. not applicable. the findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the us centers for disease control and prevention. the authors declare that they have no competing interests. key: cord-001521-l36f1gp7 authors: nan title: oral and poster manuscripts date: 2011-04-08 journal: influenza other respir viruses doi: 10.1111/j.1750-2659.2011.00209.x sha: doc_id: 1521 cord_uid: l36f1gp7 nan pandemic influenza h1n1 (h1n1pdm) virus of swine-origin causes mild disease, but occasionally is associated with acute respiratory distress syndrome and death. 1, 2 it is important to understand the pathogenesis of this new disease. previously we showed a comparable virus tropism and host innate immune responses between h1n1pdm and seasonal h1n1 influenza virus in the human respiratory tract, 3 however h1n1pdm virus differed from seasonal h1n1 influenza virus in its ability to replicate in human conjunctiva, suggesting subtle differences in receptor-binding profile and highlighting the potential role of the conjunctiva as an additional route of infection. we now compare the tropism and host responses elicited by pandemic h1n1 with that of related swine influenza viruses and a pandemic-swine reassortant virus in ex vivo and in vitro cultures of the human respiratory tract and conjunctiva. we have used recombinant virus to investigate the role of the hemagglutinin (ha) and neuraminidase (na) of h1n1pdm virus in its conjunctival tropism. these findings are relevant for understanding transmission and therapy. fragments of human conjunctiva, bronchi, and lung tissues were cut into 2-3 mm fragments within 2 h of collection and infected with influenza a viruses at a titer of 10 6 tcid 50 ⁄ ml. viruses investigated included h1n1pdm (a ⁄ hk ⁄ 415742 ⁄ 09), swine h1n2 virus (a ⁄ swine ⁄ hk ⁄ 915 ⁄ 04), which shares a common derivation for seven genes with h1n1pdm, a natural swine reassortant h1n1 (a ⁄ swine ⁄ hk ⁄ 201 ⁄ 10), which has acquired the na gene from h1n1pdm and other swine influenza h1n1 viruses. reverse genetics derived recombinant viruses with ha and na gene segments of seasonal h1n1 and pandemic h1n1 swapped were also studied. lung fragments were cultured at 37°c in culture plates; conjunctival and bronchial biopsies were cultured in air-liquid interface at 33 and 37°c respectively. tissue fragments were infected for 1 h and incubated for 1, 24, and 48 h post infection. infectious viral yield was assessed by titration in mdck cells. the infected tissues were fixed with formalin and analyzed by immunohistochemistry for influenza antigen. cytokines profiles induced by influenza virus infected respiratory epithelial cells in vitro were measured by quantitative rt-pcr and elisa. we found comparable replication in seasonal and pandemic h1n1 viruses in human respiratory tract, while the swine influenza a ⁄ swine ⁄ hk ⁄ 4361 ⁄ 99 (h1n1) virus and a ⁄ swine ⁄ hk ⁄ 915 ⁄ 04 (h1n2) virus failed to infect and replicate in human lung ex vivo culture, but it replicated productively in human bronchus ex vivo. interestingly, the swine reassortant influenza h1n1 (a ⁄ swine ⁄ hk ⁄ 201 ⁄ 10) virus (with the na from h1n1pdm) infected and productivity replicated in lung ex vivo and in vitro. pandemic h1n1pdm virus, but not seasonal h1n1 virus, was able to infect ex vivo cultures of human conjunctiva, suggesting subtle differences in receptor binding profile in h1n1pdm, seasonal viruses, and the swine related h1n2 viruses. using reverse genetics derived recombinant viruses, we were able to demonstrate that the ha and na segments of h1n1pdm, but not the polymerase genes, were required for the conjunctival tropism of h1n1pdm ( figure 1 ). in contrast with highly pathogenic influenza h5n1 virus, which induced high cytokine and chemokine decretion, the related swine viruses, a ⁄ swine ⁄ hk ⁄ 915 ⁄ 04 (h1n2), as well as the swine pandemic reassortant virus, a ⁄ swine ⁄ hk ⁄ 201 ⁄ 10 (h1n1) we studied were similar to h1n1pdm and seasonal influenza viruses in their intrinsic capacity for cytokine dysregulation. collectively, our results suggest that pandemic h1n1pdm virus differs in modest but subtle ways from seasonal h1n1 virus in its intrinsic virulence for humans, findings that are in accord with the epidemiology of the pandemic to date. the ha and na gene segments are key to the conjunctival tropism manifested by the h1n1pdm virus. the pandemic reassortant influenza h1n1 (a ⁄ swine ⁄ hk ⁄ 201 ⁄ 10) virus isolated from swine with the na from h1n1pdm shares with h1n1pdm the capacity for productive replication in lung ex vivo and in vitro. these findings are relevant for understanding transmission and therapy. isolation of influenza viruses from specimens is traditionally performed in two classical systems: embryonated chicken eggs and mdck cell culture. nevertheless, several publications are dedicated to the theme of alternative cell culture systems, which may be used for influenza virus isolation and cultivation. [1] [2] [3] this is in part because mdck cells are of animal origin, which means that they cannot be used as a proper model for estimating interactions between a human virus and a human cell culture as a host. a variety of human monolayer and suspension cell cultures have been tested on their capability to support influenza virus replication. among them, some support influenza a virus growth as well as mdck cells do, 4 others support replication of a virus, but do not enable the formation of mature viral particles, 5 whereas others show only a weak level of replication or are not permissive at all. 2 caco-2 cells, for example, represent a good substitute for mdck cells, because it has been shown that the rate of viral isolation in caco-2 cells is as effective as in mdck, and sometimes is even better. 6 the success of viral replication is determined not only by the cell culture type, but also by the virus itself. despite the accepted view that it is the type of receptor that defines the interaction between the virus and the host cell, there is evidence that it is not the only factor that predetermines the fate of the cell. 7 the fate of the infected cell can also differ. a series of articles show that apoptosis is the most probable mechanism of cell killing by influenza viruses. 8, 9 influenza a viruses of different subtypes induce apoptosis to a different extent (e.g. h3 viruses provoke more strong apoptotic response than h1 viruses do 10 ). nevertheless, it has been demonstrated that caco-2 cells do not follow the apoptotic pathway and die through necrosis. 11 the sjpl cell line also dies through necrotic pathway and not apoptosis. 12 the aim of our work was to compare growth characteristics of different flu viruses (e.g. avian, swine, and human) in various human and animal cell cultures and to evaluate their influence on cell culture growth. the parameters measured in the study were as follows: cytopathic changes of cell cultures following virus infection, hemagglutinin production, np synthesis, the dose-dependent effect of infection on cell proliferation, and the ability of viruses to induce apoptosis. influenza viruses used included: highly pathogenic avian h5n1 a ⁄ kurgan ⁄ 5 ⁄ 05, low pathogenic avian h5n1 a ⁄ gull ⁄ kostanai ⁄ 7 ⁄ 07, swine h1n1 a ⁄ swine ⁄ 1976 ⁄ 31, human h1n1v a ⁄ california ⁄ 07 ⁄ 09, human h1n1v a ⁄ saint-petersburg ⁄ 5 ⁄ 09, human h1n1 a ⁄ brisbane ⁄ 59 ⁄ 07, and human h3n2 a ⁄ brisbane ⁄ 10 ⁄ 07. the viruses were propagated in 10-days embryonated chicken eggs, the allantoic fluid was collected, the aliquots were made and stored at )70°c for further use. to evaluate tcid50 for each virus on all cell cultures, 96-well plates were used. the cells were seeded 0ae2 ml per well (concentration of 1-1ae5 · 10 cells ⁄ ml). the confluent 24-h old monolayer was used for viral inoculation. the cells were washed twice with serum-free medium, then 0ae05 ml of tenfold viral dilutions from viral aliquots were added and left for 45 min for contact at 37°c. the cells were then washed to remove the non-attached particles, and the wells were filled with tpck-trypsin (2 lg ⁄ ml)-containing medium without bovine fetal serum. the plates were observed daily for cytopathic effect, and the results were evaluated at 72 h after infection for cytopathic effect and by reaction of hemagglutination with suspension of chicken erythrocytes (0ae75%). infection of suspension cell cultures was done in centrifuge tubes. cells (concentration 3-5 · 10 5 ) were inoculated with viral dilutions (moi = 1-10). after 45 min of contact, cells were washed, resuspended in rpmi with trypsin and fetal serum, and seeded in 24-well plates (1 ml in each well). the results were fixed after 72 h, calculating the number of cells grown and estimating the rate of apoptosis by hoechst-33258 staining. 13 cells were grown in 24-well plates with seeding concentration 1 · 10 cells ⁄ ml. one millilitre of cell suspension was placed in each well, inoculated with viral dilution (moi = 1-1000) and left for 72 h. after, the cells were detached from plastic with versene and calculated in fuks-rosental camera to evaluate the number of cells. the monoclonal antibodies obtained in research institute of influenza towards viral nucleoprotein np were used following the standard protocol described in. 13 for all viruses tested, mdck turned out to be more permissive than sp cell culture. avian viruses, independently of their pathogenicity, replicated efficiently on both animal cultures tested. human h1n1 and h3n2 viruses demonstrated weaker replication in sp cells. the most significant differences were seen for swine influenza and pandemic h1n1v viruses which replicated in mdck cells at the rates comparable with other viruses, but showed poorer growth in sp cell line (see table 1 ). human cell lines displayed clear differences in their susceptibility to viruses of various origins. avian influenza viruses replicated in all cell lines except girardi heart, and the most intense replication rate was observed for ecv-304, l-41, and rd lines. a-549 and a-172 were poorly infected, as well as all suspension cell lines tested. seasonal human h1n1, as well as h3n2 viruses, replicated in all cell cultures tested, but the rate of infectivity was rather low in practically all cultures tested with the exception of rd and t-98g cell lines. strikingly, swine influenza virus and human pandemic h1n1v viruses didn't replicate well in any of human lines tested. a weak replication rate was observed in ecv-304, rd, and t-98g, but in general, human cell lines were the titers produced by swine and pandemic influenza viruses are shaded in grey. *low-pathogenic avian influenza virus; **highly-pathogenic avian influenza virus poorly susceptible to pandemic h1n1v. swine influenza virus differed because it infected weakly a-172 and girardi heart cell cultures, which was not the case for h1n1v viruses. our study has shown that all influenza viruses were able to induce apoptosis in the cell cultures tested. the degradation of chromatin found in the nucleus with hoechst-33258 staining was seen before the first symptoms of cytopathic effect (cpe) in monolayer of cells. in cell cultures where the cpe was not visible, high doses of virus still induced apoptotic response. the process of apoptosis is rather well studied in mdck cells and some other cell types, so we've focused on three human monolayer cell cultures that are relatively poorly studied: a-549, ecv-304, and flech. these cell cultures are less susceptible to viral infection, and besides, it was interesting to find out whether the viruses that do not cause any cpe do infect these cultures. a-549 turned out to be most sensitive to apoptotic response, while flech turned out to demonstrate weak reaction. time needed for apoptosis induction by different flu viruses also varied. the earliest apoptosis was noted for h5n1 and h3n2 viruses and h1n1 viruses induced apoptosis at about 20 h postinfection. it is well-known that apoptosis can be induced only by a reproducing virus, and that uv-kills viruses that are not capable of it. we tested whether swine and pandemic h1n1v viruses (that do not show cpe in these cultures) do replicate in them and induce apoptosis with the help of monoclonal antibodies against viral np. the obtained data show that they indeed do replicate in these cell cultures, as we observed np fluorescence, and that they also induce apoptosis (see table 2 ). we've shown earlier 13 thus, we've tested the ability of swine and pandemic h1n1v viruses in this aspect. it was shown that these viruses were comparable with the effect seen for seasonal h1n1 virus. moreover, swine influenza virus induced stronger apoptotic response in hemablastoid cell lines in comparison with pandemic h1n1v viruses, which also have a swine origin. we also checked the ability of flu viruses to influence monolayer cell cultures growth. the data clearly indicated that only ecv-304 endothelial line and t-98g glioblastoma line displayed cell proliferation in response to low moi. apoptosis wasn't registered in these stimulated cultures, apparently because the moi was very low. all the other monolayer cultures didn't respond to low moi by stimulation of their proliferation. interaction between an influenza virus particle and a host cell can follow several scenarios. cpe seen in infected cells is accompanied with high rates of viral particles production and leads to cell death. the death itself may be through apoptotic or necrotic pathways. 4, 8 also, infection process in low doses can stimulate cell proliferation -the effect seen for hemablastoid lines, histiocytes, peripheral blood cell lines, 14, 15 and in glioblastoma and endothelial cell lines as it was described here. considering the origin of ecv line, 16 these cells bear all the antigenic, biochemical, and physiological traits of umbilical cord and are actively used in pharmacological tests as well as glioblastoma cells; they also are of special interest for oncogenesis studies. table 2 . replication, apoptosis induction, and np synthesis of influenza viruses in a-549, ecv-304 and flech cell cultures. the numbers represent the log 10 tcid 50 ⁄ 0ae2 ml calculated by reed-muench method as described in. 17 the ()) symbol means that no cpe could be observed in any dilution and no hemagglutination could be registered. the (+) symbol means that apoptosis was observed with hoechst-33258 staining though the productive replication and production of progeny viruses in human cell lines was generally low, it is evident that viral infection does occur in these cells, even for swine and h1n1v viruses. it can be demonstrated by the presence of np de novo synthesis and by stimulation of virus-induced apoptosis. in fact, we observe a contradiction: avian influenza viruses actively reproduce in human cell lines, but we do not see their vast spreading in human population, while h1n1v viruses that hardly replicate in all human cultures tested have caused the latest pandemic. influenza viruses continue to cause problems globally in humans and their livestock, particularly poultry and pigs, as a consequence of antigenic drift and shift, resulting frequently and unpredictably in novel mutant and reassortant strains, some of which acquire the ability to cross species barriers and become pathogenic in their new hosts. long-term surveillance of influenza in migratory waterfowl in north america and europe have established the importance of anseriformes (waterfowl) and charadriiformes (gull and shorebird) in the perpetuation of all known subtypes of influenza a viruses. the available evidence suggests that each of the 16 hemagglutinin (ha) and nine neuraminidase (na) subtype combinations exist in harmony with their natural hosts, cause no overt disease, and are shed predominantly in the feces. 1, 2 in this study we determined the subtypes and prevalence of low-pathogenic influenza a viruses present on the territory of kazakhstan in 2004-2006 and further analysed the ha and na genes of these isolates in order to obtain a more detailed knowledge about the genetic variation of influenza a virus in their natural hosts. (institute for biological safety problems, gvardeiskiy, zhambyl oblast, kazakhstan)). samples that were identified as influenza a virus positive by matrix rrt-pcr were thawed, mixed with an equal volume of phosphate buffered saline containing antibiotics (penicillin 2000 u ⁄ ml, streptomycin 2 mg ⁄ ml, and gentamicin 50 lg ⁄ ml), incubated for 20 minutes at room temperature, and centrifuged at 1500 g for 15 minutes. the supernatant (0ae2 ml ⁄ egg) was inoculated into the allantoic cavity of four 9-day old embryonated hens' eggs as described in european union council directive 92 ⁄ 40 ⁄ eec. 3 embryonic death within the first 24 hour of incubation was considered as non-specific, and these eggs were discarded. after incubation at 37°c for 3 days the allantoic fluid was harvested and tested by haemagglutination (ha) assay as describe in european union council directive 92 ⁄ 40 ⁄ eec. in the cases where no influenza a virus was detected on the initial virus isolation attempt, the allantoic fluid was passaged twice in embryonated hens eggs. the number of virus passages in embryonated eggs was limited to the maximum two to limit laboratory manipulation. a sample was considered negative when the second passage ha test was negative. the subtypes of the virus isolates were determined by conventional haemagglutination inhibition (hi) test and neuraminidase inhibition (ni) test, as describe in european union council directive 92 ⁄ 40 ⁄ eec. 3 rna extraction and pcr with specific primers rna was extracted from infective allantoic fluid using rneasy mini kit (qiagene, gmbh, germany) according to the manufacturer's instructions. the rna was converted to full-length cdna using reverse transcriptase. the rt mix comprised 2ae5 ll of dmpc water, 5 ll of 5· first strand buffer (invitrogen), 0ae5 ll of 10 mm dntp mix (amersham biosciences), 2 ll of 50 mm uni12 primer, 32 u of rnaguard (amersham biosciences), 200 u of mmlv reverse transcriptase (invitrogen) and 5 ll rna solution in total volume of 25 ll. the reactions were incubated at 42°c for 60 minutes followed by inactivation of the enzyme at 95°c for 5 min. pcr amplification with ha and na gene specific primers was performed to amplify the product containing the full length ns gene. twenty-five microliter pcr-mix contained 1· platinum taq buffer (invitrogen), 200 lm dntp, 2ae5 mm mgcl 2 , 240 nm each of fw primer and rw primer, 1 u platinum taq dna polymerase (invitrogen) and 3 ll cdna. reactions were placed in a thermal cycler at 95°c for 2 min, then cycled 35 times between 95°c 20 seconds, annealing at 58°c for 60 seconds, and elongation at 72°c for 90 seconds and were finally kept at 8°c until later use. sequences of the purified pcr products were determined using gene specific primers and bigdye terminator version 3ae1 chemistry (applied biosystems, foster city, ca), according to the manufacturer's instructions. reactions were run on a abi310 tm dna analyzer (applied biosystems). sequencing was performed at least twice in each direction. after sequencing, assembly of sequences, removal of low quality sequence data, nucleotide sequence translation into protein sequence, additional multiple sequence alignments, and processing were performed with the bioedit software version 7ae0ae4ae1 with an engine based on the custal w algorithm. the phylogenetic analysis, based on complete gene nucleotide sequences were conducted using molecular evolutionary genetics analysis (mega, version 4ae0) software using neighbor joining tree inference analysis with the tamura-nei c-model, with 1000 bootstrap replications to assign confidence levels to branches. [4] [5] [6] [7] ha and na sequences obtained from genbank the ha and na gene was analyzed both with selected number of influenza isolates and in comparison with virus genes obtained from genbank were used in phylogenetic studies [22] . the nucleotide sequence data obtained in this study has been submitted to the genbank database and is available under accession numbers fj434373, fj436942ae1, fj434369, fj434370, gu982281-gu982284 for ha and fj434374, fj436943ae1, fj434371, fj434372, gu982285-gu982288 for na. avian influenza prevalence in our study h4, h5, and h13 influenza a virus subtypes were found to circulate at the same time, in the same geographic region in the kazakhstan. this finding most likely indicates the existence of a large reservoir of different influenza a viruses in kazakhstan. we analyzed the ha and na gene sequences of the eight influenza a viruses isolated in kazakhstan together with selected number of isolates, reported between year 1941 to 2008, and previously published in the genbank. 8 phylogenetic analysis of the h4 ha gene showed that all viruses separated into the american and eurasian lineages ( figure 1 ). an evolutionary tree suggests that north american isolates have diverged extensively from those circulating in other parts of the world. geographic barriers which determine flyway outlay may prevent the gene pools from extensive mixing. the lack of correlation between date of isolation and evolutionary distance suggests that different h4 ha genes co circulate in a fashion similar to avian h3 ha genes and influenza c genes, implying the absence of selective pressure by antibody that would give a significant advantage to antigenic variants. analysis of phylogenetic relationships among the ha5 ha genes reported in this study clearly shows that viruses belong to the western pacific flyway, one of the major migratory flyways in this region that have subsequently spread throughout eurasia. 9 these findings provide further evidence of the dynamic influenza virus gene pool in this region. along the western pacific migratory flyway, the influenza virus gene pool in the domestic waterfowl of southern china has 'mixed' longitudinally with viruses isolated from japan, mongolia, and siberia. however, it appears that there has also been 'mixing' latitudinally through overlapping migratory flyways, thereby facilitating interaction between the influenza virus gene pool in domestic waterfowl in the eastern and western extremities of the eurasian continent. this helps to explain the latitudinal spread of the qinghai-like (clade 2ae2) h5n1 virus in the last 2 years, while h5n1 outbreaks in korea and japan may represent the longitudinally transmitting pathway. 10 ha of subtype h13 so far has been found exclusively in shorebirds, such as gulls, and in a pilot whale (potentially a spillover from shorebirds), but not in other avian species that are natural hosts of influenza a virus, such as ducks and geese; therefore the study of the evolution of these viruses is very interesting. phylogenetic analysis h13 ha gene revealed three significantly different evolutionary lines: an american line, a european line, and a line comprising the isolates from america and eurasia. 11 further we analyzed na genes of influenza viruses (figure 2) . the na gene is important both because of its functional role in promoting the dissemination of the virus during infection, and because, like ha, it is a principal target of the immune system. it was shown that phylogeny of na genes of influenza have the same properties as hemagglutinin. na genes of kazakhstanian viruses belong to eurasian lineage of virus evolution. obtained data are important for surveillance and diagnostics because some of the lpai viruses examined in this study can infect and be shed by chickens and turkeys and may have epidemiology potential during further recombination with other influenza viruses. influenza virus is divided into different subtypes based on hemagglutinin (ha) and neuraminidase (na) on the virus surface. within each subtype, ha continues to mutate and produce immunologically distinct strains, as antigenic drift. the continuous mutation of influenza virus (iv) is important for annual epidemics and occasional pandemics of disease in humans. antigenic drift requires vaccines to be updated to correspond with the dominant epidemic strains. in humans, ivs show both antigenic drift frequently. in contrast, ivs from birds are in evolutionary stasis, 1 and they show little amino acid changes. 2, 3 the reason is that ivs in bird intestine are not subjected to strong immune selection. hemagglutinin (ha) gene of influenza a virus encodes the major surface antigen, which is the target for the protective neutralizing antibody response that is generated by infection or vaccination. in humans, influenza a viruses show antigenic drift with amino acid changes in the globular head of the ha so as to evade herd immunity of the population. on the contrary, avian influenza a viruses show evolutionary stasis in wild birds. h6 aivs have occurred frequently in chicken farms in the world. 4 although vaccination is not permitted, h6n1 aivs have circulated in taiwan for a time. 5 the seroprevalence in chicken flocks reaches about 50% in the field. h6n1 aivs invades internal organs, such as kidney and lung. 6 thus, viruses in chicken flocks are pressured into antibody selection. here, we report that h6n1 aivs in the field have showed evolutional changes instead of evolutional stasis. in response to requests from poultry farmers for diagnostic investigations of illness in poultry flocks, the authors did necropsy at the pen-site. after careful examination, tracheae were taken and kept in cold for virus isolation in the laboratory. for avian influenza virus isolation, trachea was homogenized 1:10 in tpb with antibiotics. the homogenate was frozen and thawed three times and then centrifuged at 1157 g for 15 minutes. the supernatant was passed through a 0ae45 lm filter. the homogenate was examined for the presence of virus by inoculation into five 9-to 11-day-old specific-pathogen-free (spf) chicken eggs for two passages. thirteen h6n1 aivs were isolated in this laboratory during 2000 and 2008 from different parts of taiwan. besides the viruses isolated in this laboratory, the ha sequences of 27 chicken h6n1 aivs were from the genbank. the accession numbers of hemagglutinin of aiv reference strains included in this study were as the following: g2 ⁄ 87, dq376619; g23 ⁄ 87, dq376620; 0824 ⁄ 97, dq376621; na3 ⁄ 98, dq376622; 165 ⁄ 99, dq376626; 0705 ⁄ 99, dq376624; ns2 ⁄ 99, dq376623; sp1 ⁄ 00, dq376628; 0329 ⁄ 01, dq376633; 1205 ⁄ 01, dq376630; 1212 ⁄ 01, dq376631; 1215 ⁄ 01, dq376632; 0208 ⁄ 02, dq376641; 0408 ⁄ 02, dq376642; pf1 ⁄ 02, dq376635; pf2 ⁄ 02, dq376636; pf3 ⁄ 02, dq376637; a37 ⁄ 02, dq376639; 0320 ⁄ 02, dq376638; 0107 ⁄ 02, dq376640; 0222 ⁄ 02, dq376634; 0706 ⁄ 03, dq376644; 1203 ⁄ 03, dq376645; ch1006 ⁄ 04, dq376649; 0114 ⁄ 04, dq376647; a342 ⁄ 05, dq376653 and 0204 ⁄ 05, dq376652. the viruses isolated were propagated in the allantoic cavities of 10-day-old embryonated spf eggs for 72 hour. the virus rna was extracted using qiaamp viral rna miniprep kit (qiagen) . six-week-old balb ⁄ c mice were injected emulsion intraperitoneally with 100 lg of purified and concentrated a ⁄ chicken ⁄ taiwan ⁄ 2838v ⁄ 00 (h6n1) virion with complete freund's adjuvant. every two weeks, the mice were boosted supplementary five times with 50 lg of virion in incomplete freund's adjuvant. when the mice were boosted, blood was collected from tail vein and tested by the western blot assay to check the antibody titers. the mice were then injected intraperitoneally with 50 lg of virion at week 8. five days after the last injection, the splenocytes in the mice were fused with myeloma cells (sp2 ⁄ 0-ag14). one week before fusion, the myeloma cell line was expended in dmem medium (hyclone laboratories, logan, ut) with 10% fetal bovine serum at 37°c to ensure they were in the exponential growth phase. the spleen cells from immunized mice were washed, harvested, and mixed with the previously prepared myeloma cells and fused by gradually adding 50% polyethylene glycol-1500. the resulting pellet was plated into 96 well tissue culture plates. only the fused cells grew in medium with hypoxanthine-aminopterin-thymidine (hat). with fresh medium replacement over 2 weeks, the hybridomas were ready for screening. hightiter monoclonal antibody (mab) preparations were obtained from the ascetic fluid of mice injected with the selected hybridoma clones. the antibody from mouse ascetic fluids was purified by precipitation with ammonium sulfate, then aliquoted and frozen at )70°c, avoiding repeated freezing and thawing. eventually, six mabs were obtained and named ch11-d10, eb2-b3, eb2-e5, eb2-f9, ff9-f5, and ff9-f7, respectively. the hi test was performed following a standard method. all the viruses were diluted twofold and reacted with 1% chicken erythrocytes in the v-bottomed microtiter plate by the hemagglutination test. after agglutination, four hemagglutinating units of a ⁄ chicken ⁄ taiwan ⁄ 2838v ⁄ 00 (h6n1) and ascetic fluids from the immunized mice of the six mabs were prepared for hi test. hi titers of 2 4 or more were regarded as positive. the cases submitted for diagnosis from chicken farms had respiratory signs, increase in mortality, or drop in egg production (e.g. egg production dropped from 10% to 40%). the extent of drop in egg production depended on the chicken ages. for example, the age of case 3511 was 27 weeks, a stage of increasing egg production. however, after h6n1 aiv infection, the egg production decreased 1% instead of increasing and then stayed at 65% for a week. the infected chickens showed signs of decreasing activity, anorexia from 150 g per bird to 90 g per bird, and respiratory signs. case 2829 showed infection in the second floor first and then transmitted to third and fourth floor, indicating that the virus transmitted by air or human movement. however, most cases showed air borne transmission from one flock to another in spite of enforcing restrictions of persons entering the poultry pens and changing clothes and booths. in most cases, males' mortality was higher than that of female pen mates. by comparing the sequences of ha of those h6n1 viruses, we found that amino acid changes in ha1 were higher than those in ha2, showing that antigenic changes on the globular head of ha molecule rather than randomly on the whole ha protein, indicating that h6n1 viruses in taiwan had been selected in the presence of antibody pressure. the aa residues and changes that showed yearly trends were the followings: a-5s, i19s, v30i, n35s, e39k, l45m, e54d, q66k, a94v, or t, s127n, s128r, k133n, y138d, n139t, s140i, g141d, l149v, i152v, g155e, t188n, g226s, a285v, k304e, d386n, i533m, and m535i. however, their significance on antigenic variation was previously unknown. by hemagglutinination inhibition (hi) assays, except mab ch11-d10, all other monoclonal antibodies elicited from 2838v ⁄ 00 showed different hi titers with the different h6n1 viruses (table 1 ). however, those 5 mabs showed negative hi to 2829 and 2831, the early h6n1 strains. this indicated that the epitopes recognized by those mabs were undergoing antigenic drift. introduction aquatic birds are recognized as the natural reservoirs of the influenza a virus as all known subtypes (h1-h16, n1-n9) have been found in them. phylogenetic analyses of influenza viruses found in other animals revealed that all were directly or indirectly derived from viruses resident in aquatic birds. 1 however, the prevalence, movement, and evolutionary dynamics of influenza viruses in these avian hosts have not been well defined. southern china was hypothesized to be an 'epicenter' for the generation of human pandemic influenza viruses as all major influenza pandemic viruses in the 20th century emerged from this region. 2 the ecological background that facilitates the occurrence of these pandemic influenza strains has not been fully explored. in the past two decades, four lineages, belonging to h5n1, h9n2, and h6n1 viruses, have become established and long-term endemic in different types of poultry in this region. [3] [4] [5] some of these viruses were disseminated to many countries in eurasia and africa and have continued to cause sporadic human infection, posing a persistent pandemic threat to the world. 6 in the mean time, the endemic influenza lineages have undergone extensive genetic reassortment events giving rise to many variants, dramatically increasing the genetic diversity of the influenza virus in this region. questions remain as to how and where these viruses emerged, and what were the sources of the gene segments incorporated within the novel reassortant variants of the h5n1, h9n2, and h6n1 virus lineages. to address these questions, surveillance of influenza in migratory and domestic (sentinel) ducks has been conducted since 2002 at poyang lake, the biggest fresh-water lake and the major migratory bird aggregation site in southern china. the aim of this study is to identify the prevalence, seasonality, and movement of virus between migratory and domestic ducks. migratory ducks were captured during over-wintering, from november to march. cloacal swabs and blood samples were collected from each individual bird. all birds were released after sampling. to observe the interaction between migratory ducks and domestic birds, we also sampled domestic ducks from two duck farms (designated as sentinel ducks) surrounded by rice fields and inaccessible to other types of poultry, but accessible to migratory birds. that is, the sentinel ducks share the same water body with migratory ducks and have the chance to spread viruses to each other. for sentinel ducks, sampling was conducted fortnightly, all year-round, on the two farms from august 2002 onwards. cloacal swabs and fresh fecal droppings were taken. about 200 birds were randomly sampled fortnightly from these farmed ducks. all swabs were soaked in vials containing 1ae5 ml transport medium with antibiotics and kept on ice-packs during sampling and immediately stored in )80°c freezers for further use. blood samples from migratory ducks were treated according to methods previously described. 7 serological survey and virus subtyping in migratory and sentinel ducks used hemagglutination inhibition (hi) and neuraminidase inhibition (ni) tests as previously described. 3 for isolates that were not identified by reference antisera, subtypes were determined by rt-pcr using subtype specific ha and na diagnostic primers. prevalence and seasonal patterns of influenza virus in migratory and sentinel ducks during 2002 during -2007 a total of 11 508 cloacal swabs from migratory ducks and 36 475 cloacal or fecal swabs from sentinel ducks were collected at poyang lake. from these specimens, 90 influenza isolates were obtained from migratory ducks and 1681 from sentinel ducks; isolation rates of 0ae78% and 4ae61%, respectively (table 1) . it was noted in sentinel ducks that virus occurrence formed a seasonal peak from november to february, which completely overlapped the over-wintering months of migratory ducks. this suggests that virus movement or transmission between migratory and sentinel ducks occurred during this period at poyang lake. thirty positive samples (hi titer ‡20) were identified from 715 blood samples collected during november and december in 2004. among these, 15 samples were positive to h6, 8 were positive to h9, 5 were positive to h1, and 2 were positive to h4. one serum sample was positive to both h4 and h6, which suggested co-infection of influenza virus in migratory ducks might occur in natural conditions. poyang lake, which is located in the northeastern part of jiangxi province, is the largest freshwater lake in china and is part of the eastern asia-australia migration route. 9 every year, hundreds of thousands of migratory ducks congregate at poyang lake during the migration season. 9 recent farming practice involves raising domestic waterfowl in dense populations in the poyang lake region. farmraised domestic waterfowl are allowed to feed in and share the same water body with migratory birds, thereby facilitating direct interactions between domestic waterfowl and freeranging migratory birds. this makes poyang lake an ideal site to observe the dynamics of influenza virus interactions between migratory and sentinel ducks in southern china. in our longitudinal surveillance during [2002] [2003] [2004] [2005] [2006] [2007] , the overall virus isolation rate from migratory ducks was less than 1%, which suggests a low prevalence of viral infection during the birds' southern migration. similar results have been observed in taiwan, which is also an important stopover site for migratory birds along the eastern asia-australia migration route during 10 years of surveillance. 10 the overlap in seasonal patterns of virus infection between migratory and sentinel ducks found in our study suggests that virus movement or transmission between migratory and sentinel ducks occurred during the period of time migratory birds were at poyang lake. the ha subtypes harbored in migratory and sentinel ducks were similar in our study. for migratory ducks, h4, h6, h10 were the predominant subtypes, while h3, h4, and h10 were the major subtypes in sentinel ducks. hpai h5n1 was only detected from migratory ducks in early 2005 on two sampling occasions. from phylogenetic analyses the h5n1 viruses isolated from migratory ducks were closely related to the viruses endemic in domestic poultry in southern china. 8 therefore, it appears that h5n1 viruses endemic in domestic poultry could be transmitted to migratory ducks via close contact in southern china. only lp h5 viruses were detected from sentinel ducks at poyang lake during this period. whether h5n1 virus infection was absent from sentinel ducks at poyang lake needs further investigation. serological surveys provided further evidence for the prevalence of aiv in migratory ducks at poyang lake. the serological results in 2004 did not match well with the epidemiological results during [2002] [2003] [2004] [2005] [2006] [2007] , which suggests that influenza virus infection in migratory birds could be influenced by multiple factors, such as host immune status, population size, spatial and temporal variations, and migration routes. southern china has the biggest domestic duck population in the world. our study demonstrates that dynamic interactions between migratory ducks and sentinel ducks occurred frequently throughout the surveillance period. thus, sentinel ducks could be treated as intermediate hosts between the ''real gene pool'' from migratory ducks and domestic poultry in the whole influenza virus ecosystem. a sentinel duck sampling system may be a feasible method to represent the viruses in the natural gene pool and a baseline for virus or gene interactions between migratory and domestic ducks. 11 further investigations and surveillance are required to better understand the role of the domestic duck population in facilitating virus interactions and the generation of genetic diversity. two distinct lineages of h9n2 influenza viruses represented by a ⁄ chicken ⁄ beijing ⁄ 1 ⁄ 94 (ck ⁄ bei-like) and a ⁄ quail ⁄ hong kong ⁄ g1 ⁄ 97 (g1-like) have become established and endemic in poultry in southern china. 1 these established h9n2 lineages continue evolving to generate many different reassortant variants (or genotypes) 1,2 and are causing sporadic cases of human infection. 3, 4 studies of h9n2 viruses isolated from pigs in hong kong and shandong province have also raised the possibility of reassortment with human-like viruses from pigs. 5, 6 in addition, h9n2 viruses isolated beyond the late 1990s had preferential binding with a-2,6-neuacgal human-like receptors. 7 these observations suggest that the h9n2 influenza viruses still have pandemic potential. unlike highly pathogenic h5n1 influenza viruses that have been rarely detected in the live-poultry markets in hong kong since 2002, h9n2 viruses are still frequently isolated in our surveillance program. therefore, we try to understand the continuing evolution of h9n2 viruses through genetic characterization and phylogenetic analyses of the viruses isolated in hong kong live-poultry markets from 2005 to 2009. a total of 47 421 terrestrial poultry were sampled at different live-poultry markets in the hong kong sar between january 2005 and december 2009. of those samples, 29 691 were from chickens and the others were from minor poultry species including chukar, pheasant, guinea fowl, silky chicken, and pigeon. fecal droppings, cloacal and tracheal swabs, drinking water, and environmental samples from cages were collected into transport medium. viruses were isolated in 9-to 11-day old embryonated eggs as described previously. 8 virus isolates from positive sampling occasions were selected for sequence analysis. rna extraction, cdna synthesis, and pcr were carried out as described previously. 8 dna sequencing was performed using bigdye terminator v3ae1 cycle sequencing kit on an abi 3730 dna analyzer (applied biosystems) following manufacturer's instructions. all sequences were assembled and edited with lasergene 7ae0 (dnastar, madison, wi) software. sequence alignment and residue analysis were performed with the bioedit sequence alignment editor, version 7ae0. 9 all eight gene segments of sequenced viruses were characterized and analyzed phylogenetically together with virus sequence data available in public databases. maximum-likelihood trees were constructed using garli 0ae96. 10 estimates of the phylogenies were calculated by performing 1000 neighbor-joining bootstrap replicates using paup*4ae0. 11 systematic surveillance of live-poultry in hong kong from 2005 to 2009 resulted in 1088 h9n2 isolates from 47 421 samples (overall isolation rate, 2ae3%) ( table 1 ). there were 966 strains isolated from 42 253 chicken samples (isolation rate, 2ae3%). of these viruses, four were isolated from 1556 tracheal swabs (isolation rate, 0ae3%), while 541 isolates were isolated from 29 030 cloacal or fecal swabs (isolation rate, 1ae9%). an additional 421 isolates were collected from 6926 drinking water samples (isolation rate, 6ae1%). there were 122 strains of h9n2 viruses isolated from 5168 minor poultry samples (isolation rate, 2ae4%) ( table 1) . of these viruses, only one was isolated from 1209 tracheal swabs (isolation rate, 0ae1%), whereas 72 strains of viruses were isolated from 3610 cloacal or fecal swabs (isolation rate, 2ae0%). the isolation rate in drinking water in minor poultry was again higher when compared with other sampling methods with 49 strains isolated from 297 drinking water samples (isolation rate, 16%). taken together, these findings suggest that the h9n2 viruses mainly replicated in the intestinal tract of chickens and minor poultry species. also, the high isolation rate in drinking water samples could be a sensitive indicator for monitoring the prevalence of h9n2 viruses in the field. to better understand the evolutionary pathway of h9n2 viruses in southern china, 50 representative viruses, isolated from hong kong live-poultry markets from 2005 to 2009, were sequenced and genetically characterized. phylogenetic analysis of the h9 ha gene revealed that ck ⁄ bei-like viruses were predominant and one chicken isolate had a g1-like ha gene ( figure 1 ). this is the first time the g1like h9 ha gene has been detected in chickens from livepoultry markets in hong kong. the ck ⁄ bei-like lineage is further divided into two subgroups as previously described. 1 subgroup 1 is represented by qa ⁄ st ⁄ 243 ⁄ 00 and subgroup 2 is represented by dk ⁄ hk ⁄ y280 ⁄ 97. all h9n2 viruses in this study belonged to subgroup 2 of the ck ⁄ bei-like lineage except for the virus with the g1-like ha gene. phylogenetic analysis of the na gene also showed a similar evolutionary pattern to the ha gene with all viruses clustered within the ck ⁄ bei-like lineage. these results revealed that ck ⁄ bei-like viruses are predominant in both chickens and minor poultry. all of the pb1, pa, np, ns and m genes clustered with those of h9n2 lineage viruses previously prevailing in terrestrial poultry in southern china. phylogenetic analysis of the pb2 gene revealed three different lineages; g1-like (n = 1), ck ⁄ sh ⁄ f ⁄ 98-like (n = 15), and unknown avian (n = 34). the sh ⁄ f ⁄ 98-like lineage (or f ⁄ 98-like) was previously reported in eastern china 12 and was used previously for vaccine production in an intensive vaccination program. 13 this pb2 gene lineage was also distinguishable from the ck ⁄ bei-like lineage and its presence in the viral genome may be due to reassortment between the vaccine strain and field isolates, followed by selective establishment in terrestrial poultry. gene constellation analyses of the 50 viruses revealed six genotypes. thirty-four of the viruses analyzed belonged to two genotypes, b14 and b15, which were also the prevailing reassortants found in other provinces in southern china since 2005. 2 the remaining sixteen viruses belonged to four novel genotypes that have not been identified before in this region. characterization of h9n2 influenza viruses isolated from live poultry in hong kong markets from a 5 year surveillance program revealed that ck ⁄ bei-like viruses were predominant in southern china and were continuing to evolve. two recognized and four novel genotypes were identified in this study. one characterized virus, ck ⁄ hk ⁄ nt499 ⁄ 07, had a g1like ha gene (the first time this has been detected in hong kong poultry markets) that showed a close relationship with two human h9n2 strains isolated in 2009. g1-like viruses were usually detected and caused outbreaks in chickens of middle eastern and european countries, [14] [15] [16] and minor poultry, mainly quail, in southern china. 1 whether the g1-like virus was transmitted from china to middle eastern and european countries, as the highly pathogenic h5n1 virus did in the last five years, or vice versa, is still unknown. since the ck ⁄ hk ⁄ nt499 ⁄ 07 strain clustered with other g1-like strains isolated previously in minor poultry in southern china, the g1-like viruses in chicken may be due to interspecies transmission from minor poultry species. genetic studies demonstrated that reassortants with genotypes b14 and b15 persistently occurred in either chickens or other minor poultry species from 2005 to 2009. other genotypes that were prevalent in southern china might be being gradually replaced and four novel genotypes were identified in this study. these novel genotypes were generated through reassortment of viruses with different lineages. a newly emerged f ⁄ 98-like lineage originating from eastern china is responsible for generation of some of the novel genotypes found in this study. 12 the ck ⁄ bei-like lineage is gradually being replaced by f ⁄ 98-like lineages which are becoming dominant in northern and eastern china. 12, 17 animal experiments have also demonstrated that f ⁄ 98-like viruses are more effective in replication and transmission in chickens compared with ck ⁄ bei-like viruses. 18 since the f ⁄ 98-like lineage of the pb2 gene has been introduced into southern china, this newly emerged lineage may have a higher tendency to replace the rnp genes in the circulating ck ⁄ bei-like viruses and subsequently become the endemic virus in terrestrial poultry. in vietnam, the modelling of the pandemic h1n1 progression estimates that 460 000 (260 000-740 000) pigs might be exposed to the virus on the basis of 410 000 cases among swine owners (220 000-670 000). 1 a poor level of biosecurity, high animal densities, and a mix of species could increase the risk of influenza virus flow, persistence, and emergence on swine and poultry farms. this study was set up in the red river delta, where a third of the national pig husbandry is produced. 2 the aims are to give preliminary information of the epidemiological state of swine influenza and in order to further assess the risk of infection of swiv, through cross-species transmissions from poultry to pigs. this paper will present the preliminary results on swiv and the risk factors of pig seropositivity in vietnam. a cross-sectional study was conducted in two provinces of the red river delta in april 2009. pig farms were randomly selected from nine communes representative of at risk area of avian h5n1. in each farm, pig and poultry were sampled and collected to virological and serological analyses. interviews were conducted in all farms by trained interviewees. questionnaires included closed and open questions on ª 2011 blackwell publishing ltd, influenza and other respiratory viruses, 5 (suppl. 1), 60-78 livestock husbandry ⁄ management and household characteristics, such as herd size and structure, health history and vaccination, pig housing, watering and feeding system, reproduction, purchasing of animals, biosecurity measures, pig contact with poultry, and environmental factors. the virological detection assay was performed on pools of nasal swab specimens from pigs. we investigated whether real-time rt-pcr assay could detect gene m on pools of nasal swab specimens before attempting virus isolation from individual nasal swab specimens. the poultry and pig sera were tested against influenza type a with an enzyme-like immunosorbant assay (elisa) competition test idvetª. this commercial kit is designed to specifically detect antibodies directed against the np protein antigen of influenza type a viruses. the positive serum samples were examined in hemagglutination inhibition (hi) to determine antibody titers and subtypes. the hi test was tailored for h1, h3, and h9 subtypes in pigs and h6 and h9 subtypes in poultry. seroneutralization tests by pseudo particles were used to test the presence of antibodies directed against h5 subtype. we analysed the data for relationships between influenza a serological status (the outcome variable) and possible risk factors using r version 2ae11ae1 (r development core team). the statistical unit was the individual. initially, the quantitative variables were encoded into categorical variables according to the quartiles or median. descriptive statistics (e.g., means or medians, proportions, standard deviations) were calculated for all herd-level and commune level predictors to assist in the subsequent modeling process. we also performed the independence test among all variables to determine if variables were dependant. then, univariate analysis of potential risk factors for the pigs being positive for swiv and estimation of odds ratios were performed using generalised linear mixed models with binary outcome and logit link function for each herd-level and commune-level variable to determine which variables were individually associated with influenza a seropositivity at a significance level of p < 0ae30. herd and commune of residence were included as a random effect to account for the correlation of observations at the herd level. the third stage of the analyses included the four herdlevel variables found to be significantly (p < 0ae30) associated with influenza a seropositivity. an automatic process using all possible associations between the selected variables was computed into a mixed logistic regression models, with random effects. when two variables were collinear, as determined before, only one variable was likely to enter the multivariable model, and therefore, the selection of which collinear variable to enter the model was guided by biological plausibility and statistical significance. all of the 146 pools of nasal swabs were rt-pcr negative. the maximal possible prevalence considering perfect diagnostic tests would be of 2ae03% at a confidence level of 95%, in an infinite population within these regions (win-episcope 2ae0). six hundred-and-nine pig sera were tested in 76 nonvaccinating farms. the herd seroprevalence of swine influenza in the commune previously infected by the avian h5n1 in the red river delta raised by 17ae1% [8ae7; 25ae6] in april 2009. but among 13 seropositive farms, only four had at least two seropositive pigs. the within-herd seroprevalence is very low, and no seropositivity was detected in the majority of farms. estimates had large confidence intervals due to small sample sizes. the individual seroprevalence raised 3ae62% [1ae98; 5ae27]. the subtyping of seropositive sera is still in process. descriptive statistical analyses on five major risk factors of swiv: farm size, breeding vs. fattening, purchasing, percentage of family income, and poultry production, were conducted. based on this analysis, three types of farming systems were identified and included in mixed models ( table 1) . percentage of family income by pig production and poultry production were not differentiating factors for this typology. whereas types 1 and 2 seem to be specialized in fattening, the type 3 produces and might sell piglets on the farm site. the exploration of the different variance components indicated that the random effect variances were mainly associated with the herd, while the commune did not seem to have any effect. therefore we included in all models only the herd as a random effect. the random effect term for herd was modelled, assuming a normal distribution with a table 1 . typology of farming system type 1: large fattening farms largest scale production, with more than 40 pigs per year specialized in fattening, and purchase more than 20 pigs per year type 2: small fattening farms small scale of production, with less than 20 pigs per year specialized in fattening, and purchase less than 20 pigs per year type 3: medium breeding-fattening farms medium scale of production, with less than 40 pigs per year breeding and fattening piglets, with rare purchase common variance [$n(0,r2herd)]. 3 the univariate analyses were conducted on 22 variables and typology variables, with herd as random effect. some coefficient or confidence intervals were inconsistent because of small effectives, especially for the percentage of self-product culture or the pig freegrazing because of the lack of positive results in the dataset. the only one significant (p value < 0ae1) parameter was the percentage of pig sales in the familial annual income. surprisingly, common risk factors of swine influenza infection, such as farm size, animal movements, and sanitary parameters got low odds ratio individually (without being significant); the typology provides the hypothesis of complex interactions effects that increase the risk of infection. as shown in table 2 , the farming system type 3 got a higher seroprevalence of 6ae47% [3ae00-11ae94] and a higher risk indicator, with or = 5ae26 (p-value = 0ae18) in comparison with type 1. this finding was not significant. in the multivariate mixed model, the percentage of familial income provided by pig production was the only one significant variable, with or = 0ae22 [0ae04-1ae25]. the focus on diseased animals in the winter-time is usually required in order to increase the likelihood to isolate the virus, although the isolation rate on healthy or clinical samples never exceed 6%. 4 the season and the lack of disease reports might explain the difficulties to detect influenza viruses. additionally, the pooling method tends to decrease the isolation rate because of a dilution effect, potential presence of pcr assay inhibitors, or uneven distribution of virus in the sample. 5 our seroprevalence results must be confirmed and the subtypes identified, especially because we found only one positive animal in a few farms that could be attributed to false positive results of the elisa test (performances are not known). these preliminary results are in favor of a virus circulation at low level in the spring, but must be completed by further surveys in the winter and before the new year (têt celebration) when pig production, trade, and movement increase at their maximum. no clear prior information on the expected prevalence of swine influenza in vietnam, tests sensitivity, and speci-ficity could be obtained from literature or reliable sources. bayesian methods will be carried out in the future in order to compute prevalence and ⁄ or to estimate the probabilities of freedom. the risk factors analysis was limited by the lack of positive results. further studies are necessary to identify the at-risk season and type of farming systems at risk of swine influenza infection. however, this investigation of risk factors leads to the hypothesis that medium size breeding-fattening farms had a higher risk than large or small size fattening farms. further investigation are needed to precise this typology. the risk of swiv infection increases with a combination of three major factors. poultry production does not seem to play any role on swine infection. the generalized linear mixed model afforded to take into account all the non investigated parameters at the herd level. although we investigated the most common risk factors of swine influenza infection covering different kind of fields, the herd random effect might explain risk variations. mixed models have become a frequently used tool in epidemiology. due to software limitations, random effects are often assumed to be normally distributed. since random effects are not observed, the accuracy of this assumption is difficult to check. 6 further studies, such as case-control or cohort studies could help to identify more precisely risk factors of swine influenza seropositivity, as these study designs are more adapted than cross-sectional studies. the concept that swine are a mixing-vessel for the reassortment of influenza viruses and for the emergence of pandemic influenza viruses has been re-enforced by the emergence of the recent pandemic. 1 the pandemic h1n1 virus of 2009 (h1n1pdm) is believed to have emerged through the reassortment of north american triple reassortant and eurasian avian-like swine influenza viruses. 2 since the immediate precursor of this pandemic virus has not yet been identified, it is not possible to be definite whether the reassortment leading to the pandemic occurred in swine, but swine influenza viruses are the nearest known ancestors of each gene segment of h1n1pdm. 1, 2 the mechanisms of pandemic emergence are not clear. it is believed that the pandemics of 1957 and 1968 arose through reassortment of the pre-existing human seasonal influenza virus with avian influenza viruses, and swine have been proposed to be a possible intermediate host where such reassortment between human and avian viruses may take place. 4 the 2009 pandemic was the first to arise for over 40 years and the first to occur after the understanding that pandemics arise from animal influenza viruses. 3 systematic studies of influenza virus ecology and evolution in swine are, therefore, important in order to understand the dynamics of pandemic emergence. furthermore, since swine are the likely host within which h1n1pdm virus originated, it was predicted that this virus would readily infect swine and may reassort with endemic swine influenza viruses. these predictions have now been confirmed with reports of h1n1pdm being detected in pigs in many countries and reassortment with endemic swine influenza virus being confirmed. 5 while h1n1pdm has been genetically and antigenically stable in humans, reassortment between h1n1pdm, which is well adapted to transmission in humans, and other avian or swine viruses may lead to the origin of novel viruses posing a threat to public health. in addition to endemic swine virus lineages, avian influenza viruses such as h9n2 and highly pathogenic avian influenza (hpai) h5n1 have also been occasionally identified in pigs in parts of asia. 6, 7 it has been shown that h1n1pdm readily reassorts with h5n1 to generate viable progeny in vitro. 8 it is therefore essential to monitor the ecology, evolution, and biological characteristics of swine influenza viruses so that their continued evolution and zoonotic and pandemic potential can be monitored. there is however, a paucity of surveillance data on swine influenza viruses worldwide. this is in part related to the negative commercial consequences that may arise from detection of influenza in a swine herd leading to a major economic loss to the producer. here we outline a surveillance system that has been in place in hong kong for the last decade, based on sampling animals arriving at an abattoir in hong kong. we demonstrate the feasibility of such surveillance in an abattoir setting and compare methods used for detection influenza viruses in swine. virus isolation was carried out by inoculation into mdck cells and by allantoic inoculation in embryonated eggs as previously described. 6 virus isolates were subtyped by haemagglutination inhibition tests using specific antisera and genetically characterized by sequencing and phylogenetic analysis of the haemagglutin gene. 2, 6 virus detection by rt-pcr a subset of 1154 recent specimens was tested in parallel by real time pcr using the biorobot universal system (qiagen) that enables fully-automated viral nucleic acid extraction and downstream reaction setup in a 96-well plate format. total viral nucleic acids were extracted in a 96-well plate format with the qiaamp virus biorobot mdx kit (qiagen) on the biorobot universal system (qiagen) according to the manufacturer's instructions. briefly, 140 ll of sample was lysed in 280 ll buffer al, supplemented with 5ae6 lg carrier rna in a s block (qiagen), which placed the samples into a 96 well plate format. after protease digestion, samples were transferred to silica based membrane in 96 well plate format for binding. following two washing steps, rna was eluted in 80 ll of elution buffer (buffer ave) into a 96 well elution microplate cl (qiagen) . for the synthesis of cdna, 10 ll of purified rna was used in a 20 ll reaction containing 4 ll of 5· buffer, 0ae75 nm of each deoxynucleotide triphosphate (dntp), 10 mm dithiothreitol, 3 lg random primer, 40 u of rnaseout recombinant ribonuclease inhibitor, and 200 u of superscript iii reverse transcriptase (all from invitrogen). reactions were performed in the geneamp 9700 thermocycler (applied biosystems) with the following parameters: 60 minutes at 50°c, 15 minutes at 72°c, and soak at 4°c. subsequent to the reactions, 20 ll of cdna was diluted 1 ⁄ 10 by adding 180 ll of ae buffer (qiagen) . real-time pcr was performed using the power sybrò green pcr master mix (applied biosystems) according to the manufacturer's instructions. briefly, 5 ll of 1⁄ 10 diluted cdna was amplified in a 25 ll reaction containing 12ae5 ll of 2· power sybr green pcr master mix, 800 nm of forward primer m52c (5¢-ctt cta acc gag gtc gaa acg-3¢) and 800 nm of reverse primer m253r (5¢-agg gca ttt tgg aca aag ⁄ t cgt cta-3¢). the primers have been designed to amplify the sequences in the conserved region of influenza a virus matrix gene, thereby detecting viruses from different species including swine influenza viruses. 8 real-time pcr was performed in the abi 7500 fast system (applied biosystems) with the following cycling conditions: 10 minutes at 95°c once, 30 seconds at 95°c, and 1 minutes at 50°c for 40 cycles, followed by melting curve analysis with 15 seconds at 95°c, 1 minutes at 50°c, and 15 seconds at 95°c. in each assay, serially diluted plasmids containing the full length m gene cloned from a ⁄ vietnam ⁄ 1204 ⁄ 2004 (h5n1) were included as standards to perform absolute quantification. a manual baseline was set from cycles 3-15 and a manual cycle threshold (ct) was set at 0ae2. samples that were positive or unequivocal results from the real-time pcr were confirmed by performing gel electrophoresis on the pcr products. positive visual identification was made in the presence of the target pcr product at 244 bp in length. a total of 16 566 tracheal and 11 100 nasal swabs were processed during the years january 2002-april 2010 and yielded 393 influenza virus isolates, an overall virus isolation rate of 1ae4%. of these, 352 were subtype h1 (classical swine, eurasian avian-like swine, and triple-reassortant), 25 were human-like h3 viruses, and 16 were eurasian avianlike swine h3n2 viruses. culture in mdck cells yielded 95% of h1 subtype viruses, 100% of the human seasonal-like h3n2 viruses, and 68ae8% of the avian-like eurasian swine h3n2 viruses. culture in embryonated eggs yielded 17ae9% of the h1 subtype viruses, 0% of the human seasonal-like h3n2 viruses, and 62ae5% of eurasian avian-like swine h3n2 viruses ( figure 1 ). tracheal and nasal swabs each gave comparable overall virus isolation rates (1ae4%). however, isolation rates for human-like h3n2 viruses were 4ae8 fold higher in nasal swabs (0ae17% versus 0ae036% respectively; p = 0ae002) ( figure 1) . a parallel evaluation of rt-pcr and culture was carried out in 1154 specimens. rt-pcr detected 10 ⁄ 12 (83%) of the culture positive specimens. rt-pcr was also positive in 8 ⁄ 1154 (0ae7%) culture negative specimens, but all these specimens had very low virus load in the rt pcr tests. virus could not be cultured from these culture negative specimens even by attempts at virus re-isolation from the frozen specimen. surveillance in an abattoir setting provides an acceptable yield of influenza viruses and is a feasible method of swine influenza surveillance. sampling in a large abattoir setting allows surveillance to be carried out anonymously with no negative consequences to the supplier. the supply-chain of pigs to the hong kong abattoir involves pigs being trucked in over long distances and may provide opportunity for virus amplification during transport. thus, virus isolation rates may be lower in more vertically integrated and homogenous production and slaughter systems where less mixing of pigs occurs. our results indicate that mdck cell culture is essential for optimizing virus isolation during swine influenza surveillance. allantoic inoculation of embryonated eggs by itself is sub-optimal for isolation of swine influenza viruses. it is however possible that inoculation of embryonated eggs by the amniotic route may lead to better isolation rates than allantoic inoculation. rt-pcr detection is an alternative method for virus detection. but the additional specimens detected by rt-pcr did not yield culturable virus, even following attempts at re-isolation and sequential passage. the rt-pcr positive ⁄ virus isolation negative specimens had very low virus load, and this may be the explanation for the inability to isolate such viruses. in addition, rt-pcr did not detect all viruses isolated by culture. tracheal and nasal swabs gave comparable isolation rates with the exception of human-like h3n2 viruses which were more frequently isolated from nasal swabs. this may suggest that, in contrast to endemic swine influenza virus lineages, these human-like h3n2 viruses are less adapted to replication in the lower respiratory tract. in summary, collection of nasal or tracheal swabs in an abattoir setting together with virus isolation in mdck cells provides a feasible approach to surveillance of swine influenza viruses. kong, 2003 kong, -2010 introduction wild waterfowl are the natural reservoir of influenza a viruses (aiv), and they play an important role in the genesis of pandemic influenza. it is suggested that the 1918 pandemic virus was purely derived from avian virus, which adapted to humans and caused efficient human-to-human transmission, 1 while the pandemics of 1957 and 1968 had acquired the viral haemagglutinin, pb1 polymerase, and in 1957, the neuraminidase gene segments from the avian gene pool. 2 the major regional outbreaks of highly pathogenic avian influenza (hpai) h5n1 in asia, europe, and africa highlight the potential role played by migratory waterfowl in disseminating highly pathogenic influenza viruses. 3 therefore defining the influenza virus gene-pool in wild birds is of vital importance. surveillance was carried out 2-3 times weekly from 2003 to 2010 during the winter months of october to april in the hong kong mai po nature reserve and lok ma chau, hong kong. the hong kong mai po nature reserve and lok ma chau are along the east asia-australian flyway where a peak of more than 30 000 ducks and grebes congregate every winter. 4 fecal droppings were collected and transported in vials containing 1ae0 ml of vtm, which was prepared from m199 (9ae5 g ⁄ l), penicillin g (2 · 10 6 u ⁄ l), polymyxin b (10 · 10 6 u ⁄ l), gentamicin (250 mg ⁄ l), nystatin (0ae5 · 10 6 u ⁄ l), ofloxacin hcl (100 mg ⁄ l), and sulfamethoxazole (1 g ⁄ l). an aliquot of 100 ll from each swab sample was inoculated into the allantoic cavity of a 9-to 11-day-old chicken embryonated egg, and incubated for 3 days at 37°c. positive ha isolates were subtyped using standard antisera 5, 6 and rt-pcr was performed with the used of one-step rt-pcr assay (invitrogen) described earlier, 7 followed by sequencing on abi prism 3730xl dna analyzer. the determination of species of origin was performed by dna barcoding of the mitochondrial cyto-chrome oxidase i gene from dna extracted from the fecal droppings. 8 during the 7-year surveillance period, a total of 193 influenza viruses were isolated from 39 354 samples collected, an overall isolation rate of 0ae49%. a total of 192 isolates were obtained from 39 152 specimens collected during the winter period coinciding with the southern migration of waterfowl along the east asian flyway and one isolate obtained from 202 samples collected in spring during the period when northern migration of waterfowl took place along the east asian-australasian flyway. the isolation in hong kong was slightly lower than a similar study conducted in south korea in which the isolation rate of migratory birds was 0ae8% in 2003-2008. 9 this suggested a slightly lower prevalence of influenza virus present in hong kong as the birds migrated southwards. the viruses isolated in hong kong, representing hemagglutinin (ha) subtypes of h1-h12 and neuramidinase (na) subtypes of n1-n9, were all from wild waterfowl ( table 1) . out of the twelve ha subtypes isolated, h10 and h6 were the two subtypes that were isolated frequently every year for h10 and in six out of seven years for h6, respectively. h10 and h6 viruses accounted for 16ae7% and 22ae4% of all virus isolated, respectively. on the other hand, h3, h9, and h12 were the least prevalence (0ae5%) and were only isolated once in 7 years. of the na subtypes, n1 and n2 were isolated most often (24ae4% and 19ae2% of all isolates, respectively) and n5 was the least (0ae5%). november was the month that had the highest prevalence of influenza virus (0ae83% of samples being positive) compared to only 0ae3% in march. the subtype's variation was the most diverse in december during our 7 years of surveillance. this suggested that more of these wild migratory birds may be carrying influenza virus when they arrive in hong kong. however the continued isolation of viruses suggests continued circulation of these viruses in the vicinity of mai po. the study of dna barcoding for the mitochondrial cytochrome oxidase i gene retrieved from fecal droppings revealed that the isolates originated mainly (89ae2%) from birds of the order anseriform, family anatidae including eurasian wigeon, northern shoveler, northern pintail, common teal, and garganey. non-anseriformes which were found to have shed aiv viruses were cormorant, grey heron, and stint. none of the water samples collected from the ponds where these birds congregate were found to be positive for the virus. phylogenetic analyses of the ha gene of the lpai h5 viruses isolated in this study clustered with that of the other lpai h5 viruses isolated from hokkaido, mongolia, and siberia and were not closely related to the hpai h5n1. satellite tracking of 47 eurasian wigeons and northern pintails in dec 2008 and 2009 revealed their flyway from hong kong to as far north as eastern russia, eastern mongolia, and northern china. no hpai h5n1 viruses were isolated in this study from apparently healthy birds. however, as part of the surveillance of dead wild birds carried out by the department of agricultural, fisheries and conservation of the government of hong kong during this same period, over 50 dead wild birds were tested positive for hpai h5n1 and has been reported elsewhere. 10 our influenza surveillance in hong kong has revealed a diversity of influenza virus subtypes the migratory waterfowl infected within the region. the result of the phylogenetic analysis correlated with the findings from satellite tracking that viruses isolated in hong kong were closely related to those isolated in areas along the migratory route. no healthy bird was isolated with hpai h5n,1 although dead wild birds have been regularly found to have hpai h5n1 virus, suggesting that infected birds might not live for a long period. introduction a novel swine-origin h1n1 influenza virus emerged in mexico in april 2009 and rapidly spread worldwide, causing the first influenza pandemic of the 21st century. 1 most confirmed human cases of h1n1 ⁄ 2009 influenza have been uncomplicated and mild, but the increasing number of cases and affected persons worldwide warrant optimal prevention and treatment measures. today, almost all of the pandemic h1n1 ⁄ 2009 viruses tested are resistant to m2blockers. 2 therefore, only the neuraminidase (na) inhibitors are currently recommended for treatment of this pandemic influenza. for the control of influenza infection, the clinical use of oseltamivir has increased substantially during the 2009 pandemic. to date, the majority of tested clinical isolates have remained susceptible to na inhibitors, oseltamivir and zanamivir, 2 but oseltamivir-resistant variants with h275y na mutation (n1 numbering) have been isolated from individuals taking prophylaxis, from immunocompromised patients, and from a few community clusters. 3, 4 in view of the high prevalence of oseltamivirresistant seasonal h1n1 influenza viruses in 2007-2008, the isolation of resistant h1n1 ⁄ 2009 viruses without known oseltamivir exposure raised great concern about the transmissibility and fitness of these resistant viruses. here we studied the transmissibility of a closely matched pair of pandemic h1n1 ⁄ 2009 clinical isolates, one oseltamivir-sensitive and one resistant, in both direct contact and respiratory droplets routes among ferrets. viral fitness was evaluated by co-infecting a ferret with both the oseltamivir-sensitive and -resistant viruses. the viruses were also characterized by full genome sequencing, susceptibility to na inhibitors, and growth in mdck and mdck-siat1 cells. oseltamivir-resistant influenza a ⁄ denmark ⁄ 528 ⁄ 09 (h1n1) virus (a ⁄ dm ⁄ 528 ⁄ 09) was isolated from the throat swab of a patient who had influenza-like symptoms and received post-exposure oseltamivir prophylaxis (75 mg once daily). 5 wild-type influenza a ⁄ denmark ⁄ 524 ⁄ 09 (h1n1) virus (a ⁄ dm ⁄ 524 ⁄ 09) was isolated from a patient in the same cluster of infection as the a ⁄ dm ⁄ 528 ⁄ 09 virus. to assess growth kinetics of viruses, confluent mdck or mdck siat1 cell monolayers were infected with viruses at a multiplicity of infection (moi) of approximately 2ae0 pfu ⁄ cell (single-step) or 0ae001 pfu ⁄ cell (multi-step). supernatants were collected every 2 h or 12 h p.i. for 6 time points. a modified fluorometric assay using the fluorogenic substrate 2¢-(4-methylumbelliferyl)a-d-n-acetylneuraminic acid (munana) was used to determine viral na activity. 6 the drug concentration required to inhibit 50% of the na enzymatic activity (ic 50 ) was determined by plotting the percent inhibition of na activity as a function of compound concentration calculated in the graphpad prism 4 (la jolla, ca) software from the inhibitor-response curve. na enzyme kinetics were determined by measuring na activity every 60 seconds for 60 minutes under the same conditions as above, when all viruses were standardized to an equivalent dose of 10 6ae0 pfu ⁄ ml. the k m and v max were calculated by fitting the data to the appropriate michaelis-menten equations using nonlinear regression in the graphpad prism 4 software. young adult ferrets (4-5 months of age) were obtained from the ferret breeding program at st. jude children's research hospital. all ferrets were seronegative for influenza a h1n1 and h3n2 viruses and for influenza b viruses. for transmission studies, the donor ferrets were lightly anesthetized with isoflurane and inoculated intranasally with 10 6 tcid 50 virus in 1ae0 ml sterile pbs . after the donor ferrets were confirmed to shed virus on day 2 p.i., each donor was then housed in the same cage with two naïve direct-contact ferrets. two additional recipient ferrets were placed in an adjacent cage isolated from the donor's cage by a two layers of wire mesh (approximately 5 cm apart) that prevented physical contact but allowed the passage of respiratory droplets. ferret weight and temperature were recorded daily for 21 days. nasal washes were collected from donors and recipients on day 1, 2, 4, 6, 8, 10, 12, and 14 p.i. by flushing both nostrils with 1ae0 ml pbs, and tcid 50 titers were determined in mdck cells. serum samples were collected 3 weeks after virus inoculation, and were tested for seroconvention by hi assay. full genome sequencing revealed that the pair of h1n1 ⁄ 2009 viruses differed only at na amino acid position 275, where the pandemic a ⁄ dm ⁄ 528 ⁄ 09 virus had an h275y amino acid mutation caused by a single t-to-c nucleotide substitution at codon 275. the wild-type a ⁄ dm ⁄ 524 ⁄ 09 was susceptible to oseltamivir carboxylate (mean ic 50 : 5ae0 nm), but the a ⁄ dm ⁄ 528 ⁄ 09 carrying the h275y na mutation had ic 50 values approximately 200-300 times of the wild-type viruses (mean ic 50 : 972 nm). the ic 50 of zanamivir was comparable for both viruses and were uniformly low (mean ic 50 £ 1ae3 nm). the h275y na mutation confers resistance to oseltamivir carboxylate but did not alter susceptibility to zanamivir. to understand the impact of the h275y mutation on the na enzymatic properties, na enzyme kinetics was determined. the na of the oseltamivir-resistant virus had a slightly higher k m (mean = 55 lm) and lower v max (mean = 101 u ⁄ sec) than na of the sensitive virus (k m , mean = 80 lm; vmax, mean = 86 u ⁄ sec). the results suggested that the h275y na mutation reduced na affinity for substrate and na catalytic activity, although the function of na was not severely impaired. to further evaluate the impact of the h275y na mutation on virus growth in vitro, single-and multi-cycle growth studies of both viruses were performed in mdck and mdck-siat1 cells. in the both single-and multiple-cycle growth curves, the two viruses reached comparable levels eventually, but the initial growth of the resistant virus was significantly delayed by at least 1-2 logs in comparison to that of wild-type virus (p < 0ae05). the donor ferrets inoculated with wild-type a ⁄ dm ⁄ 524 ⁄ 09 or oseltamivir-resistant virus shed virus productively until day 6 or day 8 p.i., with a peak virus titer comparable to that of a ⁄ dm ⁄ 524 ⁄ 09 virus (table 1 ). in a ⁄ dm ⁄ 524 ⁄ 09 virus group, two of 2 direct-contact ferrets the weight loss in ferrets is the maximum percentage loss compared with the initial weight. virus shedding is indicated as number of virus-shedding animals ⁄ total number; mean peak virus titer (log 10 tcid 50 ⁄ ml) in nasal wash samples is indicated in parentheses. serum hemagglutination inhibition (hi) titer to homologous virus in ferret serum was determined on day 21 p.i. duan et al. and 1 of 2 respiratory droplet-contact ferrets were infected through virus transmission, as indicated by the virus titers and inflammatory cell counts in their nasal washes and also by sero-conversion. under identical conditions, in a ⁄ dm ⁄ 528 ⁄ 09 group, only 2 of 2 direct-contact ferrets were infected through virus transmission, but neither respiratory droplet-contact ferrets was infected, as confirmed by the absence of sero-conversion (table 1) . virus shedding in the direct-contact ferrets was lower and peaked after a longer interval in this group than in the oseltamivir-sensitive a ⁄ dm ⁄ 524 ⁄ 09 group (table 1) , but the resistant viruses appeared to cause a similar disease course in ferrets without apparent attenuation of clinical signs. these results showed that an oseltamivir-resistant h275y mutant of pandemic h1n1 virus, a ⁄ dm ⁄ 528 ⁄ 09 virus could be only transmitted efficiently by direct contact. to compare the relative fitness, growth capability, and transmissibility of the sensitive and resistant h1n1 ⁄ 2009 viruses within host, a donor ferret was co-inoculated with a 1:1 ratio of the sensitive and resistant viruses, and another two naive ferrets were housed with the donor to test direct contact. during co-infection, the pattern of virus shedding and the clinical signs were similar to those in ferrets inoculated with either a ⁄ dm ⁄ 524 ⁄ 09 or a ⁄ dm ⁄ 528 ⁄ 09 virus (table 1 ). in the inoculated donor ferret, the virus population in the nasal washes remained mixed but wild-type viruses outgrew the resistant virus progressively ( figure 1 ). two of 2 direct-contact ferrets were infected through virus transmission, but only wild-type virus was detected in both direct-contact ferrets ( figure 1 ). in summary, oseltamivir-sensitive a ⁄ dm ⁄ 524 ⁄ 09 virus possessed better growth capability in the upper respiratory tract than did resistant a ⁄ dm ⁄ 528 ⁄ 09 virus, and thus had an advantage in directcontact transmission. our study determined the comparative transmissibility of two naturally circulating oseltamivir-sensitive and -resistant pandemic h1n1 ⁄ 2009 viruses; we demonstrated inefficient respiratory-droplet transmission of an oseltamivir-resistant h275y mutant of pandemic h1n1 virus among ferrets, although it retained efficient direct-contact transmission. we suggest that the lower fitness of resistant virus within the host along with its reduced na function and delayed growth in vitro may in part explain its less efficient transmission. notably, the h275y mutant of h1n1 ⁄ 2009 used in this study was the first oseltamivir-resistant h1n1 ⁄ 2009 isolate from a patient on oseltamivir prophylaxis to be characterized for transmissibility. our observation in the animal model is consistent with the epidemiological data collected from humans, which showed no evidence of predominant or continued circulation of oseltamivir-resistant viruses. 4 as this study was undertaken, additional h275y mutants of h1n1 ⁄ 2009 viruses have emerged in the absence of oseltamivir use. 7, 8 the emergence of these viruses should raise concerns as to whether resistant h1n1 ⁄ 2009 viruses will acquire greater fitness and spread worldwide as the naturally resistant h1n1 viruses did during the 2007-2008 season. two independent studies have evaluated the pathogenecity and transmission of other oseltamivir-resistant pandemic h1n1 ⁄ 2009 clinical isolates in the animal models. 9, 10 one of the studies, 9 which also used an oseltamivir-resistant virus isolated from a patient under oseltamivir prophylaxis, observed similar results as ours: although the respiratory-droplet route of transmission was not investigated, it was shown that the resistant isolate was transmitted though direct-contact route and was as virulent as wild-type virus in ferrets. 9 in another study, 10 two oseltamivir-resistant isolates were transmitted through the respiratory-droplet route in ferrets, and the dynamics of transmission were different between the two isolates. apparently, these two oseltamivir-resistant isolates were still unequal in their transmissibility and were disparate from the resistant isolate in our study. the isolation history of the two resistant isolates was unclear in this study, and this would be an important factor to understand the fitness of drug-resistant viruses. further studies with more clinical isolates of diverse isolation background are warranted to identify how these novel h275y mutants of pandemic h1n1 ⁄ 2009 virus have changed to retain their full transmissibility. taken together, all these related studies underline the necessity of continuous monitoring of drug resistance and characterization of potential evolving viral proteins. this study was supported by contract hhsn266200700005c from the national institute of allergy and infectious diseases, national institutes of pigs have been considered as hypothetical ''mixing vessels'' facilitating the genesis of pandemic influenza viruses. 1, 2 the pandemic h1n1 ⁄ 2009 virus (ph1n1 ⁄ 09) contained a very unique genetic combination and was thought to be of swine origin, as each of its eight gene segments had been found to be circulating in pig populations for more than a decade. 3 however, such a gene constellation had not been found previously in pig herds all around the world. only after its initial emergence in humans has this virus been repeatedly detected in pigs, and found to further reassort with other swine influenza virus. [3] [4] [5] a primary question remaining to be answered is whether the ph1n1 ⁄ 09-like and their genetically related viruses could become established in pig populations, thereby posing novel threats to public health. despite the fact that ph1n1 ⁄ 09 first appeared in mexico and the united states, and six of its eight gene segments were derived from the established north american triple reassortant swine influenza virus (trig), its neuraminidase (na) and matrix protein (m) genes belonged to the eurasian avian-like swine lineage (ea), which had never been detected in north america previously. 3, 6 likewise, the trig-like viruses were never reported in europe. 6 in contrast, both lineages of virus were frequently detected in asia, and reassortants between them have also been documented in recent years. 3, 5 this has given rise to a complicated ecological situation, i.e. the simultaneous prevalence of multiple genotypes of h1n1 and h1n2 viruses in pigs. 3, 5 among them, two representative reassortants showed the most similar genotypic characterization to the ph1n1 ⁄ 09 virus, the sw ⁄ hk ⁄ 915 ⁄ 2004 (h1n2) and sw ⁄ hk ⁄ 201 ⁄ 2010 (h1n1), which respectively harbor seven and six gene segments closely related to the pandemic strains. 3 , 5 to understand their in vivo characteristics and zoonotic potential, these two viruses, together with a human prototype strain and a swine ph1n1 ⁄ 09-like isolate, were chosen for a study of their pathogenicity and transmissibility in domestic pigs, ferrets, and mice. the prototype ph1n1 ⁄ 09 virus, a ⁄ california ⁄ 04 ⁄ 2009 (ca04), was provided by the world health organization collaborating centers for reference and research on influenza (atlanta, ga, usa). three ph1n1 ⁄ 09-related swine influenza viruses were isolated through our surveillance program in south china as previously described. 3, 5 the a ⁄ swine ⁄ guangdong ⁄ 106 ⁄ 2009 (h1n1, gd106) virus was a ph1n1 ⁄ 09-like swine isolate. a ⁄ swine ⁄ hong kong ⁄ 915 ⁄ 2004 (h1n2, hk915), the closest pandemic ancestor known to date, possesses an m gene derived from the ea lineage, with the other gene segments from trig viruses. 3 a ⁄ swine ⁄ hong kong ⁄ 201 ⁄ 2010 (h1n1, hk201), a recent pandemic reassortant progeny, had a ph1n1 ⁄ 09like na gene (also belonging to the ea lineage), an ea-like hemagglutinin (ha) gene, and six trig-like internal genes. 5 all viruses were propagated in madin-darby canine kidney (mdck) cells for three passages, and their titers were determined by plaque assays. all experiments with live viruses were conducted in biosafety level 3 (bsl-3) containment laboratories. pigs (4-6 week old, n = 5-6) and ferrets (5 month old, male, n = 3) were intranasally infected with 10 6 pfu of each virus, and mice (8) (9) week old, female balb ⁄ c, n = 10) with a dose of 10 4 pfu. naïve uninfected pigs (n = 2) were co-housed in the same cage with the inoculated ones from each group. body weights and clinical signs were recorded daily. virus replication was determined by titration of the virus in nasal and rectal swabs (pigs), nasal washes (ferrets), as well as from lungs and other organs (pigs and mice). seroconversion was tested by hemagglutination inhibition (hi) assays. histopathological and immunohistochemical analysis were performed as previously described. 7 statistical analysis was performed by mean analysis with pasw statistics 18 (spss inc., chicago, il, usa). the probability of a significant difference was computed using anova (analysis of variance). results were considered significant at p < 0ae05. the pathogenicity of the four viruses tested differed significantly in inoculated mice. animals infected with 10 4 pfu of hk915 experienced the most severe body weight loss (25ae1 ± 4ae7%) but started to recover after 8 days post-infection (dpi). hk201 caused similar peak body weight loss (16ae9 ± 4ae6% on 8 dpi) in mice as did ca04 (17ae3 ± 2ae4%, on 7 dpi), but the onset of clinical signs and weight loss (on 4 dpi) was 1 day later than those caused by the other three viruses. the gd106-infected group suffered the least body weight loss (6ae9 ± 1ae9%, 5 dpi) and was the earliest to recover. although all four viruses were detected in the lungs with comparable virus titers on 3 dpi (p > 0ae05), mice inoculated with gd106 consistently showed the lowest lung index (lung weight ⁄ body weight, %) on 3, 6, and 14 dpi (p < 0ae01), suggesting the slightest injury and consolidation of the lungs. in concordance with the body weight change, the lung index from the hk915 group was higher than that from any other groups on 6 and 14 dpi, indicating the marked virulence of hk915 in mice. notably, virus titer of hk201 in the nasal turbinate was lower than the other groups both on 3 and 6 dpi (p < 0ae01), but virus replication in the lower respiratory tract was either higher (in the trachea) or similar (in the lungs). observations of the body weight changes caused by infection of ph1n1 ⁄ 09 or its genetically related swine viruses in ferrets have come to a similar conclusion as that for the mouse experiment. after nasal inoculation with 10 6 pfu of each virus, all groups of ferrets experienced transient body weight loss for 2-3 days, except for those infected with gd106, which showed no significant weight loss (p > 0ae05). although ferrets from the ca04-infected group reached their peak weight loss (6ae2 ± 0ae8%, 2 dpi) one day earlier than those from the hk201 and hk915 groups, they began to regain body weight quickly thereafter. hk201-infected ferrets also recovered rapidly and their body weights reached the same level as those of the gd106-infected group at 6 dpi. comparatively, ferrets inoculated with hk915 had the most retarded body weight recovery, which did not get back to the baseline level until 11 dpi. hk201 was only detectable in the nasal wash on 2 dpi, whereas the duration of virus shedding for gd106, hk915, and ca04 was 4-6 days. by combining the data obtained from the virus titration in the mouse turbinate and ferret nasal washes, a possible conclusion can be made that hk201 may have lower transmissibility than the other three viruses. after inoculation or exposure by direct contact (physical contact) with the ph1n1 ⁄ 09 virus and its close relatives, most pigs experience no or mild symptoms, such as slight loss of appetite and inactivity. body weight loss was only recorded in pigs inoculated with hk915 during the second week post-inoculation, but not in their contact pigs or in the other groups. diarrhea was observed intermittently in each of the inoculated or contact groups throughout the experiment, and viruses could be recovered in the rectal swabs, saliva, drinking water, and environmental swabs (inner cage walls accessible to the pigs) at various time points. however, virus titers in the positive rectal swabs were just slightly above the detection limit, while those from the environment sometimes could be higher. whether these viruses can replicate in the digestive tract or were just carried-over by contaminated foods and water requires further investigation. although virus could be detected in the nasal swabs of all infected or contact animals, the lowest peak titer was from pigs inoculated or in contact with hk201 (0ae5-1ae5 log tcid50 ⁄ ml lower than the other groups), suggesting unfavorable replication in the nasal cavity for this virus. postmortem examination on 4 and 7 dpi revealed that pigs infected with hk915 had the most extensive gross lesions in the lungs, and histochemical staining of viral nucleoprotein (np) in lung tissues on 4 dpi also suggested the best replication for hk915 in the lower respiratory tract. on 11 days post-contact (dpc), all pigs exposed to the inoculated animals developed sero-conversions (hi = 80-160) except for one from the gd106 contact group. however, on 17 dpc, its hi titer reached 40, indicating slower seroconversion. this study revealed that both the 2009 pandemic h1n1 and its genetically related swine viruses could readily infect mice, ferrets, and pigs causing mild to moderate clinical symptoms. they could also transmit efficiently between pigs. when compared with the pandemic stains and its reassortant progeny (hk201), the hk915 (h1n2) virus containing the ea-like m gene in the genetic context of the trig virus showed consistently higher virulence in all three mammalian models tested, but it is still unknown what might happen if such a virus further reassorts to obtain the pandemic-like or ea-like na gene. however, our findings suggest that pigs could likely maintain the prevalence of different genotypes of pandemic-related influenza viruses, and highlight the zoonotic potential of multiple strains of swine influenza virus. pandemic influenza viruses emerge from the animal reservoirs. 1 among the three pandemics that occurred in the last century, we learned that the 1957 h2n2 and the 1968 h3n2 pandemic viruses emerged by reassortment between circulating human virus and avian-origin influenza virus(es). 1 studies on the emergence of the catastrophic 1918 spanish h1n1 virus suggest that the virus may have obtained all of its eight gene segments from the avian reservoir, 2,3 or alternatively is a reassortant between mammalian and a previously circulating human influenza virus. 4 over 40 years since the last pandemic, the first pandemic in the 21st century arose in 2009 and was caused by a swine-origin influenza virus containing a unique gene combination, with gene segments derived from the circulating north america ''triple reassortant'' (pb2, pb1, pa, ha, np, and ns) and the ''eurasian'' (na and m) swine influenza viruses. 5, 6 analysis of the pandemic h1n1 viruses failed to identify known molecular markers predictive of adaptation to humans. 6 the ''triple reassortant'' swine influenza viruses emerged in late 1990s in north america is a reassortant between classical swine (descendent of the 1918 virus after adaptation in swine population), avian, and human influenza viruses. 7 the eurasian influenza virus was originally an avian influenza virus that was introduced into the european swine population in the late 1970s. 8, 9 while incidents of zoonotic infection with triple reassortant or eurasian influenza in humans have been reported, 10, 11 sustained human-to-human transmission has never been established. these results suggest that the unique gene combination seen with the pandemic h1n1 viruses may confer its transmissibility among humans. we have carried out systematic prospective surveillance of swine influenza in southern china over that last 12 years through samples routinely collected at an abattoir in hong kong. during this time, the surveillance results suggest co-circulation of classical swine h1n1, triple reassortant h1n2, eurasian swine h1n1, and a range of reassortants between these three virus lineages. 4, 12 ferrets have been reported as a suitable model for the study of influenza transmission as they are naturally susceptible to influenza infection, exhibit similar clinical signs (including sneezing), and possess receptor distribution in the airway similar to that of humans. [13] [14] [15] to identify molecular determinants that enable sustained human-to-human transmission, we compared the pandemic virus with genetically related swine influenza viruses obtained from this surveillance program for their ability to transmit from ferret to ferret by direct contact or aerosol transmission. viruses human h3n2 influenza virus [a ⁄ wuhan ⁄ 359 ⁄ 95 (wuhan95)] and pandemic h1n1 influenza viruses [a ⁄ california ⁄ 4 ⁄ 09 (ca04)] were included for the study. swine influenza viruses that are genetically related with the pandemic h1n1 virus were selected from our surveillance system, including classical swine-like influenza virus a ⁄ sw ⁄ hk ⁄ 4167 ⁄ 99 (h1n1) (swhk4167), triple reassortant-like a ⁄ sw ⁄ arkansas ⁄ 2976 ⁄ 02 (h1n2) (swar2976), and one reassortant between triple reassortant and eurasia swine influenza viruses [a ⁄ sw ⁄ hk ⁄ 915 ⁄ 04 (h1n2) (swhk915)]. swhk915 contains seven gene segments (pb2,pb1,pa,ha,np,m,ns) closely related to the pandemic h1n1 viruses. transmissibility was tested in 4-to 6-month-old male ferrets obtained from triple f farm (sayre, pa); all ferrets were tested to have hi titer £40 against human seasonal influenza h1n1 (a ⁄ tennessee ⁄ 560 ⁄ 2009), h3n2 (a ⁄ brisbane ⁄ 10 ⁄ 2007), and influenza b (b ⁄ florida ⁄ 4 ⁄ 2006) prior the experiments. in each virus group, three ferrets were inoculated with 10 5 tcid 50 of the virus. at 1 day postinoculation (dpi), we introduced one naïve direct contact ferret to share the cage with inoculated ferret, and one naïve aerosol contact ferret into the adjacent compartment of the cage separated by a double-layered perforated divider. nasal washes were collected every other day and tested for influenza virus antigen and to determine viral titers (tcid 50 ). weight changes, temperature, and clinical signs were monitored daily. transmission is defined by detection of virus from nasal washes and ⁄ or by seroconversion (>4fold rise in the post-sera collected after 16-18 days post contact). experiments were performed in the p2+ laboratory at st. jude children's research hospital. all studies were conducted under applicable laws and guidelines and after approval from the st. jude children's research hospital animal care and use committee. at 10 5 tcid 50 inoculation dose, all viruses replicated efficiently in the ferret upper respiratory tract with peak titers detected from inoculated ferrets at 2 dpi. lower peak titers were detected from swhk4167 and swhk915 inoculated ferrets, however, the differences were not statistically significant (table 1) . tissues collected from inoculated ferrets at 3 dpi showed that pandemic h1n1 and swine influenza viruses replicated both in the upper and lower respiratory tract of the ferrets, while the replication of human seasonal influenza wuhan95 was restricted in the upper respiratory tract. direct contact transmission from inoculated donor ferrets to their cage-mates was observed for all viruses studied, albeit at different efficiency. human seasonal influenza (wuhan95) and pandemic h1n1 viruses (ca04) transmitted most efficiently via direct contact route as the virus can be detected on 2 dpi from direct contact ferrets, and the peak titers were detected on 4 dpi from direct contacts. moderate direct contact transmission efficiency was detected from swar2976 and swhk915 viruses as the virus can be detected from direct contact ferrets at 4 dpi, with peak titers detected at 4 dpi or 6 dpi. classical swine-like swhk4167 showed least efficient contact transmission as virus could be detected from all direct contacts only at 6 dpi, and the peak titer detected on 8 dpi. aerosol transmission was detected in groups of human seasonal influenza virus wuhan95 (2 ⁄ 3), pandemic h1n1 influenza virus ca04 (3 ⁄ 3), as well as swine precursor virus swhk915 (1 ⁄ 3). transmission of wuhan95 and ca04 to aerosol contacts was detected at 4 dpi or 6 dpi, while transmission of swhk915 was detected later at 8 dpi, suggesting that the swhk915 virus possessed aerosol transmission potential, but may require further adaptation to acquire efficient aerosol transmissibility. in addition to viral detection from nasal washes, we also detected viruses from the rectal swabs of ferrets inoculated or infected with pandemic h1n1 viruses (ca04) or classical swine-like virus (swhk4167), which share the common origin for the ha, np, and ns gene segments. while many of the swine influenza viruses studied were able to transmit via the direct contact route, swhk915, which shares a common genetic derivation for seven genes with h1n1pdm, possessed capacity for aerosol transmission, albeit of moderate efficiency. swhk915 differed from swine triple reassortant viruses in the origins of its m gene. it is possible that the m gene derived from eurasian avianlike swine viruses also contributes to the transmissibility of h1n1pdm influenza viruses. outbreaks of highly pathogenic avian influenza (hpai) of the h5n1 subtype are of extreme concern to global health organisations as human infection can result in severe acute respiratory distress syndrome, multi-organ failure, and coma. hpai viruses of either h5 or h7 subtypes contain a characteristic multi-basic cleavage site in the hemagglutinin glycoprotein 1 as well as other virulence factors 2 that expand the viral tropism beyond the respiratory tract of poultry. there is also emerging evidence of viral rna or antigen in multiple organs and the cns of humans infected with h5n1 that is consistent with systemic infection 3, 4 and raises the question of the role of the cleavage site in dissemination of the virus in this species. the majority of human cases with h5n1 have involved contact with sick or contaminated poultry and exposure to respiratory secretions of birds that can be inhaled and ingested. particular risk factors for h5n1 infection include bathing with sick birds, improper hand washing after handling sick birds, or slaughtering poultry. 5 viral inoculum may also be consumed directly during a variety of religious and cultural practices, such as drinking contaminated duck blood and kissing of merit release birds. h5n1 infection is lethal in 60% of human cases, and the pathogenetic mechanisms leading to this level of mortality are unclear. to date 505 cases have been reported to the who, although many more people have potentially been exposed to h5n1 through contact with infected bird populations. 6 some studies have suggested that genetic factors may predispose an individual to severe h5n1 disease, 7 but little is known about the influence of route of virus exposure on morbidity and mortality. in ferrets, an animal model frequently used to study influenza because of its similar disease profile to humans, 8 swayne et al. 9 observed that exposure to a virulent h5n1 strain a ⁄ vietnam ⁄ 1203 ⁄ 2004 by intra-gastric gavage did not lead to disease and did not generate an antibody response, whereas ferrets that experienced a more natural exposure by being fed contaminated meat developed severe signs of infection. in this study we further assessed the disease profile of h5n1 following a natural oral exposure in the ferret model. to achieve this inoculation condition, conscious ferrets voluntarily consumed a liquid inoculum of h5n1 hpai strain a ⁄ vietnam ⁄ 1203 ⁄ 2004. as a comparison anesthetised ferrets were exposed by intranasal administration of inoculum and the ensuing disease profiles of the different routes of infection were compared. eight ferrets per group were inoculated with 10 6 egg infectious dose 50 of a ⁄ vietnam ⁄ 1203 ⁄ 2004 in a volume of 500 ll that was given to the nares of anaesthetized ferrets to establish a total respiratory tract (trt) infection or voluntarily consumed by conscious ferrets to establish an oral infection. ferrets were culled at a predetermined humane endpoint that was defined as either a > 10% weight loss and ⁄ or evidence of neurological signs, discussed in 10 ; animals that did not reach the humane endpoint were euthanased on day 14 after challenge. nasal washes and oral swabs collected during the course of infection and organ homogenates were assessed for the presence of replicating virus by growth in embryonated-chicken eggs; viral loads were determined by titration on vero cells and expressed as tcid 50 . tissue samples were fixed with formalin and embedded in paraffin for sectioning. viral lesions were identified by hematoxylin and eosin staining of the sections and the presence of viral antigen in the sections was determined by staining with antibody to influenza a nucleoprotein. pre-and post-exposure antibody responses were assessed by hemagglutination-inhibition assays using irradiated a ⁄ vietnam ⁄ 1203 ⁄ 2004 virus. the majority (75%) of ferrets infected by the trt route rapidly became inactive, developed severe disease, and were euthanased at the humane endpoint following infection ( figure 1 ). ferrets infected orally had an improved chance of survival, as only 25% of animals developed severe disease (figure 1 ), and the surviving ferrets were more active than ferrets infected by the trt throughout the stage of acute infection (data not shown). the improved survival rate and wellbeing of ferrets infected orally was not a result of poor infection rates by this route, as 5 of 6 surviving ferrets developed h5 specific antibodies by day 14 post-infection, and they did not have pre-existing antibodies to h5n1 (data not shown). the two ferrets that developed severe disease after oral infection had similar disease profiles to ferrets infected by the trt route; they both progressed to a > 10% weight loss and exhibited neurological signs (data not shown). viral loads in organs of these two ferrets confirmed dissemination to extra-pulmonary sites (table 1) : replicating virus was detected at high titres in the spleen, pancreas, liver, and brain. similar findings were recorded in ferrets with trt infections in this study (not shown) and elsewhere. 10 viral load in nasal washes and oral swabs taken at days 3, 5, and 7 post-infection by the oral route did not correlate with the development of severe disease, and virus was isolated only sporadically and at low titre from the nasopharynx of these animals (data not shown). interestingly, the two ferrets with severe disease after being infected orally had no detectable viral antigen or lesions in the olfactory epithelium and bulb (table 1) , whereas 5 of 6 ferrets culled after infection by the trt route had lesions and viral antigen in both the olfactory epithelium and bulb (data not shown). trt oral 50 100 figure 1 . percentage of ferrets that survived infection after oral or trt infection. ferrets were exposed to a ⁄ vietnam ⁄ 1203 ⁄ 2004 by the total respiratory tract (trt) route (circles) or the oral route (triangles). the percentages of ferrets that survived infection are indicated at each day following challenge. ferrets exposed orally were more likely to survive h5n1 infection than ferrets exposed to the same dose of virus by the trt. the improved survival rates that were observed after an oral infection could be a consequence of low-level viral replication in the upper respiratory tract in combination with delivery of a substantial portion of the inoculum directly to the stomach where it may have been inactivated by the harsh environment of the gastro-intestinal tract. 11 most ferrets infected orally developed an h5-specific antibody response which differs from the studies of swayne et al. 6 in which ferrets gavaged with a liquid inoculum neither developed signs of disease nor an antibody response. however swayne et al. administered virus to anaesthetized ferrets by gastric gavage that would have bypassed the oropharynx. in our study virus was administered to the oral cavity directly and would have had access to the oropharynx. low level of replication at this site may have been sufficient to trigger an antibody response. the two ferrets that developed severe disease following oral infection had a similar profile of viral dissemination as ferrets infected by the trt route. differences were seen in the olfactory epithelium and bulb as lesions, and viral antigen did not occur in these sites following oral infection, although cerebral involvement was identified. one route of dissemination of h5n1 into the cns may be by transport within nerves through the olfactory bulb into the cerebrum. 12 due to the absence of lesions and antigen in these sites following oral infection the spread of virus into the brain in these two animals may be occurring through involvement of other cranial nerves or the hematagenous routes. nasal turbinates ) 3ae00 ) ) ) ) pharyngeal lymph node interactions of oseltamivir-sensitive and -resistant highly pathogenic h5n1 influenza viruses in a ferret model <2ae5 b ) + + + + olfactory epithelium nd a nd ) ) ) ) olfactory bulb nd nd ) ) ) ) trachea <2ae5 ) ) nd ) nd lung ) <2ae5 + + ) + spleen 3ae75 ) + + ) + small intestine ) ) ) ) ) + pancreas ) 3ae50 + ) + + the pandemic potential of highly pathogenic h5n1 influenza viruses remains a serious public health concern. while the neuraminidase (na) inhibitors are currently our first treatment option, the possibility of the emergence of virulent and transmissible drug-resistant h5n1 variants has important implications. clinically derived drug-resistant viruses have carried mutations that are na subtype-specific and differ with the na inhibitor used. 1 the most commonly observed mutations are h274y and n294s in the influenza a n1 na subtype (n2 numbering here and throughout the text); e119a ⁄ g ⁄ d ⁄ v and r292k in the n2 na subtype; and r152k and d198n in influenza b viruses. 2 h5n1 influenza viruses isolated from untreated patients are susceptible to the na inhibitors oseltamivir and zanamivir, 3 although oseltamivir-resistant variants with the h274y na mutation have been reported in five patients after 4,5 or before 6 drug treatment; and the isolation of two oseltamivir-resistant h5n1 viruses with n294s na mutation from an egyptian girl and her uncle after oseltamivir treatment were described. 6 the impact of drug resistance would depend on the fitness (i.e., infectivity in vitro, virulence, and transmissibility in vivo) of the drug-resistant virus. if the resistance mutation only modestly reduces the virus' biological fitness and does not impair its replication efficiency and transmissibility, the effectiveness of antiviral treatment can be significantly impaired. the recombinant wild-type h5n1 influenza a ⁄ vietnam ⁄ 1203 ⁄ 04 (vn-wt), a ⁄ turkey ⁄ 15 ⁄ 06 (tk-wt) viruses, and oseltamivir-resistant viruses with h274y na mutation (vn-h274y and tk-h274y) were generated by using the 8-plasmid reverse genetics system. susceptibility to na inhibitors was tested by using a fluorescence-based na enzyme inhibition assay with munana substrate at a final concentration of 100 lm. viral fitness was studied in vivo in a ferret model: groups of three ferrets were lightly anesthetized with isoflurane and inoculated intranasally with vn-wt, vn-h274y, or mixtures of the two at a different ratios at a dose of 10 2 pfu in 0ae5 ml pbs; they were inoculated with tk-wt, tk-h274y, or mixtures of the two at a different ratios at a dose of 10 6 pfu in 0ae5 ml pbs. respiratory signs (labored breezing, sneezing, wheezing, and nasal discharge), neurologic signs (hind-limb paresis, ataxia, torticollis, and tremor), relative inactivity index, weight, and body temperature were recorded daily. virus replication in the upper respiratory tract (urt) was determined on days 2, 4, and 6 p.i. the competitive fitness (i.e., co-inoculation of ferrets with different ratios of oseltamivir-resistant and -sensitive h5n1 viruses) was evaluated by the proportion of clones in day-6 nasal washes that contained the h274y na mutation. na mutations were analyzed by sequence analysis of individual clones ($20 clones ⁄ sample) created by ligation of purified pcr products extracted from nasal wash samples into a topo vector. introduction of the h274y na mutation conferred high resistance to oseltamivir carboxylate in vitro; the mean ic 50 of the vn-h274y and tk-h274y viruses was 3375 and 1208 times, respectively, that of the corresponding wildtype viruses. the oseltamivir ic 50 of the tk-wt virus was $16 times that of the vn-wt virus. all four recombinant h5n1 viruses were susceptible to zanamivir. introduction of the h274y na mutation reduced $90% and 60% of the na activity of vn-h274y and tk-h274y viruses, respectively, as compared to the wild-type virus activity (p < 0ae01; two-tailed t-test). all ferrets inoculated with either vn-wt or vn-h274y virus exhibited acute disease signs (high fever, marked weight loss, anorexia, extreme lethargy), rapid progression, and death by day 6-7 p.i., and no differences in clinical signs and replication in the urt of ferrets were observed between wild-type and oseltamivir-resistant viruses ( table 1) . both of the tk viruses caused milder illness than did the vn viruses, despite a much higher dose (10 6 pfu ⁄ ferret), and the tk-h274y virus caused less weight loss and fever than the tk-wt virus (table 1) . however, competitive fitness experiments revealed a disparity in the growth capacity of vn-h274y and tk-h274y viruses as compared to their wild-type counterparts: clonal analysis established the uncompromised fitness of vn-h274y virus and the impaired fitness of tk-h274y virus (table 2) . although, the trend towards an increase ⁄decrease in the frequency of the h274y na mutation relative to the wild-type was statistically significant (p > 0ae05) for two studied groups only. mutations within the na catalytic (r292k) and framework (e119a ⁄ k, i222l, h274l, n294s) sites or near the na active enzyme site (v116i, i117t ⁄ v, q136h, k150n, a250t) emerged spontaneously (without drug pressure) in both pairs of viruses (results not shown). the na substitutions i254v and e276a could exert compensatory effect on the fitness of vn-h274y and tk-h274y viruses. the lethality and continuing circulation of h5n1 influenza viruses warrants an urgent search for an optimal therapy. our results showed that the h274y na mutation affects the fitness of two h5n1 influenza viruses differently: the oseltamivir-resistant a ⁄ vietnam ⁄ 1203 ⁄ 04-like virus outgrew its wild-type counterpart, while the oseltamivir-resistant a ⁄ turkey ⁄ 15 ⁄ 06-like virus showed less fitness than its wild-type counterpart. we used a novel approach to compare the fitness of oseltamivir-sensitive and -resistant influenza viruses that included analysis of virus-virus interactions within the host (competitive fitness) during co-infection with these viruses. although mixed populations were present in the urt of ferrets on day 6 p.i., the fitness of vn-h274y virus was uncompromised as compared to that of its drug-sensitive counterpart, while that of tk-h274y virus was impaired. a minor population of na inhibitor-resistant variants may gain a replication advantage under suboptimal therapy in two ways: (i) preexisting variants less sensitive to the drug are selected from the quasispecies population, leading to an increase of the number of resistant clones, and (ii) outgrowing variants may acquire additional compensatory mutations that enhance their fitness. it is possible that use of antiviral drugs (particularly at suboptimal concentration) against mixtures of oseltamivir-resistant and sensitive viruses will promote the spread of drug-resistant variants * ferrets in all groups inoculated with a ⁄ vietnam ⁄ 1203 ⁄ 04 virus died by day 6-7 p.i. and were observed once daily for 7 days. ** results obtained from one ferret. *** by inhibiting drug-sensitive variants that are competing with them for the dominance in the infected host. the influence of multiple genes on the fitness of viruses carrying h274 na mutation cannot be excluded. in our study we focused on additional na mutations, and sequence analysis of individual na clones 7 was done to identify potential host-dependent and compensatory na mutations. we found that the na mutations e119a and n294s, which confer cross-resistance to oseltamivir and zanamivir, 1,2 can emerge spontaneously in clade 2.2 h5n1 influenza virus in ferrets. further, we observed that mutations at na catalytic (r292k) and framework (i222l and n294s) sites and in close proximity to the na enzyme active site (v116i, i117t ⁄ v, q136h, k150n, a250t) emerged without drug pressure in both pairs of h5n1 viruses. compensatory mutations in na or other genes may mitigate any fitness cost imposed by resistance mutations. our study identified six potential compensatory na changes (d103v, f132s, i254v, e276a, h296l, and f466s) that may affect the fitness of viruses with the h274y na mutation. we suggest that na mutations at residues i254v and e276a are of importance. interestingly, we observed differences in predominance of i254v and e276a na mutations in different genetic backgrounds: i254v mutation was identified in a ⁄ vietnam ⁄ 1203 ⁄ 04 (h5n1)-like and e276a in a ⁄ turkey ⁄ 15 ⁄ 06 (h5n1)-like genetic background. moreover, i254v na mutation was identified only when ferrets were inoculated with the mixtures of vn-wt and vn-h274y viruses, but not in ferrets inoculated with vn-h274y virus. none of the potential compensatory na mutations was identified in the original inoculum used to infect ferrets. the h274y na mutation causes a large shift in the position of the side chain of the neighboring e276 residue, 8 which must form a salt bridge with r224 to accommodate the large hydrophobic pentyl ether group of oseltamivir. residue i254 is located near the na active site, and although it does not alter polarity, it results in a shorter side-chain and, thus, may indirectly affect the residues in the na active site. we suggest that antigenic and genetic diversity, virulence, the degree of na functional loss, and differences in host immune response and genetic background can contribute to the observed differences in the fitness of h5n1 influenza viruses. therefore, the risk of emergence of drugresistant influenza viruses with uncompromised fitness should be monitored closely and considered in pandemic planning. this study was supported by contract hhsn266200700005c from the national institute of allergy and infectious diseases, national institutes of health, and by the american lebanese syrian associated charities (alsac). the data presented in the manuscript have been published at: govorkova ea, ilyushina na, marathe bm, mcclaren laninamivir (r-125489) is a strong na inhibitor against various influenza viruses, including oseltamivir-resistant viruses. [1] [2] [3] [4] [5] [6] we discovered a single intranasal administration of laninamivir octanoate (cs-8958), a prodrug of laninamivir, showed a superior anti-virus efficacy in mouse and ferret infection models compared to repeated administra-tion of oseltamivir and zanamivir. [2] [3] [4] 7 this suggested that cs-8958 works as a novel long-acting na inhibitor of influenza virus in vivo. a single inhalation of cs-8958 proved noninferiority in adult patients 8 and significantly superior in child patients, 9 compared to an approved dosage regimen of oseltamivir for treatment. cs-8958 has been commercially available as an inhaled drug, inavir ò , for the treatment of influenza in japan since october 2010. the long-acting characteristics of cs-8958 are explained by several reasons. first, cs-8958 was quickly hydrolyzed to an active metabolite, laninamivir, after an intranasal administration to mice, and was retained for a long time as laninamivir in target organs, such as lung and trachea. 10 however, with an intranasal administration of laninamivir, it disappeared quickly and did not demonstrate its longlasting characteristics. 10 another reason is a strong binding of laninamivir to nas of seasonal influenza viruses compared to other three na inhibitors, oseltamivir carboxylate, zanamivir, and peramivir. 3 in the following, the tight-binding ability of laninamivir to pandemic (h1n1)2009 na, as well as to the seasonal influenza virus nas, was demonstrated. in addition, we present a hypothesis of the mechanism of the long-lasting property of cs-8958 in mouse based on a localization of an enzyme that hydrolyzes cs-8958 to laninamivir. the influenza viruses, pandemic(h1n1)2009 (inf139), a ⁄ new caledonia ⁄ 20 ⁄ 99 (h1n1), a ⁄ panama ⁄ 2007 ⁄ 99 (h3n2), and b ⁄ mie ⁄ 1 ⁄ 93 were treated with excess na inhibitors, such as oseltamivir carboxylate, zanamivir, peramivir, and laninamivir, and then unbound na inhibitors were removed from the mixtures with a bio-spin column bio-gel p-6 (bio-rad laboratories, hercules, ca, usa). the na substrate, 4-methylumbelliferyl-n-acetyl-a-d-neuraminic acid (nacalai tesque, japan) was added to the virus-na inhibitor complex, and the na activities were followed for 6 hours at room temperature by measuring the fluorescence at an excitation wavelength of 360 nm and an emission wavelength of 460 nm. the enzyme which hydrolyzes cs-8958 to laninamivir was partially purified from rat lungs using ion exchange column chromatography, and almost all bands separated by an sdspolyacrylamide gel electrophoresis were identified by mass spectrometry. the gene expression profiles of the enzyme were investigated by the bioexpress database (genelogic inc., gaithersburg, md, usa). the enzyme gene cloned from mouse lung mrna was transiently expressed in cos cells. antiserum to the esterase was prepared by immunizing rabbits, and immunostaining was done using histomouse-tm-max kit (invitorgen corp., carlsbad, ca, usa) according to the manufacturer's manual. binding stability of na inhibitors to the four viruses are shown in figure 1 the enzyme that hydrolyzes cs-8958 to laninamivir in rat lungs was identified as carboxyesterase. this esterase was shown to be expressed in epithelial cells of rat lung by in situ hybridization. 11 the mouse homolog of the rat esterase was carboxylesterase 3 (ces3). the mrna of the mouse ces3 was shown to be highly expressed in lung and liver by the gene expression profile, and ces3 was also found to contain signal sequences for retention in endoplasmic reticulum (er) and golgi at the c-terminus. the cloned ces3 gene and the ces3 gene lacking the signal sequence were exogenously expressed in the cos cells. the cs-8958-hydrolyzing activity associated with the cos cells expressing ces3 was recovered from the culture sup of the cos cells expressing ces3 lacking the retention signal sequence. localization of ces3 was immunohistologically confirmed inside the airway epithelium cells of mice, which are the target cells for influenza virus infection. the long acting property of intranasal administration of cs-8958 in mice can be explained both by the long retention of laninamivir in the respiratory tract and by the stable binding of laninamivir to influenza virus na. again, stable binding of laninamivir to na of pandemic (h1n1)2009 virus was also observed similar to that of seasonal h1n1 virus. the following are speculated as the mechanisms for the long-lasting characteristics of cs-8958 in mice. we explain the mechanism by clarifying a cs-8958 hydrolyzing enzyme and its localization inside cells. the hypothesis of the mechanism is presented in figure 2 . briefly, hydrophilic laninamivir may not enter easily inside cells, whereas hydrophobic cs-8958 may enter inside cells. ces3 with er ⁄ golgi retention signal hydrolyzes octanoate of cs-8958 figure 1 . difference of binding stabilities of various na inhibitors to influenza virus neuraminidases. the na substrate was added to the influenza virus-na inhibitor complex (oseltamivir carboxylate, n; zanamivir, h; peramivir, s; laninamivir, •; distilled water, ¤), and the na reaction was followed for 360 minutes. the background (only the na substrate [d] ) is also shown. a part of data from. 3 to generate the hydrophilic drug, laninamivir, and then it is trapped inside er ⁄ golgi because of its high hydrophilicity. the glycoprotein, na, which matures in er ⁄ golgi, meets laninamivir there and efficiently makes a stable complex with it. there are some questions that remain. how does cs-8958 move from the cell membrane to er ⁄ golgi? is laninamivir indeed trapped inside er ⁄ golgi, and does it make a complex with na in mice? we are now making an attempt to clarify these concerns. in our study, we have explored the antiviral potential of two newly synthesized compounds to provide protection against the novel pandemic influenza virus h1n1 (2009) strain. the compounds were reconstituted in dimethylsulphoxide (dmso), and so the initial studies began with cytotoxicity determination of solvent on uninfected and untreated madin-darby canine kidney (mdck) cells. on obtaining an upper limit for dmso, the compounds were tested for estimation of their maximum non-toxic dose to the mdck cells. thereafter, the effective dose of the compounds was evaluated and validated by a number of assays and gene expression profiling at both nucleic acid and protein level. we found that these newly synthesized compounds possess potent inhibitory activity towards the novel pandemic influenza h1n1 (2009) virus. these findings are being evaluated in vivo for a better understanding of their inhibitory capabilities and also their effect on the host metabolism. this will be required in the course of development of new drugs for use in the prophylaxis and treatment against the influenza virus. the mdck cell line (from nccs, pune) was maintained in 1· dmem media (sigma, st. louis, mo, usa) supplemented with 10% fetal calf serum and antibiotics viz. 100 unit ⁄ ml penicillin and 100 lg ⁄ ml streptomycin at 37°c ⁄ 5% co 2 . 5 the synthesized compounds used in this study were kindly provided by the department of chemistry, university of delhi, delhi, india. the pandemic influenza h1n1 (2009) virus was isolated and propagated in the allantoic cavities of embryonated chicken eggs during the pandemic period. the virus stocks were prepared and stored at )80°c. plaque assay was performed as previously described by hui et al., 2003. 6 briefly, 0ae5 · 10 6 mdck cells ⁄ ml were seeded in six-well plates and maintained in dmem for 24 hours at 37°c ⁄ 5%co 2 . the monolayer of the cells was inoculated with serially diluted virus samples for 45 minutes at 37°c ⁄ 5%co 2 . subsequently, a mixture of agar overlay was added, and the plates were incubated at 37°c for 5 days or until formation of plaques. the plaques were visualized after removal of the agar plug and staining with 0ae1% crystal violet or neutral red solution. the virus titre was expressed as plaque forming unit (pfu) per milliliter. the in vitro cytotoxicity analysis was performed to determine the 50% cytotoxic concentration (cc 50 ) of the compounds on mdck cells. the compounds were dissolved in dimethylsulfoxide (dmso), and so a prior cytotoxicity analysis was performed to determine the toxic concentration of dmso on the cells. various concentrations of compounds were mixed with dmem containing 1% fcs before addition to the preformed monolayer of mdck cells in 96-well plates. a series of suitable controls for in vitro cc 50 determination was included in every plate, and the plates were incubated in the optimum environment for mdck cell culture. the cc 50 of test compounds was analyzed by estimation of percentage cell viability of the compound-and mocktreated mdck cells by performing a colorimetric assay using tetrazolium salt 3-(4, 5-dimethylthiazol-2-yl)-2, 5diphenyl tetrazolium bromide (mtt) at end-point of 48 hours post-incubation. the assay was performed as described by mosman 1983. 7 briefly, mtt stock at a concentration of 0ae5 mg ⁄ ml was prepared in 1· pbs. the media was aspirated from the wells and 100 ll of mtt dye from the stock was added to each well. following incubation at 37°c ⁄ 5% co 2 for 2-4 hours, the dye was very carefully removed from the wells, and the cells were incubated with 100 ll of stop solution (dmso) per well at 37°c ⁄ 5% co 2 for 1 hour. the absorbance of the supernatants from each well was measured at 540 nm, and the percentage cell viability was calculated. madin-darby canine kidney cells were maintained overnight in a 12-well tissue culture plate at 37°c ⁄ 5% co 2 . the cells were inoculated with various virus dilutions at 37°c ⁄ 5% co 2 for 45 minutes and observed for cytopathic effect (cpe). the media from the experimental wells were aspirated after 48-72 hours of infection and were subjected to plaque assay. the percentage cell viability was determined by performing mtt assay. the results of both these tests were used to assess tcid 50 of the virus. the pre-formed monolayer of mdck cells was inoculated with the 10-fold dilution corresponding to tcid 50 of the virus for 1 hour at 37°c ⁄ co 2 . the experimental setup included control wells for the cells, virus, and compound. meanwhile, the concentrated stocks of the synthesized compounds were diluted with dmem (with1% fcs) to various concentrations within their respective cc 50 ranges. one hour post-infection, the cells were incubated with these diluted solutions. the cells were observed at various time intervals post-inoculation for cpe, and 200 ll media was collected from each experimental well for performing hemagglutination test. after 48 h, the media was collected for plaque assay and the cells were subjected to mtt cell viability assay. preformed monolayers of mdck cells were infected with virus and treated with the respective inhibitory concentration of the compounds. forty-eight to 72 hours post-incubation, total cellular rna was isolated using ribozol (amresco, solon, oh, usa) and treated with 50 lg ⁄ ml of dnase (promega, madison, usa). the concentration and quality of the rna from each well were determined by measuring their absorbance at 260 and 280 nm. one microgram of the cdna synthesized from each rna sample was used for sybr green-based real-time pcr detection of the ha gene of pandemic influenza h1n1 (2009) virus. as a control, human glyceraldehyde-3-phosphate dehydrogenase (hgapdh) was also amplified using gene specific primers. 8, 9 immunoblotting immunoblotting was performed to further validate the antiviral potential of the compounds. the experimental protocol was the same as for real time rt-pcr analysis. the cells were harvested 48 hours post-treatment with the compounds to prepare whole cell lysates in mammalian cell lysis buffer [0ae1 m nacl, 0ae01 m tris cl (ph 7ae6), 0ae001 m edta (ph 8ae0), 1 m m protease inhibitor cocktail, 100 lg ⁄ ml pmsf]. the protein concentration was determined by bca protein assay. the cell lysates were fractionated on 12% polyacrylamide for western blotting. the blot was developed using sheep monoclonal antibody (santa cruz biotechnology, ca, usa) against ha protein of influenza virus and horseradish peroxide conjugated rabbit-anti sheep igg (1:1000 dilutions) as secondary antibody. 10 the median cytotoxic concentration for compound meuh came out to be 150 lm, and that for flh was 160 lm. compounds showing potent antiviral effect on the pandemic influenza h1n1 (2009) virus propagation in madindarby canine kidney cells ( figure 1 ). the viral titres remained constant in cells treated with the compounds, while they increased in the untreated virus infected cells. ed 50 for the compounds meuh and flh were 150 and 160 lm, respectively. fifty-two percent (meuh) and 45% (flh) inhibition against the pandemic influenza h1n1 (2009) virus was achieved using ed 50 of the test compounds. both the compounds were able to reduce the rna levels of the ha gene by approximately 40-45%, whereas approximately 50% inhibition was seen when both the compounds were used in combination. similar results were obtained by the immunoblotting analysis ( figure 2 ). antiviral therapy has shown to be a promising tool in the management of various respiratory diseases, including those caused by influenza viruses. we have already shown inhibition of influenza virus replication in our earlier studies using catalytic nucleic acids, 11 which can be used as an approach in the development of new therapeutic strategy. these therapies are very useful as the influenza virus vaccines need annual renewals due to frequent genetic drifts in the viral surface proteins. in pandemic situations the existing vaccines do not provide complete protection against the novel virus as the population generally remains naïve for the newly mutated surface antigens. the antiviral drugs play an important role in the control of novel viral strains for which there are no vaccines available. however, the key obstruction in the extensive use of antiviral drugs is their cost and relative therapeutic efficacy provided. two classes of drugs were being used for treatment and control of the influenza virus infection in humans, the m2 ionchannel blockers 12, 13 (amantadine and rimantadine), which prevent viral uncoating, and the neuraminidase inhibitors 14, 15 (zanamivir and oseltmivir), which prevent the release of influenza virions from the cytoplasmic membrane. but widespread resistance to these antiviral drugs 16, 17 has limited their use. thus, novel drugs are required for the effective therapy against the emerging strains of influenza virus. the novel chemical compounds used in our study were tested for their antiviral efficacy against the pandemic influenza h1n1 (2009) virus. a reduction in the cpe in compound treated virus infected mdck cells indicated presence of antiviral activity in chemical compounds. the persistence of constant viral titers in the compound treated cells provided evidence for the interference posed by the compounds in the replication of influenza virus. inhibition in the ha gene expression further validated our hypothesis for the antiviral effect of compounds. the efficacy of these compounds in animal models is currently being validated in our laboratory. further, molecular studies are required to ameliorate the awareness regarding the mode of action of these chemical compounds against the viruses. 4 and is now licensed in japan, 5 while another, laninamivir, is being developed as an inhaled prodrug. 6 resistance to nais among circulating influenza viruses was previously low (<1% worldwide). [7] [8] [9] however, the 2007-2008 influenza season was marked by a worldwide emergence of oseltamivir-resistant seasonal influenza a (h1n1) viruses with the h275y (h274y in n2 numbering) in the na. [9] [10] [11] [12] [13] [14] the prevalence of oseltamivir resistance was even higher in the subsequent 2008-2009 influenza season with many countries reporting up to 100% oseltamivir resistance, 15 seasonal and pandemic influenza viruses collected globally between october 1, 2008 and september 30, 2009 were submitted to the who collaborating center for surveillance, epidemiology and control of influenza at the centers for disease control and prevention (cdc) in atlanta, ga, usa, and propagated in madin-darby canine kidney (mdck) cells (atcc, manassas, va, usa). reference viruses representative of oseltamivir-sensitive and -resistant seasonal and pandemic viruses were also propagated in mdck cells. susceptibilities of virus isolates to the nais oseltamivir carboxylate (hoffman-la roche, basel, switzerland) and zanamivir (glaxosmithkline, uxbridge, uk) were assessed in the chemiluminescent ni assay using the na-star tm kit (applied biosystems, foster city, ca, usa) as previously described. 9 additionally, subsets of virus isolates were tested for susceptibility to peramivir (biocryst pharmaceuticals, birmingham, al, usa). fifty percent inhibitory concentration (ic 50 ) values were calculated using jaspr curve fitting software, an in-house program developed at cdc. curve fitting in jaspr was done using the equation: v = vmax · (1 ) ([i] ⁄ (ki + [i]))), where vmax is the maximum rate of metabolism, [i] is the inhibitor concentration, v is the response being inhibited, and ki is the ic 50 for the inhibition curve. box-and-whisker plot analyses 7 of log-transformed ic 50 s were performed for each virus type ⁄ subtype and nai using sas 9.2 software (sas institute, cary, nc, usa) to identify viruses with extreme ic 50 values (outliers). outliers were characterized based on a statistical cutoff of ic 50 greater than three interquartile ranges from the 75th percentile. outliers were subjected to genetic analysis by pyrosequencing 17 and ⁄ or conventional sequencing 18 to detect known or novel markers of nai resistance. those harboring previously characterized mutations in the na associated with nai resistance were considered drug-resistant; their descriptive statistics were determined separately from naisusceptible viruses. descriptive statistics to compute the mean, median, and standard deviation (sd), and a one-way analysis of variance were performed on original scale ic 50 data, using sas 9.2 software (sas institute) for each nai and virus among seasonal influenza a (h1n1) viruses tested for oseltamivir susceptibility (n = 1533), 1431 (93ae3%) were outliers for the drug (table 1 ) and harbored the oseltamivir-resistance conferring h275y mutation in the na. by contrast, only a small proportion (0ae7%) of tested h1n1pdm viruses (n = 2259) were resistant to oseltamivir. all influenza a (h3n2) viruses (n = 834) were sensitive to oseltamivir except for one outlier, a ⁄ ontario ⁄ rv0442 ⁄ 2009 with d151v mutation in the na, whose ic 50 of 2ae50 nm was beyond the statistical cut-value off and >10-fold the mean ic 50 for the drug (0ae24 nm). all influenza b viruses (n = 914) were sensitive to oseltamivir with exception of an outlier b ⁄ texas ⁄ 38 ⁄ 2008, with d197e (d198e in n2 numbering) mutation in the na, whose ic 50 was beyond the cut-off, but only fourfold greater than the mean ic 50 for the drug. all virus types ⁄ subtypes tested for zanamivir were sensitive to the drug (table 1) , except for some outliers among seasonal influenza a (h1n1) and a (h3n2) outliers. the seasonal influenza a (h1n1) outliers included a ⁄ thailand ⁄ 1035 ⁄ 2008 (h1n1) and a ⁄ hawaii ⁄ 20 ⁄ 2008 (h1n1), both with combined h275y and d151d ⁄ g mutations in their na. the presence of concurrent mutations at na residues h275 and d151 in seasonal influenza a (h1n1) virus isolates substantially enhances resistance to oseltamivir and peramivir and ⁄ or zanamivir, however, the changes at d151 are typically cell-derived and not present in clinical specimens. 19 influenza a (h3n2) outliers for zanamivir included a ⁄ ontario ⁄ rv0442 ⁄ 2009 with d151v mutation in the na, as well as a ⁄ maryland ⁄ 02 ⁄ 2009 and a ⁄ vladivostok ⁄ 53 ⁄ 2009 with d151g and mixed d151d ⁄ g mutations, respectively. some mild outliers for zanamivir among a (h3n2) viruses with ic 50 beyond the statistical cutoff but <10-fold mean ic 50 for the drug were also identified; their genetic analysis revealed presence of wildtype and mutant sequences at residue 151 namely, d151d ⁄ g, d151d ⁄ n, or d151d ⁄ a. mutations at residue d151 of the na are associated with reduced susceptibility to zanamivir in a (h3n2) viruses, 9 but were reported to be cell-culture derived in recent h3n2 viruses. 20 all virus isolates tested for peramivir (n = 1058) were sensitive to the drug, except for h275y variants among seasonal influenza a (h1n1) and h1n1pdm viruses, which exhibited reduced susceptibility to the drug. in addition, one influenza a (h3n2) isolate, a ⁄ ontario ⁄ rv0442 ⁄ 2009 with d151v mutation in the na, showed reduced susceptibility to peramivir. the ic 50 values determined in functional ni assays provide valuable information for detection of resistant viruses, but should not be used to draw direct correlations with drug concentrations needed to inhibit virus replication in the infected human host, as clinical data to support such inferences are inadequate. 21 nevertheless, combining elevated ic 50 values with the presence of established molecular markers of resistance in the na of virus isolates and their matching clinical specimens provides a reliable and reasonably comprehensive approach of identifying nai-resistant isolates for surveillance purposes. in this study, outliers with elevated ic 50 values for oseltamivir among seasonal influenza a (h1n1) and h1n1pdm viruses were confirmed to be oseltamivir-resistant based on the presence of the h275y mutation in the na. outliers for oseltamivir and ⁄ or zanamivir among influenza a (h3n2) viruses in this study were shown to harbor mutations at d151, which were earlier associated with reduced susceptibility to zanamivir, 9 and were cell-culture derived. 20 the effects of d151 mutations on nai susceptibility appear to be strain-specific; however, there are no conclusive supporting data and further investigations are required. outliers among the influenza a viruses in this study exhibited changes in the na, derived naturally or through cell-culture, which altered their susceptibility to nais. however, mild outliers for oseltamivir and ⁄ or zanamivir among influenza a viruses with slightly elevated ic 50 s, but without apparent changes in the na are sometimes identified. in such instances it is imperative to exclude the potential presence of influenza b among such outliers, using conclusive genetic tests such as real time pcr, since influenza b viruses exhibit higher ic 50 values for oseltamivir and zanamivir than influenza a viruses. 21 viruses exhibiting such mixes are typically excluded from statistical analyses of ic 50 s for respective drugs and virus type ⁄ subtype. establishment of a clinically relevant ic 50 cutoff value which could be used to differentiate statistical outliers from truly resistant viruses is imperative. global surveillance for nai susceptibility of influenza viruses circulating globally should be sustained to reflect the impact of seasonal and pandemic of influenza, given the limited pharmaceutical options available for control of influenza infections. nasopharyngeal swab specimens from patients with acute respiratory infection were collected at 164 influenza sentinel surveillance units (outpatient and hospital-based) all over mongolia. specimens were transported to the virology laboratory, nccd, ulaanbaatar, and rt-rt pcr positive samples were grown in a mdck cell culture according to the protocol developed by cdc. 5 6 and influenza virus gene segment 7 (m genes) sequencing (7 strains-genbank accession numbers: cy053364, cy053365, cy054547, cy054549, cy055171, cy065990, and cy065998) and influenza virus gene segment 6 (na gene) sequencing (2 strains genbank accession numbers: cy073447 and cy073448) by the standard methods with applied biosystems 3130xl genetic analyzer using primers supplied by who collaboration centers. a chemiluminescent na inhibition assay was performed with veritas microplate luminometer using the commercially available kit, na-star (applied biosystems, foster city, ca, usa), according to the manufacturers protocol. the na inhibitor susceptibility of influenza virus isolates was expressed at the concentration of na inhibitor needed to reduce na enzyme activity by 50% (ic 50 ). oseltamivir carboxylate, was provided by f. hoffman-la roche ltd (basel, switzerland). na inhibition assay data were analyzed using robosage software comparing test data with the data produced by the reference na inhibitor sensitive and resistance strains, which were provided by the who influenza collaboration center, melbourne, australia. all viruses tested were sensitive to oseltamivir with two exceptions: a seasonal influenza virus a ⁄ ulaanbaatar ⁄ 1735 ⁄ 2009(h1n1) with 137ae1 nm ic 50 value and a pandemic influenza virus a ⁄ dundgovi ⁄ 381 ⁄ 2010(h1n1) with 65ae6 nm ic 50 value ( figure 1 ). there was oseltamivir resistance detected in 2ae7% (37 ⁄ 1) of seasonal a (h1n1) and in 0ae4% (262 ⁄ 1) of a (h1n1) pdm viruses. the oseltamivirresistant viruses were collected from untreated patients. in total, 18 influenza b viruses were analyzed by na inhibition assay and all were sensitive to oseltamivir. the na of both oseltamivir-resistant strains contained h275y mutation based on the sequencing analysis. the difference in the na amino-acid sequences between the mongolian oseltamivir-resistant viruses and the respective oseltamivir-sensitive reference viruses is shown in table all 37 a(h1n1) viruses analyzed for m2 channel inhibitor resistance by pyrosequencing contained the s31n mutation and, thus, were resistant to this class of anti-influenza drugs. the segment 7 sequencing revealed that 6 seasonal a(h1n1) viruses possess the common s31n mutation. of note, a single strain a ⁄ zavkhan ⁄ 8299 ⁄ 2009(h1n1) contained an unusual s31d change in the m2 protein. our study shows that the same prevalence [2ae7% (1 ⁄ 37)] of seasonal a(h1n1) viruses with h275y mutation in 2008 ⁄ 2009 season in mongolia with the published data for 2007 ⁄ 2008 season from japan. 7, 8 however the prevalence of oseltamivir resistance in japan has dramatically increased in 2008 ⁄ 2009 season to 100% (77 ⁄ 77). the observed double mutations: h275y and d354g in a ⁄ ulaanbaatar ⁄ 1735 ⁄ 2009(h1n1) strain, which have been also found in japan in 2008 ⁄ 2009 season. 7 the patient from whom the oseltamivir resistant seasonal influenza h1n1 virus has been isolated was a 1-year-old boy, living in ulaanbaatar, the capital city, without history of using oseltamivir. the patient from whom the oseltamivir resistant a(h1n1)pdm virus was isolated was a 59 year-old man, residing in the dundgovi, the southern province, also without history of antiviral treatment. according to the who data, isolation of the pandemic viruses carrying h275y change from untreated patients has been uncommon. circulation of amantadine-resistant seasonal a (h1n1) viruses has been increasing in mongolia since 2006 ⁄ 2007 influenza season. 4 all pandemic influenza a(h1n1) strains (16) tested were resistant to m2 channel inhibitors due to the presence of the s31n mutation in the m2 protein. among seasonal a(h1n1) viruses, one contained a s31d change whereas the others had s31n, the well established marker of resistance to both amantadine and rimantadine. this is the first report of detecting the s31d change in the seasonal a(h1n1) viruses. according to the cdc data (unpublished), the s31d change conferred the drug resistance in the a(h3n2) viruses according to the virus yield reduction assay. it is essential to continue the antiviral resistance surveillance of influenza virus strains circulating in mongolia to ensure the efficiency of a proper clinical management of influenza patients. (conferred by the s31n mutation). of note, the genotype 2 and genotype 3 dual resistant viruses from asia appear to be genetically similar to those previously reported dual resistant viruses from hong kong, sar. 4, 5 the genotype 4 virus was the only dual resistant virus with a nearly complete 2c genome. oseltamivir-resistance for this virus appears to be the result of a reassortment as demonstrated by the presence of the oseltamivir-resistant clade 2b na gene. although the detection of dual resistant seasonal influenza a (h1n1) viruses is still rare, there has been an increased prevalence of dual resistance viruses during the last three seasons: 0.06% (1 of 1753 tested in 2007-2008), 1.5% (21 of 1426 in 2008-2009) , and 28% (7 of 25 in 2009-2010) (v 2 p < 0.001). while the continued circulation or co-circulation of seasonal a (h1n1) viruses is uncertain, the emergence of dual resistant influenza viruses in five countries does present a public health concern, especially since dual resistant viruses would limit the options for antiviral treatment to a single licensed antiviral drug: zanamivir. moreover, the markers of resistance seen in seasonal a (h1n1) viruses also confer resistance in the more widely circulating 2009 pandemic a (h1n1) virus. and, since the acquisition of mutations in influenza a viruses typically occur through drug selection, spontaneous mutation, or genetic reassortment with another drug resistant influenza a viruses, the detection of influenza a (h1n1) viruses that are resistant to both adamantanes and oseltamivir warrants close monitoring, even if only detected at low frequency. new antiviral agents and strategies for antiviral therapy are likely to be necessary in the future. 1 heightening concern that drug resistance will likewise become prominent in pandemic viral strains and highlighting the need for antiviral drug resistance surveillance. the h275y mutation in h1n1 neuraminidase is the most common mutation conferring resistance. however, due to the high mutation rates of viruses, new mutations can be expected that will also render viral neuraminidase less sensitive to antiviral drugs. pcr methods can be used to detect previously identified mutations; however, functional neuraminidase enzyme activity inhibition testing is necessary for detecting drug resistance that results from novel mutations. the two neuraminidase enzyme inhibition assays using either the fluorescent munana or chemiluminescent na-star ò substrate are robust tools for ni susceptibility testing. the munan-a-based assay is broadly used by many groups, including many regional health organizations for ni susceptibility testing, yet no standardized protocol or dedicated kit has been in place for this assay, making comparison of data generated between different laboratories difficult. borrowing from multiple neuraminidase inhibitor susceptibility network (nisn)-published munana-based neuraminidase assay protocols, 2 we have developed a kit-based fluorescent neuraminidase assay that offers both standardization and off-the-shelf quality-controlled reagents for ni susceptibility testing and other neuraminidase assay applications. the na-fluor tm influenza neuraminidase assay reagents and protocols were optimized in comparison to published nisn protocols according to the criteria of assay performance, ease-of-use, consideration of historically used assay conditions, reagent storage stability, and environmental impact. our optimized assay conditions consists of 100 lm munana, 33ae3 mm mes, 4 mm cacl 2 , and ph 6ae5 in a 100 ll assay volume, and performing the assay for 60 minutes at 37°c following a 30 minutes preincubation of drug with the virus. these conditions are consistent with the majority of published influenza ni screening data in publication. 2 the standard na-fluor tm assay workflow for screening viral isolates for sensitivity to nis includes first titering the viral sample by neuraminidase activity to determine optimal virus concentration to be used in subsequent ic 50 determination assays. the na-fluor tm assay is an ideal tool for titering virus based on neuraminidase activity in the viral coat. titering of viral samples prior to running the ic 50 determination assays insured that assays would be performed within the fluorescence detection dynamic range of both the assay and the fluorometric instrument being used. viral titers giving rfus in the range of 2000-5000 were used for subsequent assays. comparison to traditional munana assays a primary goal of developing a standardized munana assay was to provide a standardized protocol and set of reagents that would allow for comparison of ni surveillance data between laboratories and over time. in addition, the assay should provide data comparable to historical data sets based on traditional munana-based protocols. to insure that our newly developed na-fluor tm assay met these criteria we performed side-by-side comparisons of the na-fluor tm assay to munana-based nisn protocols, as well as our na-xtd tm and na-star ò chemiluminescent neuraminidase assays to compare assay sensitivity and dynamic range and for ni ic50 determination with multiple viral isolates. for all assay comparisons, assays were performed according to respective published protocols. for direct comparison of results, an equivalent amount of virus (and concomitant neuraminidase activity) was used for each assay. the na-fluor tm assay provides low-end sensitivity (by signal to noise ratio) and dynamic range similar to nisnpublished, munana-based protocols (data not shown). these assays all show a low-end detection of approximately 0ae002 u ⁄ well and dynamic range of 2-3 orders of magnitude when performed simultaneously side-by-side using serial dilutions of bacterial (clostridium perfringes) neuraminidase. these assays show approximately onefold less dynamic range and approximately fivefold less low-end sensitivity than chemiluminescent assays under these conditions. given the large amount of archived ni inhibition data for viral isolates over the past decade, it is very important for a standardized assay to generate data similar to established protocols so that data can be compared in relative terms. when run side-by-side, na-fluor tm assay provided oseltamivir carboxylate and zanamivir ic 50 values similar to nisn-published, munana-based protocols. ic 50 values vary somewhat for munana assays versus chemiluminescent assays depending on the viral isolate, as previously described. 3 the na-fluor tm assay also exhibited similar sensitivity for detecting ni sensitive virus compared to nisn-published fluorescent assays as shown in figure 1 . the large shift in ic 50 values between oseltamivir-sensitive and resistant virus using the na-fluor tm assay enables detection of mutant virus in mixed viral samples ( figure 2 ). this capability is critical for identifying resistant virus in clinical isolates presenting mixed populations of resistant and sensitive virus during ni susceptibility surveillance. several characteristics of the na-fluor tm assay make it an ideal assay for processing large numbers of viral isolates for ni sensitivity surveillance or for using the assay for high throughput screening for lead discovery of new antiviral reagents. the na-fluor tm assay signal was found to remain stable for up to 4 hours after stop solution addition when stored at room temperature and for several days when stored at 4°c (data not shown). ic 50 values did not change over these times, indicating that the assay is compatible with processing many samples in a short time frame. the na-fluor tm assay was also found to be highly reproducible giving a z' of 0ae78 or above indicating that the assay can be used confidently to identify nis in high throughput screening mode. 4 the assay can tolerate up to 5% dmso, a common compound delivery reagent used in high throughput screens (data not shown). we have developed a standardized na-fluor tm assay suggested protocol that gives data similar to established mun-ana protocols. however, we have also found that several protocol adaptations can be made that generate comparable data while allowing the user more flexibility in assay mode, use of additional reagents, and to meet user-specified assay time requirements. the na-fluor tm assay can be run in either the standard 60 minutes ⁄ 37°c endpoint mode described above or as real-time kinetic assay 5 with repeated reads taken over time without the addition of stop solution, which both serves to terminate neuraminidase activity and to enhance the fluorescence of the product. for typical ni-sensitive viral strains, the rate of munana substrate turnover at 37°c is linear for at least 4 hours (data not shown). as would be expected, rates of substrate turnover decrease in the presence of nis reflected in a decreased slope exhibited by real-time kinetic reads. real-time acquired rfus are typically 5-6 fold lower than rfus acquired after addition of stop solution at the same time point. ic 50 values obtained using slope analysis for real-time assays are similar to values obtained by endpoint analysis. whether run in real-time or end-point mode, the linear rate of substrate turnover allows the user to run the assay for shorter or longer assay times than the standard protocol without compromise to assay performance. the na-fluor tm assay is also compatible with standard methods used in many laboratories to inactivate virus. 6 we have shown that ni ic 50 values for multiple viral strains remain unchanged when the assay is performed in the presence of 0ae1% np-40 or 1% triton x-100 (data not shown). similar results are also obtained by adjusting the na-fluor tm stop solution to 40% ethanol prior to addition for assay termination. the assay is unaffected by phenol red concentrations present in cell culture media. we have developed a standardized munana-based fluorescent neuraminidase assay, the na-fluor tm influenza neuraminidase assay kit, which has been optimized for ni susceptibility screening. the assay provides data that can be compared to data generated using traditional munanabased protocols. the assay is economical, highly reproducible, easy to use, and environmentally friendly. the assay is flexible and amendable to user-specific adaptations including assay mode, assay timing, and reagent compatibility. trademarks ⁄ licensing ª 2010 life technologies corporation. all rights reserved. relenza is a registered trademark of glaxo to test the prophylactic potency of h5-vhhb, mice were treated intranasally with pbs, 100 lg of h5-vhhb, or negative control rsv-vhhb at 4, 24, or 48 hours before infection with one ld 50 of nibrg-14ma virus. body weight loss was monitored daily, and on day 4 mice were sacrificed to determine the viral load in the lungs. all mice that received h5-vhhb retained their original body weight, whereas those receiving pbs or rsv-vhhb gradually lost weight (data not shown). intranasal administration of h5-vhhb at 4 or 24 hours before challenge resulted in undetectable lung virus titers. when animals were treated with h5-vhhb 48 hours before challenge, virus titers were 50fold lower compared to pbs and rsv-vhhb treated mice, and three out of seven animals still had undetectable virus titers ( figure 1 ). we next determined if h5-vhhb nanobody ò could be also be used therapeutically. we administered 60 lg of this nanobody ò intranasally to mice up to 72 hours after chal-lenge with 1 ld 50 of nibrg-14ma virus. four days after challenge, animals that received h5-vhhb 4, 24, or 48 hours after challenge had significantly higher body weight (data not shown) and lower lung virus loads than control mice. although mice treated with h5-vhhb nanobody ò 72 hours after challenge were not clinically protected compared to control mice, they had significantly lower lung virus titers (figure 1 ). to identify the ha amino acid residues that are potentially involved in h5-vhh binding, escape viruses were selected by growth and plaque purification of nibrg-14ma virus in the presence of h5-vhhm or h5-vhhb nanobodies ò . the ha sequences of six independently isolated h5-vhhm escape viruses revealed substitution of a lysine by a glutamic acid residue at position 189 in ha1 (h5 numbering). in addition, two h5-vhhm escape mutants carried an n154d and four carried an n154s substitution. the three-dimensional structure of nibrg-14 ha shows that n154d ⁄ s and k189e are close to each other as part of the corresponding antigenic site b in h3 ha. 4, 5 interestingly, the n154d ⁄ s mutations remove an n-glycosylation site, which is surmised to have evolved in h5n1 ha as a strategy to mask an antigenic site. 6 escape viruses selected in the presence of h5-vhhb carried k189n (n = 4) or k189e (n = 2) substitutions. these results indicate that residues in antigenic site b, at the top of ha and very close to the receptor binding domain (rbd), are essential for neutralization of the virus by h5-vhhm ⁄ b nanobodies ò (figure 2 ). the virus titer was measured in lung homogenates prepared on day 4 after challenge. the x axis refers to the time points in hours relative to the challenge (time = 0 hours) when ha-specific nanobodies (h5-vhhb), control nanobodies (rsv-vhhb) or pbs was administered to the mice. # below detection limit, n not determined [n = 4-7 mice per condition: p values < 0ae05 (*)]. here we demonstrated that prophylactic and therapeutic treatment with llama-derived immunoglobulin single variable domain fragments is effective to control infection with h5n1 influenza virus in a mouse model. we demonstrate that pulmonary delivery is a highly effective route of administration to treat or prevent influenza virus infection. in addition, we demonstrate that a homobivalent h5-vhhb has powerful h5n1-neutralizing activity in vivo. it is important to note that we used a mouse-adapted derivative of the non-highly pathogenic nibrg-14 virus in our challenge model. nevertheless, this virus induces severe morbidity and lethality in mice. compared to conventional neutralizing monoclonal antibodies, vhhs offer the advantage that they are easy to produce in escherichia coli, typically with high yield. in addition, their small size (15 kda for a monovalent vhh) and high folding capacity allow the generation of oligovalent vhh derivatives. in vitro escape selection revealed that a k189e substitution in ha1 abolished the neutralizing effect of h5-vhhm ⁄ b. a lys or arg residue at this position is conserved in all human h5n1 virus isolates. of note, all selected escape mutants contained a glutamic acid or serine residue at position 189, which suggests that the conserved positively charged amino acid is important for neutralization by h5-vhh nanobodies ò . interestingly, escape mutants selected with h5-vhhm also carried an n154d ⁄ s co-mutation that removes an n-glycosylation site in this antigenic site of ha. the predicted n-glycosylation site at n154 in a ⁄ hong kong ⁄ 156 ⁄ 97 ha was shown to be glycosylated and may have evolved to mask an antigenic site near the rbd. 7, 8 the selected amino acid changes are located near the receptor binding site of ha. therefore, it is possible that enhanced receptor binding properties of these escape viruses contribute to or are responsible for the loss of neutralizing activity of h5-vhh nanobodies ò . 9, 10 we conclude that influenza virus neutralizing nanobodies ò have considerable potential for the treatment of h5n1 virus infections. although we focused on vhhs that presumably recognizes an epitope near the rbd, it is possible to select vhh molecules that bind to other epitopes in ha, including more conserved domains. more, a novel na (i117m) substitution was discovered in a series of specimens from a patient. for the amantadine resistance, 1113 samples were tested, and all of them were confirmed to be resistant. we collected respiratory specimens from patients who had been clinically refractory to antiviral treatment since october 2009 upon ethical approval from the relevant institutions. to investigate the resistant pattern, sequence analysis to the na and matrix (m2) genes were conducted by reverse transcription (rt)-pcr and sequencing reaction. the obtained sequences were analyzed by the influenza sequences and epitopes database, which was developed in korea. eleven patients were found to be having oseltamivir-resistant pandemic (h1n1) 2009 viruses with the h275y substitution in the viral na genes (tables 1 and 2 ). some cases were associated with oseltamivir treatment on the basis of h275y change from the oseltamivir-sensitive genotypes to oseltamivir-resistant genotypes in consecutive samples from the same patient. furthermore, a novel na (i117m) substitution that may be associated with oseltamivir resistance was detected in specimens from one patient (patient g) who had myelodysplasia and received oseltamivir and peramivir (tables 1 and 2 ). in addition, we obtained viruses from clinical specimens (patients a and c) and evaluated antiviral susceptibility by measuring the dose of oseltamivir and zanamivir required for 50% inhibition (ic 50 ) of na activity. these viruses (from patients a and c) were resistant only to oseltamivir (ic 50 713ae2 and 359ae4 nmol ⁄ l, respectively). susceptibility to zanamivir was not altered whether na contained y275 or h275 (ic 50 0ae13 and 0ae78 nmol ⁄ l, respectively). one isolate of pandemic (h1n1) 2009 virus with an oseltamivir-sensitive genotype (h275 in its na) was susceptible to oseltamivir (ic 50 1ae18 nmol ⁄ l) and zanamivir (ic 50 0ae42 nmol ⁄ l). patients with oseltamivir-resistant pandemic (h1n1) 2009 were treated during hospitalization with oseltamivir alone or with a combination of other antiviral drugs ( we found 11 patients of oseltamivir resistance with h275y mutation in the na gene of pandemic (h1n1) 2009 virus through the surveillance of patient refractory to antiviral treatment. in addition, novel amino acid change (i to m) at position 117 in the na gene, which might influence oseltamivir susceptibility, was detected in sequential specimens of a patient. these data showed that generation of oseltamivir resistance could be associated with oseltamivir treatment. therefore, it needs to strengthen the antiviral monitoring by supplementation of the clinical data including antiviral treatment. during the pandemic, oral oseltamivir was the primary antiviral medication used for treatment of hospitalized patients with ph1n1 infection. many physicians worried that clinical deterioration or failure to respond to treatment with oseltamivir was due to either oseltamivir resistance or oseltamivir failure. in the united states, two investigational intravenous (iv) nais were available during 2009-2010: peramivir through emergency use authorization and zanamivir by investigational new drug application. peramivir would be an option for patients with oseltamivir failure, but would not be appropriate for patients infected with h275y oseltamivir resistant mutants. iv zanamivir was available in limited supply, but would be appropriate for severely ill patients infected with an oseltamivir-resistant ph1n1 virus. during the pandemic, clinicians had few options for antiviral resistance testing in the united states. to respond to this need, the us centers for disease control and prevention (cdc) offered antiviral resistance testing for patients suspected to have clinical failure due to oseltamivir resistance. we describe the methods that cdc used to prioritize patients for testing during the pandemic and to detect markers for oseltamivir resistance, as well as the results from this testing. to facilitate decisions on which patients to test, we developed testing algorithms that were shared with state labora-tories, epidemiologists, and the emergency operation center at cdc. we prioritized patients who might benefit the most from antiviral testing given the inherent delay in providing antiviral results, e.g. patients who might have prolonged ph1n1 shedding. patients that were critically ill [intensive care unit (icu) admission] or patients with severe immunocompromising conditions with clinical evidence for oseltamivir treatment failure (persistent detection of virus and clinical unresponsiveness to the drug) were prioritized. in addition, we tested specimens from patients that failed oseltamivir chemoprophylaxis. standard forms with information regarding specimen and minimal clinical information were collected on all patients. all protocols were validated and approved by clinical laboratory improvement amendments, e.g. quality standards to ensure accuracy, reliability, and timeliness of patient test results. information collected on patients was deemed public health response, not research, at cdc. clinical specimens, confirmed as 2009 pandemic influenza a (h1n1), were tested for the h275y mutation in the na using pyrosequencing. 5 results were returned to sender within 24-48 hours of specimen receipt. from october 2009 until july 2010, a total of 116 specimens from 73 patients were submitted for testing. viruses from 26 (36%) of 73 patients had h275y mutation in the na in at least one submitted specimen. clinical information was available for 66 patients (table 1) . most patients had received oseltamivir for treatment prior to obtaining the specimen sent for antiviral testing. four patients received oseltamivir for chemoprophylaxis, all were immunosuppressed, and all had the h275y mutant; duration of chemoprophylaxis until ph1n1 infection was detected varied (4-39 days). among the patients with an h275y mutant who were treated with oseltamivir, the median time on oseltamivir prior to collection of specimen with h275y mutation was 18 days (range 3-36 days). three patients were part of a hospital cluster of oseltamivir-resistant virus infections and were infected with h275y mutants prior to oseltamivir treatment. 6 patients with immunocompromising conditions accounted for almost half of all patient specimens tested, but they accounted for the majority of oseltamivir-resistant ph1n1 virus infections (table 1) ; among 32 individuals with severe immunocompromising conditions and clinical failure while on oseltamivir therapy, 23 (72%) had the h275y mutant detected. among the immunosuppressed patients with an oseltamivir-resistant virus, 16 (70%) had hematologic malignancies reported. in contrast, among the subset of icu patients without immunocompromising conditions and clinical failure while on oseltamivir therapy, we found little resistance: 2 (5ae9%) of 34 icu patients had oseltamivir resistance detected. during the 2009 pandemic, we were able to provide timely and useful information to clinicians regarding suspected cases of oseltamivir resistance. our testing algorithm limited the number of specimens to specimens from the highest risk patients that would benefit the most from antiviral treatment. such an approach allowed us to offer this service without compromising our public health duties. in addition, the information we collected on patients from this service complimented our data on the national surveillance for antiviral resistance. we also performed national antiviral resistance surveillance from april 2009 to july 2010. overall, 67 resistant ph1n1 viruses were identified from april 2009 to july 2010 in the united states among 6812 tested samples, including specimens described above, surveillance specimens, and resistant viruses reported in the literature. 7 further studies to understand risk factors for oseltamivir-resistant ph1n1 infection in patients with severe immunocompromising conditions are needed. while efforts to provide antiviral testing technology and materials to state laboratories are ongoing, clinicians still have limited options for such testing. rapid and inexpensive assays that could be performed by clinical laboratories, especially those caring for immunosuppressed patients, would be useful to inform patient care. the applied biosystems ò na-xtd tm influenza neuraminidase assay kit provides the next-generation na-xtd tm 1,2-dioxetane chemiluminescent neuraminidase (na) substrate, together with all necessary assay reagents and microplates, to quantitate sensitivity of influenza virus isolates to neuraminidase inhibitors. like the na-star ò influenza neuraminidase inhibitor resistance detection kit, the na-xtd tm influenza neuraminidase assay provides highly sensitive detection of influenza neuraminidase activity. in addition, the na-xtd tm assay provides extended-glow light emission that eliminates the need for reagent injection and enables signal measurement either immediately or up to several hours after assay completion. the na-xtd tm assay is also used to quantitate influenza na activity directly in cellbased virus cultures to monitor viral growth or inhibition. global monitoring of influenza strains for resistance to neuraminidase inhibitors (nis) is essential for understanding their efficacy for seasonal, pandemic, or avian influenza, and studying the epidemiology of viral strains and resistance mutations. functional neuraminidase inhibition assays enable detection of any resistance mutation, making them extremely important for global monitoring of virus sensitivity to nis. the first-generation chemiluminescent na-star ò influenza neuraminidase inhibitor resistance detection kit has been widely used for virus ni sensitivity assays, 1-8 including identification of a ⁄ h1n1 pandemic virus resistant to oseltamivir. 9, 10 in addition, this assay has been used for identification of new ni compounds, 11 ni characterization, 12 studies of virus transmission, 13 drug delivery, 14 na quantitation of virus-like particles, 15 and cell-based virus quantitation. 16 neuraminidase assays performed with chemiluminescent 1,2-dioxetane substrates, including na-star ò and na-xtd tm substrates, typically provide 5-to-50-fold higher sensitivity by signal-to-noise ratio than assays performed with the fluorescent munana substrate. in addition, chemiluminescent assays provide linear results over 3-4 order of magnitude of neuraminidase concentration compared to 2-3 orders of magnitude with the fluorescent assay. 2 the high assay sensitivity achieved with chemiluminescent assays enables use of lower concentrations of viral stocks, and the wide assay range minimizes the need to pre-titer virus stocks prior to ic 50 determination. chemiluminescent reactions result in conversion of chemical energy to light energy, as light emission. the na-xtd tm substrate is a 1,2-dioxetane structure bearing a sialic acid cleavable group. to perform the na-xtd assay, virus dilutions (from cell culture supernatant) are pre-incubated in the presence of neuraminidase inhibitor. then na-xtd substrate is added and incubated for 30 minutes for substrate cleavage to proceed. finally, light emission is triggered upon addition of na-xtd accelerator, which provides a ph shift and a proprietary polymeric enhancer, both required for efficient light emission. chemiluminescent assays are performed in solid white microplates, and light emission is measured in a luminometer. the na-xtd tm substrate has a single structural difference from the na-star ò substrate that provides a much longer-lasting chemiluminescent signal, with a signal half-life of approximately 2 hours (not shown), compared to $10 minutes with the na-star assay, eliminating the need for luminometer instruments equipped with reagent injectors and enabling more convenient batch-mode processing of assay plates. the na-xtd tm assay kit also provides a new accelerator solution, containing a next-generation polymer enhancer, and a triton ò x-100-containing sample prep buffer providing enhanced na activity. read-time flexibility is demonstrated by determination of oseltamivir ic 50 values using data collected over 3 hours after addition of na-xtd tm accelerator. although signal intensity slowly decreases over time, the ic 50 curves and values are identical at each time point, shown using influenza b ⁄ lee ⁄ 40 ( figure 1) . triton x-100 detergent at 1% has been shown to inactivate flu virus while increasing neuraminidase activity. 17 the addition of na sample prep buffer (containing 10% triton x-100) to virus stocks (at 1 ⁄ 10 volume, achieving a final concentration of 1%) provides increase in na activity up to fourfold, but is not consistently observed, and seems to be most effective with more concentrated virus stocks. ic 50 values are unaffected by the addition of triton x-100 to the virus stock prior to virus dilution (not shown), so the assay is compatible with known virus inactivation reagents. assay sensitivity and ic 50 values determined with the na-xtd assay have been compared to those obtained with both the chemiluminescent na-star assay and the fluorescent na-fluor assay (not shown). the chemiluminescent assays provide 2-to 50-fold higher sensitivity by signal-to-noise ratio, depending on the virus strain, wider assay dynamic range, and better low-end detection limit than the fluorescent assay. the wide assay range with the chemiluminescent assays enables determination of ic 50 values over a range of virus concentrations, eliminating the need to titer virus prior to performing ic 50 determination assays. ic 50 values obtained with the na-xtd assay are nearly identical to those obtained with the na-star assay, with both oseltamivir and zanamivir neuraminidase inhibitors, and tend to be slightly lower than ic 50 values obtained with the fluorescent assay. viral na quantitation provides a convenient read-out to measure viral growth or inhibition, including inhibition in the presence of inhibitory compounds or antibodies, described as accelerated viral inhibition with na as readout assay (avina). 16 bation in the presence of varying concentrations of oseltamivir carboxylate. samples of culture media were assayed 24 hours later. quantitation of na activity with the na-xtd tm assay demonstrates inhibition of viral growth by oseltamivir carboxylate in cell culture ( figure 2 ). different volumes of culture media were assayed with the na-xtd assay, either in the culture plate or in a separate assay plate (not shown). performing the assay using the entire well contents (100 ll) reduces assay sensitivity due to the high concentration of phenol red. assaying a smaller volume of culture medium (either in culture plate or a separate assay plate) provides higher sensitivity, and enables temporal monitoring or use of remaining culture medium for other assays. the applied biosystems ò na-xtd tm influenza neuraminidase assay kit is a next-generation chemiluminescent neuraminidase assay providing high assay sensitivity and ''glow'' light emission kinetics for improved ease-ofuse. the applied biosystems ò na-fluor tm influenza neuraminidase assay kit, based on the fluorescent mun-ana substrate, has also been developed to complement the na-xtd tm and na-star ò chemiluminescence assays, for users lacking luminometer instrumentation or choosing to use fluorescence assay detection. together these kits offer: • standardized reagents and protocols • choice of detection technology • simple instrumentation requirements • high sensitivity for use with low virus concentrations • compatibility with batch-mode processing and largescale assay throughput • broad specificity of influenza detection • flexibility in assay format • additional na assay applications -cell-based viral assays, screening for new nis, detection of na from other organisms functional neuraminidase inhibition assays enable detection of any resistance mutation and are extremely important in conjunction with sequence-based screening assays for global monitoring of virus isolates for ni resistance mutations, including known and new mutations. together, these assays provide highly sensitive, convenient and versatile assay systems with standardized assay reagents, and simple assay protocols for influenza researchers. 15 over 200 000 hospitalizations and 36 000 deaths in the us annually are attributable to seasonal influenza, primarily in chronically ill persons and the elderly. 1-3 following the emergence of pandemic 2009 h1n1 influenza, severe illnesses have also been observed in children and young healthy adults. 4 the occurrences of staphylococcal and pneumococcal pneumonia complicating influenza pandemics are well described. [5] [6] [7] although temporal associations of bacterial pneumonia and influenza circulation have been reported, there is little precise data on rates of bacterial complications of seasonal or pandemic influenza. the study of bacterial lung infection has been hampered by insensitive tests for invasive disease and the difficulty of interpreting routinely obtained sputum culture results. 8, 9 procalcitonin (proct), the prohormone of calcitonin, can discriminate viral and bacterial infections. 10 this 116-aminoacid precursor protein normally produced by neuroendocrine cells of the lungs and thyroid gland was first shown to be elevated in bacterial infections in patients with pulmonary injury and pneumonitis. 11 stimuli of proct include tnf-a, endotoxin, and other bacterial products. 12 several studies indicate that bacterial infections commonly induce hyperprocalcitonemia, but that viral infections, including 2009 h1n1, are associated with only minimal increases. 10, 12, 13 of note, proct induction is attenuated by viral-induced interferon-c. 14 a meta-analysis of studies comparing proct and crp as markers for bacterial infection found that proct was more sensitive and specific than crp for differentiating bacterial from other causes of inflammation. 15, 16 therefore, we measured proct levels in patients with seasonal and pandemic influenza and compared results with conventional methods for bacterial diagnosis. adults ‡21 years of age admitted to rochester general hospital (rgh) from november 1st to june 30th for two winter seasons (2008-2010) with an admitting diagnosis compatible with acute respiratory tract infection were recruited for the study. patients were screened within 24 hours of admission, and those with prior antibiotic use, immunosupression, or pregnancy were excluded. subjects or their legal guardian provided written informed consent. the study was approved by the university of rochester and rgh research subjects review board. at enrollment demographic, clinical and laboratory information was collected. influenza testing included nosethroat swabs (nts) for rapid antigen, viral culture, and reverse transcription-polymerase chain reaction (rt-pcr) and serology. testing for bacterial pathogens included blood cultures, sputum for culture and gram stain, nts for mycoplasma pneumoniae and chlamydophila pneumoniae pcr, s. pneumoniae antigen testing, and pneumococcal serology. if patients were unable to expectorate, sputum was induced with normal saline and bronchodilators. specimens were considered adequate by the standard criteria of >25 neutrophils (pmns) and <10 epithelial cells per high power field. serum was collected at admission and hospital day 2 for proct measurements. influenza infection was defined a positive result for any of the following tests: . cloned proteins were coated on eia plates at 2 ug ⁄ ml in bicarbonate buffer. after overnight incubation, plates were washed and two-fold dilutions of serum were incubated overnight at room temperature. plates were washed and incubated with alkaline phosphatase conjugate for 3 hours, followed by substrate. a greater than or equal to fourfold rise in titer was considered evidence of infection with s. pneumoniae. urinary antigen for s. pneumoniae samples were assayed for antigen using the binax now kit. (binax inc, scarborough, me, usa). the proct was measured using time resolved amplified cryptate emission technology (kryptor pct; brahms, henningsdorf, germany). functional sensitivity is 0ae06 ng ⁄ ml (normal levels are 0ae033 ± 0ae003 ng ⁄ ml). 18 mycoplasma and chlamydia pcr real-time pcr targeting the p1 adhesion gene for m. pneumoniae and the ompa gene for c. pneumoniae was used to detect atypical bacteria. 19 results fifty-one of 529 (9ae6%) illnesses evaluated tested positive for influenza virus. of these, 20 were due to ''seasonal influenza'' (16 influenza a ⁄ h3n2 and 4 influenza b), and 31 were identified as ''pandemic influenza'' (2009 h1n1). demographics of both groups were similar: mean ages 55 ± 18 and 50 ± 11 years, respectively, and equivalent sex and racial characteristics. other than a higher incidence of underlying lung disease in the seasonal group (75% versus 45%, p = 0ae05), pre-existing medical conditions including obesity were similar. symptoms, physical findings, and discharge diagnoses did not differ, and chest radiographs (cxr) showed infiltrates in 20% and 13% of seasonal and pandemic subjects, respectively. two pandemic and one seasonal influenza patient developed respiratory failure, and none died. overall, bacterial infections were diagnosed in 8 (16%) subjects (4-seasonal and 4-pandemic), and none were bacteremic. bacterial infections included: 5-s. pneumoniae, 1-m. pneumoniae, 1-s. aureus, and 1-h. influenzae. all seasonal patients were diagnosed with asthma or bronchitis, whereas three pandemic patients had pneumonia. mean serum proct (ng ⁄ ml) levels in seasonal versus pandemic patients on admission and day 2 were: 0ae14 ± 0ae17 versus 1ae05 ± 2ae86 and 0ae17 ± 0ae17 versus 1ae39 ± 4ae37, respectively, and were not significantly different (table 1) . several patients in the pandemic group had high proct levels, and there was a trend toward more pandemic patients having admission proct values ‡0ae5 ng ⁄ ml than seasonal subjects [1 (5%) versus 6 (19%), p = 0ae22] ( figure 1a , b). of the four patients with proc-t > 1ae5 ng ⁄ ml, two had dense infiltrates on cxr, one had a peripheral wbc of 17 200 ⁄ ml with a threefold increase in s. pneumoniae antibody, and one developed respiratory failure associated with copd exacerbation. reliable sputum samples (within 6 hours of antibiotics) were collected in only 22 (43%) subjects. of these, proct was ‡0ae5 ng ⁄ ml in two with influenza alone and three associated with bacterial infection, and <0ae5 ng ⁄ ml in 12 with influenza alone and five associated with bacterial infection. in the 22 with reliable sputa and accepting the conventional bacterial diagnosis, sensitivity of a proc-t ‡ 0ae5 ng ⁄ ml for bacterial infection was 38%, specificity 86%, positive predictive value 60%, and negative predictive value 71%. notably, one patient considered to have influenza alone (proct -0ae87 ng ⁄ ml) had group a streptococcus and s. aureus in a contaminated sputum and bilateral infiltrates on cxr. three of five patients with bacterial infections and proct < 0ae25 ng ⁄ ml had a clinical diagnosis of bronchitis. mean proct values were significantly higher in patients with infiltrates versus those with atelectasis or no acute disease on cxr (2ae4 ± 4ae8 ng ⁄ ml versus 0ae36 ± 1ae2 ng ⁄ ml, p = 0ae02). combining patients with proct values ‡0ae5 ng ⁄ ml with those having positive bacterial tests, rates of bacterial infection associated with seasonal and pandemic influenza were 20% and 26%, respectively. notably, antibiotics were administered to 88% of subjects despite 73% having no acute disease on cxr. in our study, bacterial infections were diagnosed in approximately 25% of adults hospitalized with influenza with no significant difference in rates noted between seasonal and pandemic influenza infected subjects. previous reports of bacterial infection rates of 4-15% with seasonal influenza are difficult to compare with recent studies of pandemic influenza, because the latter tended to focus on more severely ill patients. [20] [21] [22] bacterial pneumonia has been suspected or diagnosed in 13-24% of patients in intensive care associated with 2009 h1n1 infection and up to 38% of patients who died. 23, 24 despite aggressive pursuit of specimens for bacterial testing, diagnoses could be confirmed in only 8 (15ae6%) of patients using conventional methodology. given the difficulty in establishing a diagnosis of bacterial infection, elevated proct values may be helpful to identify patients at high risk for invasive disease. in a study of 25 patients with severe 2009 h1n1 or bacterial infection necessitating intensive care, a threshold proct level of 0ae8 ng ⁄ ml, demonstrated 100% sensitivity and 62% specificity for bacterial infection. 25 among 38 patients with 2009 h1n1 associated pneumonia, many of whom had respiratory failure, a threshold proct value of 0ae5 ng ⁄ ml provided a sensitivity of 100% and specificity of 53% for bacterial infection. 26 access to samples from lower airways in ventilated patients in these studies may have improved recovery of bacteria and account for the different results we observed. it should be noted that none of our patients were bacteremic, which is a very strong stimulus for proct release. proct levels have been used successfully to guide therapy in community acquired pneumonia, and our data showing high proct levels in patients with infiltrates on cxr suggests proct may be most useful for excluding invasive disease. 27, 28 elevated proct levels were not observed in patients with purulent sputum and clear cxr. it is notable that a proct level of <0ae25 ng ⁄ ml did not exclude patients with bacterial bronchitis since proct has been used to guide antibiotic therapy in copd exacerbations. 29 while it could be argued that healthy patients with bacterial bronchitis do not require antibiotic treatment, physician behavior in our study indicates antibiotics are frequently prescribed. combining patients with proct values ‡0ae5 ng ⁄ ml and those with a positive bacterial test, approximately 25% in patients in our study had bacterial complications associated with influenza infection. efforts should be made to curtail antibiotic use in hemodynamically stable patients with clear cxrs. given physician discomfort regarding discontinuing antibiotics, proct measurements in combination with routine bacterial cultures should be useful tools to guide therapy. influenza, mrsa, cytokines: diagnosis, treatment, prevention -a possible strategy for outpatient care we started the antiviral treatment of influenza in humans using neuraminidase inhibitors on january 29, 1999 in a successful attempt to cure a 96-year-old patient. since then, we have used the inhalant antiviral drug zanamivir, and later (october 1, 2002) changed to the use of oseltamivir with systemic bioavailability for treating patients with influenza. after 10 years of experience with antiviral treatment of outpatients, we highlight the importance of early diagnosis and early treatment. 1 the necessity of an earliest possible diagnosis was confirmed in the pandemic of 2009. large hospitals reported that patients with an h1n1 ⁄ 09 infection had to be treated with extracorporeal membrane oxygenation. we are convinced this is due to delayed recognition of infection in most cases. valuable time is lost when the patient with a sudden onset has to be brought to a hospital for emergency treatment. the point at which the patient goes to the doctor is decisive, and this problem of timing and the delivery of early treatment is not specific to germany. in our medical office, we assessed patients with suspected influenza (to date 262 seasonal infections, and in 2009, 25 h1n1 ⁄ 09) through clinical diagnosis, 2 and then proven by point of care rapid test (quickvue; quidel, san diego, ca, usa) followed by pcr. all of the patients undergo concomitant lab tests: leukopenia, serum iron level, and the humoral inflammation status [sum of the c-reactive protein (crp) and fibrinogen levels]. because of the constant threat of a bacterial superinfection, a bacterial swab and antibiogram is carried out on every patient. in all cases positive for influenza, oseltamivir was given immediately. nowadays it is important that a double infection with influenza and mrsa must be recognized immediately and treatment started at once with antivirals and, when appropriate, with a suitable antibiotic. we pay particular attention to an extremely low iron level (signum mali ominis). in addition we monitor oxygen saturation and the course of the humoral inflammation status every 1-2 hours for every of our outpatients. among our patients with seasonal influenza, we saw 132 within 12 hours, 103 within 36 hours, and 27 within 48 hours after disease onset. for pandemic influenza, it was 12 patients within 12 hours, 11 within 24 hours, and two within 36 hours. for all patients, we measured crp < 3ae5 mg and fibrinogen < 350 mg ⁄ dl (12 hours), crp < 10 mg and fibrinogen < 450 mg ⁄ dl (36 hours), and crp > 10 mg and fibrinogen < 500 mg ⁄ dl (48 hours, only seasonal cases). antibiotics were necessary in 130 cases, heparin and oxygen administration in 27 cases. one hundred forty-eight patients had a superinfection following influenza. the most common strains were haemophilus parainfluenzae and staphylococcus aureus. the subsequent use of a suitable antibiotic was only necessary in 50% of the patients. in all cases diagnosed, treatment (including heparin and oxygen administration) and monitoring were conducted in our medical office. none of our patients (seasonal and pandemic) had to be admitted to hospital. the early decision of whether or not antiviral and antibacterial treatment is taking effect is the only way the threat of a cytokine storm can be averted. not only does the primary care physician have to be aware of the pathophysiology involved, but also the necessary diagnostic and therapeutic options have to be made available to him. the result will lead to a saving of both lives and healthcare costs. this applies both in epidemic as well as in pandemic times. today we know that influenza leaves behind a defenceless immune system, and that the proteases of s. aureus contribute to influenza associated pneumonia. mark von itzstein, who discovered neuraminidase inhibitors, emphasized the synergistic cooperation of viruses and bacteria (personal communication, 2009). mrsa and influenza viruses are posing problems worldwide. the case of a 17-year-old boy with h1n1 ⁄ 09 infection demonstrates how fatal developments can be prevented. due to his constantly recurring colds, we had already detected the mrsa colonization years earlier and had always worked on boosting his general health and resistance. both the patient and his family were included in dealing with the problem. the patient was, and is, always vaccinated early with a virosomal vaccine (baxter). during the oktoberfest in munich in september 2009, when h1n1 infections were increasingly occurring, we learned that our patient had come down with an extremely acute feverish illness. with the help of the rapid test, we diagnosed an h1n1 ⁄ 09 virus infection and started treatment with oseltamivir immediately. the humoral inflammation status, which had increased very rapidly to more than 4ae5 mg ⁄ dl within hours, was treated with the effective cotrimoxazol from the antibiogram. at the same time, the patient was heparinized. the following day the patient had no fever and was symptom-free. it was only through our early knowledge of what could develop pathophysiologically that we were in a position to make the right decision at the right time. every doctor treating outpatients can follow this procedure if he is familiar with the pathophysiology of the disease and has the available tests on hand: virus rapid test, additional laboratory parameters (leukopenia, iron), and the humoral inflammation status. the decisive factor, however, is the constant clinical alertness towards the course of every acute feverish cold with acute onset. the patient has to remain in the care of the attending physician, and the chosen treatment has to be administered and monitored. this means constant spo2 measurements and checking the humoral inflammation status every 2 hours. if a clinical worsening occurs during monitoring, the treatment regime has to be changed immediately, which means the administration of an appropriate antibiotic. this outpatient care on the part of the doctor has to be available 7 days a week so that no time will be lost. reports from the netherlands and denmark show that, with the help of this preventive strategy under the motto 'search and destroy,' the dangerous, fatal course of infections reported in germany with at least four deaths a day, can be avoided. however, the doctor has to be adequately remunerated for the elaborate amount of time this intensive outpatient care requires. with our strategy, we have moved from divergence to convergence in the care of our patients. we reported on our 10 years of clinical experience with this approach at the antivirals congress in peking. our main message was early diagnosis and early treatment. we were able to demonstrate this in 287 outpatients with seasonal influenza and 25 h1n1 ⁄ 09 outpatients. our creed is: as much outpatient care as possible and as little hospitalization as possible. virological and autopsy findings in suspected and confirmed fatal cases of 2009 h1n1 pandemic influenza in the czech republic -preliminary results influenza viruses cause substantial morbidity and mortality. pandemic influenza may have a serious impact on certain (mainly younger) age groups in comparison with seasonal flu. influenza is one of few viral infections capable of causing a pneumonia that is difficult to cure and ⁄ or leads to sudden death. the aim of this study was to analyze and compare virological and autopsy findings in patients who died with suspected or confirmed 2009 h1n1 pandemic influenza virus infection. there were 2477 virologically confirmed cases of pandemic influenza and 102 deaths in the czech republic during pandemic wave. more than 400 influenza strains belonging to the new pandemic variant were isolated in the national influenza reference laboratory. postmortem biological samples were collected from any patient who died with suspected influenza infection to test for respiratory viruses. the samples were screened for 2009 h1n1 pandemic influenza virus by real-time pcr (rt pcr), and when rt pcr positive, by virus isolation assay. no immunohistochemical staining for influenza antigen was done on the rna pcr positive cases. other important respiratory viruses such as respiratory syncytial virus, parainfluenza viruses, and adenoviruses were detected by virus isolation assay in a suitable cell culture. epidemiological analysis of postmortem histopathologic findings in the airway tissue was carried out in 57 of 61 fatal cases. virological findings were subsequently correlated with histological changes and available demographic and clinical data. statistical analysis was performed by t-test using spss software. sixty-one deaths (34 males, 27 females) were analyzed. the rna of the 2009 h1n1 pandemic influenza virus was detected by pcr in 38 cases, while 23 cases remained negative. five respiratory syncytial viruses and two adenoviruses were detected in the influenza negative group. the mean age of 38 confirmed 2009 h1n1 pandemic influenza victims was 46ae0 years, age range 7-74 years and median 47ae5 years. the mean age of 23 influenza negative victims was 55ae1 years, age range 1-89 years and median 57ae0 years. the 95% ci for the difference in the age between the two groups is )1ae7; 19ae9. the test is statistically significant at the 10% level. the obtained significance (p = 0ae07) can be explained by the relatively small size of the study group. the most common postmortem histopathologic finding in the lung tissue of the 2009 h1n1 pandemic influenza virus-positive victims was diffuse alveolar damage (often bilateral) and ⁄ or hyaline membrane formation, possibly with signs of respiratory distress syndrome (in 18, i.e., 51ae4%, of 35 autopsied patients). in the 2009 h1n1 pandemic influenza virus negatives, the most common finding was pneumonia or bronchopneumonia with the detection of various bacterial species (in 12, i.e., 54ae5% of 22 autopsied patients). the cause might be either primary bacterial infection or superinfection following primary infection with influenza virus that remained undetected. the 2009 h1n1 pandemic influenza victims were younger than the patients who died with suspected but undetected 2009 h1n1 pandemic influenza. the majority of deaths were primarily linked to rapidly developing respiratory failure. this result supports the previous reports of severe respiratory outcomes in younger age groups that are typically linked to the spread of a pandemic strain of influenza. 1 due to limited amount of pandemic vaccine, especially at the beginning of pandemic, it is advisable to assess experiences with antiviral treatment, mainly dosing, and way of antiviral administration. primers specific for each of the eight genes of pandemic h1n1 ⁄ 2009 were adopted from assays as described previously to discriminate against seasonal human h1n1 and h3n2 viral segments (table 1) . 9 the primers were allowed to cross-react specifically with the sister clade viral segments of pandemic h1n1 ⁄ 2009. 2 the method we employed in this study was a 2-step singleplex sybr green-based real-time rt-pcr. this approach helped lower the running cost of the assays and facilitated downstream molecular analyses (e.g., sequencing) by using screened cdna samples. viral rna was extracted from viral cultures or clinical samples as described 3, 10 and was converted to cdna in a universal rt-pcr. each 10 ll rt reaction containing 5ae5 ll of purified rna, 2 ll of 5· firststrand buffer (invitrogen), 100 u of superscript ii reverse transcriptase (invitrogen), 0ae1 lg of uni12 (5¢-ag-caaaagcagg-3¢), 11 0ae5 mm of deoxynucleoside triphosphates and 10 mm of dithiothreitol was incubated at 42°c for 50 minutes, followed by 72°c for 15 minutes for heat inactivation. for each segment-specific real-time pcr, the 20 ll reaction contained 1 ll of a 10-fold diluted cdna samples, 10 ll of fast sybr green master mix (applied biosystems), and 0ae5 lm of the corresponding primer pair. the thermocycling conditions of all eight segment-specific pcrs were optimized as 95°c for 20 seconds, followed by 30 cycles of 95°c for 3 seconds and 62°c for 30 seconds, and all eight assays were performed simultaneously in a 7500 sequence detection system (applied biosystems). at the end of the amplification step, pcr products went through a melting curve analysis to determine the specificity of the assay (60-95°c; temperature increment: 0ae1°c ⁄ seconds). cdna of a ⁄ california ⁄ 04 ⁄ 2009 virus was used as a positive control. robust and specific amplification was achieved in all eight segment-specific real-time rt-pcr reactions. 9 pcr product for each segment of pandemic h1n1 ⁄ 2009 yielded unique melting curve pattern with distinctive melting temperature (tm), which was not observed in negative and water controls ( figure 1 ). reactions with tm value within 2 sds of the mean tm were determined as positive. we evaluated the assays with a number of serologically confirmed human clinical samples. all pandemic h1n1 ⁄ 2009 samples (n = 67) were positive in all eight assays, while all seasonal samples (h1n1 = 38; h3n2 = 66) were negative in all assays, as expected ( figure 1 and data not shown). these results showed that no reassortant of pandemic h1n1 and seasonal viruses was present in the tested human isolates. we applied these assays to our on-going influenza virus surveillance program in swine. nasal and tracheal swab samples were collected at an abattoir in hong kong and cultured in madin darby canine kidney cells or embryonated eggs as described. 12 positive viral cultures in hemagglutination assays were tested with the established segmentspecific real-time rt-pcr assays. among 59 swine viral isolates collected from 2009 to september 2010, 10 of them were recognized as pandemic h1n1 ⁄ 2009 in all eight segments. they were confirmed to be of pandemic h1n1 ⁄ 2009 origin by subsequent full genome sequencing analyses, showing that there were interspecies transmissions of the virus from humans to pigs. 13, 14 the remaining 49 viruses had one to seven gene segments positive in the segment-specific real-time rt-pcrs. thirty of them were selected as representative samples for full genome sequencing analyses based on the genotyping data generated in our assays. they were swine h1n1 or h1n2 viruses with their gene segments derived from tr or eurasian avian-like swine lineages. it should be highlighted that all of their positive gene segments in our assays belonged to the sister groups of pandemic h1n1 ⁄ 2009. their melting curve patterns were very similar to those derived from segments of pandemic h1n1 ⁄ 2009, except for ha of tr lineage. 9 our results successfully demonstrated the use of these segment-specific real-time rt-pcrs to recognize gene segments of contemporary tr (pb2, pb1, pa, ha, np, and ns) and ea (na and m) swine viruses. 2 the ha-specific assay was able to discriminate pandemic h1n1 ⁄ 2009 from other contemporary swine viruses in the same lineage. nevertheless, to confirm the identity and to examine all the genetic variations in the viruses of interest, full genome sequencing analyses were necessary. in this study, the biggest obstacles in primer design were sequence similarity and diversity of influenza viruses. we attempted to use degenerated primers, but they were highly non-specific. the finalized non-degenerated primers crossreacted with genes from pandemic h1n1 ⁄ 2009 and its sister clade tr (pb2, pb1, pa, ha, np, and ns) and ea (na and m) swine viruses with some minor sequence mismatches. three avian (h5n1, h7n7, and h9n2) and 1 classical swine (h1n1) were also tested with our assays. all of these animal viruses were negative, except for ns gene of the classical swine virus. our segment-specific real-time rt-pcr assays might be used in high throughput genotyping. they detected pandemic h1n1 ⁄ 2009 viruses and acted as a preliminary screen-ing tool to select virus reassortants of interesting genotypes for further sequencing analyses. in fact, we identified a novel reassortant in january 2010 during the course of this study. this sw ⁄ hk ⁄ 201 ⁄ 2010 has a previously unidentified viral gene combination as shown in figure 1 . it was confirmed to be a reassortant between pandemic h1n1 ⁄ 2009 and other swine viruses in full genome sequencing characterization. it has a pandemic h1n1-like n1 gene, an ea-like h1, and the other six internal genes derived from tr swine viruses. 13, 14 the eight established real-time rt-pcrs can rapidly reveal the gene-origins of influenza viruses. we are currently using these assays in influenza surveillance in humans and other animals. it is believed that similar strategy might be applied to detect and genotype other influenza viruses and possible reassortants in the future. pandemic influenza a ⁄ h1n1 ⁄ 2009 infects millions of people around the world. a significant fraction of the world's population may also already have been exposed to the virus and, although asymptomatic, may be at least partially immune to the disease. a precise assessment of the number of people exposed to the influenza a ⁄ h1n1 ⁄ 2009 virus is epidemiologically relevant. however, assays typically used to estimate antibody titers against a particular influenza strain, namely hi and neutralization, require use of the actual virus. this seriously limits broad implementation, particularly in regions where high biosafety facilities are unavailable. we developed an elisa method for the evaluation of presence of specific 2009 h1n1 influenza virus-antibodies in serum samples. mouse anti-histidine tagged antibodies (100 ll; 5 lg ⁄ ml; abd serotec ò , uk) in pbs (ph 7ae2) were dispensed into standard 96-well plates and incubated for 12-16 hour at room temperature. excess antibody was removed by at least two successive alternate washings with pbs-tween 0ae05% and pbs. commercial blocking solution (300 ll, superblock ò t20; pierce ò , usa) was added and incubated for at least 2 hour at room temperature. after successive washing steps with pbs-tween 0ae05%, non-glycosylated histidine-tagged recombinant protein (100 ll; 10 lg ⁄ ml) was added to each well. this protein consisted of the receptor-binding domain of the hemagglutinin of the influenza a ⁄ h1n1 virus. 1,2 after 1 hour incubation, wells were washed for at least two alternating 5 minutes cycles with pbs-tween and pbs. a 1:50 dilution of the serum or plasma sample to be assayed (100 ll) was added to each well and incubated at room temperature for 1 hour. after repeated alternating 5 minutes pbs-tween 0ae05% and pbs washes, anti-human igg antibody solution (100 ll ⁄ well; 1:30 000 dilution in pbs-tween 0ae05%) marked with horse radish peroxidase (pierce ò , usa) was added and incubated for 1 hour at room temperature. after repeated alternate washes with pbs-tween 0ae05% and pbs), substrate solution (100 ll; 1-step ultra tmb-elisa; pierce ò ) was added to each well. after incubation for 15 minutes at room temperature in darkness, the enzymatic reaction was stopped by addition of 1 m h 2 so 4 (50 ll ⁄ well). yellow color produced by the enzymatic reaction was evaluated by absorbance at 450 nm in a biotek ò microplate reader (usa). blank assays using albumin in place of human sera established the elisa background signal, which was subtracted from sample absorbance signals: abs serum sample ¼ abs serum sample before correction à abs albumin sample : absorbance values were normalized based on the average signal of 103 non-exposed subjects (uninfected subjects), and expressed as normalized absorbance (abs norm ): where abs serum ample is the sample absorbance signal, abs albumin sample is the albumin control absorbance signal, abs non exposed subjects is the average absorbance signal of non-exposed subject samples. for ferret serum samples, the same basic protocol was followed, with minor modifications. an anti-igg anti-ferret polyclonal antibody preparation was used at a dilution of 1:30 000 in pbs-tween 0ae05%. 2 a recombinant receptor-binding domain of the ha of the influenza a ⁄ h1n1 ⁄ 2009 virus, expressed in escherichia coli strains, 2 was used as the elisa antigen. this 27 kda protein, designated here as ha 63-286 -rbd, contained amino acids 63-286 of the influenza a ⁄ mexico ⁄ indre4114 ⁄ 2009(h1n1) hemagglutinin. a sequence coding for a series of six histidines at the n-terminus of the protein was included in the genetic construct to allow purification using immobilized metal affinity chromatography (imac) and attachment to assay surfaces treated with anti-histidine antibodies (or alternatively co +2 or ni +2 ). a panel of four samples (kindly provided by st. jude from ferrets exposed to different influenza strains, namely h3n2, h1n1 swine, and h5n1, was also tested by the elisa method using 1:50 dilutions. protein ha 63-286 -rbd specifically and selectively recognizes antibodies from serum samples from convalescent h1n1 ⁄ 2009 influenza subjects. dubois et al. 4 demonstrated that this protein, produced in e. coli, folds properly into a 3-d structure practically indistinguishable from the analogous region in the ha of the influenza a ⁄ h1n1 ⁄ 2009 virus. ha 63-286 -rbd preserves three of the conformational immunogenic epitopes (sa, sb, and cb) described for influenza a ⁄ h1n1 hemagglutinins. 5 the recombinant protein was used as the antigen, attached through histidine tags to microplate surfaces treated with anti-histidine antibodies to discriminate between serum samples from subjects exposed and non-exposed to influenza a ⁄ h1n1 ⁄ 2009. samples collected before the pandemic onset, and therefore presumed to exhibit low specific antibody titers against influenza a ⁄ h1n1 ⁄ 2009, were analyzed by elisa using the antigen ha 63-286 -rbd. the histogram of normalized absorbance values from this sample set displayed a normal behavior with a standard deviation of 0ae57 units. only 12ae5, 9ae7, and 3ae88% of these samples exhibited normalized absorbance values higher than 1ae5, 1ae75, and 2ae0, respectively. no sample from non-exposed individuals presented an absorbance value higher than 2ae5. variability among samples from non-exposed subjects was much lower than in samples with high specific serum antibody titers from convalescent h1n1 ⁄ 2009 patients. exposure to the h1n1 ⁄ 2009 influenza virus with this elisa method can be predicted by absorbance values normalized to those of abs norm ¼ ðabs serum ample à abs albumin sample þ=ðabs non exposed subjects à abs albumin sample þ ð1þ serum from uninfected subjects. consequently, for reliable results, inclusion of samples from non-exposed subjects on every assay microplate is necessary. figure 1 shows the analysis of 20 human serum samples, including 17 samples from convalescent patients with positive diagnosis by rt-pcr. three positive (dark gray bars) and two negative controls (light gray bars) were included in the same microplate. all serum samples corresponding to convalescent subjects exhibited absorbance values 1ae75-4ae5 times higher than negative samples ( figure 1 ). normalized absorbance values above 1ae5 suggested exposure to the virus, although, a more conservative threshold value of 1ae75 units is proposed for discrimination between exposed and non-exposed subjects. the elisa method described here yields adequate reproducibility and a high signal ⁄ noise ratio within determinations in the same microplate and among different microplates. using a normalized absorbance value of 1ae75, the method was able to discriminate samples from convalescent patients, preferably after the third week of infection, and at least up to the twentyfourth week of exposure. assay sensibility was further validated against results from hi assays. a previously reported study showed that all members in a pool of fourteen samples diagnosed as positive by hi exhibited normalized absorbance values higher than 1ae5, and 85% of them exhibited normalized absorbance values higher than 2ae0. 1 in general, high hi titers (>320) were correlated with normalized absorbance values higher than 4ae0. figure 2a shows results using the ha-rbd elisa method and the hi assay on a pool of seventeen known positive serum samples corresponding to convalescent h1n1 ⁄ 2009 patients. all samples determined as positive by hi (10 samples) were also positive by elisa. while sensitivity of the hi assay was 10 ⁄ 17 = 58ae88%, the elisa method recognized all samples correctly as positive (100% sensitivity) when a threshold of 1ae5 or 1ae75 was used. figure 2b shows that sera from ferrets infected with other influenza strains (h3n2, h1n1 swine, and h5n1) showed no cross-reactivity when analyzed by elisa. in summary, the ha-rbd elisa method presented here consistently distinguished influenza a ⁄ h1n1 ⁄ 2009 infected and non-infected individuals, particularly after the third week of infection ⁄ exposure. since no actual viral particles are required, this assay can be readily implemented in any basic laboratory. in addition, should sufficient vaccine be unavailable, this elisa could determine the level of specific antibodies against the virus and presumably the extent of partial protection in a subject. therefore, the elisa protocol might allow better administration of vaccination programs during pandemic or seasonal influenza outbreaks. in april 2009, a novel h1n1 influenza virus emerged in north america and caused the first influenza pandemic of the 21st century. [1] [2] [3] [4] the 2009 pandemic h1n1 (pdmh1n1) has a unique gene constellation that was not previously identified in any species or elsewhere. it is genetically related to the triple reassortant swine h1n1 influenza viruses currently circulating in north america, with the exception of the neuraminidase (na) and matrix (m) genes, which are derived from a eurasian swine influenza virus. swine h1n1 influenza viruses were first isolated in 1930 and continued to circulate in north america with very little antigenic changes (classical swine h1n1) until 1998. since 1999, however, the antigenic make up of swine h1 viruses has shown increased diversity due to multiple reassortment events and the introduction of h1n1 genes from human influenza viruses. currently, four swine h1 clusters (a, b, c, d) are found endemic in the north american swine population. 5, 6 these swine h1 viruses show substantial antigenic drift compared to the classical swine h1 viruses. cluster d swine h1 is derived from current human h1 viruses, and there is a substantial antigenic divergence between classical swine h1 and human seasonal h1 viruses. epidemiological evidence shows a two-way transmission of influenza viruses between swine and humans, and such events lead to the emergence of the pdmh1n1 virus. 5, 7, 8 phylogenetic analysis have suggested that possible ancestors of the eight genes of pdmh1n1 were circulating in the swine population for at least 10 years prior to the emergence of the pdmh1n1 virus in humans, although the pdmh1n1 virus itself was not isolated from pigs until after the pandemic. interestingly, pdmh1n1 infections have been reported not only in humans and pigs, but also in other animal species such as turkeys, cats, ferrets, cheetahs, and dogs. [9] [10] [11] after the first report of pdmh1n1 infection in swine in canada, other countries, including argentina, australia, singapore, northern ireland, finland, iceland, england, united states, japan, and china reported outbreaks of pdmh1n1 in swine as well. 9, [12] [13] [14] the ample geographic range of pdmh1n1 outbreaks in swine, its apparent broad host range, and the possibility of two-way transmission between swine and humans poses a tremendous challenge for controlling the virus. therefore, to differentiate pdmh1n1 from other h1 strains, particularly in swine and human populations, is an important issue to ascertain the magnitude of the disease caused by the pdmh1n1. in this study, we developed an elisa assay to discriminate pdmh1n1 strains from other swine and human h1 viruses. madin-darby canine kidney (mdck) cells (atcc, manassas, va, usa) were maintained in modified eagle's medium (mem) containing 5% fbs. a ⁄ california ⁄ 04 ⁄ 09 ⁄ h1n1 virus (ca ⁄ 04) was kindly provided by the centers for disease control and prevention (cdc), atlanta, georgia. other viruses are listed in table 1 . viruses were propagated in mdck cells and stored at )70°c until use. viruses were titrated by the reed and muench method to determine the median tissue culture infectious dose (tcid 50 ). 15 three monoclonal antibodies (3b2, 5h7, and 12f3) against ha of 2009 pandemic h1n1 were prepared in our laboratory following previously described methods (shao and perez et al., unpublished). purification and labeling of mabs mab 3b2, 5h7 and 12f3 were purified on a protein g-sepharose affinity column (upstate biotechnology, lake placid, ny, usa). biotinylation of the detection antibody in the elisa was performed using sulfo-nhs-lc-biotin (sulfosuccinimidyl-6-(biotinamido)hexanoate; pierce, rockford, il, usa) according to the manufacturer's instructions. purified 5h7 and 12f3 were selected as the capture antibody, and biotin-conjugated 3b2 was selected as the detection antibody, and hrp-conjugated streptavidin (abcom, cambridge, ma, usa) was developed using the tmb substrate system (kpl, gaithersburg, md, usa). in brief, the mixture of the purified 5h7 and 12f3 (2ae0 and 2ae2 lg ⁄ ml respectively, in carbonate ⁄ bicarbonate buffer, ph 9ae6) was coated to 96-well plates (test well, t) for 12 h at 4°c. at the same time, a control antibody was coated to 96-well plates (control well, c). after blocking the plates with 5% (w ⁄ v) non-fat milk in pbs for 1 hour at 37°c, the samples were diluted in extract buffer (1%tween-20, 0ae5%bsa in pbs) and added to the wells (100 ll ⁄ well, each sample was table 1 . specificity assay of the sandwich elisa result (t ⁄ c) added to four wells-two for t wells and two for c wellsand the mixture was incubated at 37°c for 1 hour. after four washes, 100 ll biotin-conjugated 3b2 (0ae25 lg ⁄ ml) in dilution buffer (0ae5% bsa in pbs) was added to the wells and the mixture was incubated for 1 h at 37°c. following three washes, 100 ll diluted hrp-conjugated streptavidin (62ae5 ng ⁄ ml) in dilution buffer was added to the plates. after incubation for 1 h at 37°c, the plates were washed five times, and the binding developed using the tmb substrate system for 30 minutes. the ratio of the average od 650 value of the t wells to that of the c wells (t ⁄ c) of individual samples was calculated. t ⁄ c values >1ae5 were considered positive in the sandwich elisa. we developed three monoclonal antibodies, 3b2, 5h7, and 12f3, against a prototypical pdmh1n1 strain, a ⁄ california ⁄ 04 ⁄ 2009 (h1n1) (ca ⁄ 04). these monoclonals were used to develop a rapid sandwich elisa for specific diagnosis of pdmh1n1 strains. purified 5h7 and 12f3 were used as capture antibodies, whereas the biotin-conjugated 3b2 was used as detection antibody. the sandwich elisa showed strong reaction with different pdmh1n1 strains as described in in order to evaluate if the sandwich elisa could distinguish the pdmh1n1 from other swine h1 clusters (a, b, c, d), 14 swine influenza strains spanning these clusters were tested. these viruses were first diluted 1:10 in extract buffer, and then added to the coated plates. as shown in table 1 , the t ⁄ c ratios of these viruses were <1ae5, and therefore showed negative elisa result. likewise, testing of human seasonal virus strains a ⁄ brisbane ⁄ 59 ⁄ 2007 (h1n1), a ⁄ malaya ⁄ 302 ⁄ 1954(h1n1), a ⁄ wsn ⁄ 1933 (h1n1), and a ⁄ brisbane ⁄ 10 ⁄ 2007 (h3n2) also showed negative elisa results. furthermore, the sandwich elisa showed no cross reaction with avian influenza viruses, including strains of the h2, h3, h5, h6, h7, h8, h9, h10, h11, h12, and h13 subtypes. more recently, the mutation d222g in the ha of some pdmh1n1 strains has been associated with exacerbated disease and altered receptor binding. [16] [17] [18] [19] [20] to evaluate if such mutant could be detected in our sandwich elisa, we tested a mutant of a ⁄ netherland ⁄ 602 ⁄ 2009 (h1n1) carrying the d222g mutation (engineered by reverse genetics). as described in table 1 , our elisa could still capture the d222g mutant virus and showed a positive reaction, which highlights the specificity of our assay for pdmh1n1 strains, even those with mutations. to evaluate the sensitivity of the elisa, we used the serially diluted pdmh1n1 viruses to determine the limit of detection (lod). as shown in table 2 , in our elisa the highest positive dilutions of nl ⁄ 602 and ca ⁄ 04 were 1:320 and 1:160, respectively. the lod of the sandwich elisa by tcid 50 was 3ae2 · 10 3 and 1ae5 · 10 4 tcid 50 ⁄ ml, for nl ⁄ 602 and ca ⁄ 04, respectively. it is important to note that the t ⁄ c ratio from nl ⁄ 602 and ca ⁄ 04 viruses showed clearly a dose dependent effect, while the t ⁄ c ratio of a ⁄ swine ⁄ iowa ⁄ 30 (h1n1) did not show the same dependence and was always <1ae5, corroborating the high specificity of the sandwich elisa for pdmh1n1 strains. although we did not compare our elisa with other current commercial rapid influenza detection kits, the lod of our elisa assay is similar to other commercial kits that detect human seasonal influenza virus. 21 comparison of the sandwich elisa with the ''gold standard'' -virus isolation in order to further evaluate the feasibility of the application of the elisa to clinical samples, 70 nasal wash samples 1ae58 · 10^6 )(0ae8) )(0ae8) )(0ae9) )(1ae0) )(0ae9) )(0ae8) )(1ae1) )(0ae9) -from ferrets, 56 of those previously infected with ca ⁄ 04 and shown positive by virus isolation, were tested. the samples were diluted 1:1 in extract buffer and then tested using the sandwich elisa. result showed 51 out of 56 positive samples by virus isolation were positive also by the sandwich elisa (sensitivity 90ae1%). the 14 samples tested that were negative by virus isolation were also negative in the elisa, indicating 100% specificity for our assay. these results show not only that our elisa has high compatibility with the virus culture method, but also indicates this application can be used for clinical samples. although real time rt-pcr targeting the ha gene has been used for specific diagnosis of pdmh1n1 with high sensitivity, [22] [23] [24] [25] [26] [27] it is a method that requires manipulation of the sample to extract viral rna, and it is prone to crosscontamination during the pcr steps. in this study, we described a convenient sandwich elisa based on three mabs developed against the pdmh1n1 strain. the elisa not only shows high specificity for pdmh1n1 strain, but also shows great sensitivity. the elisa could distinguish pdmh1n1 strains from human seasonal h1 and h3 viruses and, more importantly, from other swine h1 viruses. we must note that current rapid diagnostic tests cannot be used to differentiate pdmh1n1 from swine or human h1 viruses. 28 it is also worth noting that the sensitivity of commercial rapid antigen-based diagnostic tests for detecting pdmh1n1 is lower than that for human seasonal influenza viruses. 28, 29 a study by kok et al. 32 showed that sensitivity of the current rapid antigenic tests for pdmh1n1 is only 53ae4%, whereas that for seasonal influenza a is 74ae2%. chen et al. 33 developed a dot-elisa and increased the sensitivity for influenza rapid antigen detection. however, the dot-elisa developed by chen cannot distinguish among subtypes. the lod of our elisa is between 3ae2 · 10 3 to 1ae5 · 10 4 tcid50 ⁄ ml, comparable to the lod of rapid diagnostic tests for human seasonal influenza viruses. 21 compared to the ''gold standard''-virus isolation-our sandwich elisa showed 90ae1% sensitivity using ferret nasal washes. our results highlight the potential application of our sandwich elisa for the specific diagnosis of pdmh1n1 viruses. 17 the timely and reliable laboratory evidences are vital factors for field epidemiologists trying to control outbreaks of infectious diseases and for the practicing clinicians to properly manage disease cases. therefore, analysis of new detection methods in comparison to the routine ''classical'' methods is essential to select new methods to be introduced into health service practices, especially in developing countries. in this study we have compared rt-rt-pcr detection of influenza viruses and direct fluorescent-antibody assay using r-mix hybrid cells (a549&mv1lu) with the ''classical'' cell culture methods in developing country settings. in this study, we analyzed 503 nasopharyngeal swabs col the detection of influenza h1, h3, b, and pandemic influenza (h1)pdm virus-specific nucleic acids was performed by rt-rt-pcr in abi7500 fast real time pcr system using primers recommended by cdc, usa, and super-scriptô iii one-step rt-pcr and platinum ò taq dna polymerase kits (invitrogen). the cycling protocol was: 30 minutes at 50°c, 2 minutes at 95°c, and 45 cycles of 15 seconds at 95°c, 30 seconds at 55°c. 1 rapid detection of influenza infected cells has been performed by dfa using the infected hybrid cells of r-mix within 48 hours after inoculation, according to the manufacturers instruction (diagnostic hybrids, inc., usa). 2 the isolation of influenza viruses was performed on mdck cell culture by the protocol recommended by cdc, usa. 3 we detected 1891(25ae2%) influenza virus-specific nucleic acid fragments from all tested samples by rt-rt-pcr. among the positive samples, there were 25ae6% a(h1n1), 0ae4% a(h3n2), 5ae6% influenza b, and 19ae5% a(h1n1)pdm with different distributions by time series in different age-groups. inoculation of the cell lines by rt-rt-pcr positive samples selected randomly has detected influenza virus in 52ae3% (521 ⁄ 995) on mdck cell culture and 51% (77 ⁄ 151) on r-mix hybrid cell culture with varying distribution for different strains. in other words, mdck cell culture technique was better for isolation for pandemic influenza viruses and dfa using r-mix hybrid cell culture technique for detection of seasonal influenza viruses (table 1) . average times needed for the final results for different methods were: 4 hours for rt-rt-pcr, 48 hours for dfa on r-mix and 10 days for mdck cell culture with two passages at least. the peak of the seasonal influenza a virus detection occurred in the 7-8th weeks of 2009, however the pandemic influenza detection peak was observed in the 43-44th weeks of 2009 ( figure 1 ). the outbreaks by seasonal influenza viruses was observed mostly among the children of 0-15 years of age, and pandemic influenza virus outbreak was observed mostly in the adults of 16-64 years of age. the results of this study indicate that rt-rt-pcr is the most suitable method for decision makers in epidemiological and clinical settings by sensitivity and timeliness. the final results show that r-mix dfa requires 12 times longer, and by mdck cell culturing, 60 times longer periods, than by rtrt-pcr. mdck cell culture technique has a higher isolation of pandemic influenza viruses, and r-mix dfa has a greater detection rate of seasonal influenza viruses by our results. according to our study, with rtrt-pcr, the isolation of positive samples by tissue culture of influenza a viruses was 24% and influenza b viruses was 66ae5%, which is lower than in similar spanish study. 4 however our study illustrates similar results with a canadian study 5 where the sensitivity of dfa method and tissue culture technique was shown to be lower than rtrt-pcr sensitivity. as recorded by a study of american researchers, 6 r-mix hybrid and conventional cell culture techniques have had similar sensitivity, which does not match the results of our study. however, the results of our study match with the results of italian and american scientists 7, 8 where the r-mix hybrid method for seasonal influenza viruses is higher than mdck cell culture technique. background: viral kinetics is increasingly used to study influenza infectiousness. the choice of the study design, i.e. when and how many times nasal samples are to be collected in individuals depending on the sample size, is crucial to efficiently estimate the viral kinetics (vk) parameters. material and methods: we performed a model based optimal design analysis in order to determine the minimal number of nasal samples needed to be collected per subject and when to collect them in order to correctly estimate the vk parameters. the model used was a non linear mixed effect model developed with data collected from 44 patients sampled nine times in 9 days (initial design -504 samples collected), and we used d-optimization for design identification. we also computed the minimal number of participants necessary. results: considering that 25% of the influenza-like illness cases are not due to volunteer challenge studies have been used since the 60's to provide data on virus shedding from the respiratory tract during influenza infection. 1 recently, vk was studied in naturally acquired influenza infection. 2, 3 these data are invaluable to describe the natural history of influenza-infection and to compute natural history parameters such as the latent period, generation time, or the duration of infectiousness. [1] [2] [3] [4] however, among the 56 studies used in a meta-analysis about viral shedding kinetics, 1 the designs varied greatly from one to another. these differences led to variable amount of available information concerning the vk. the lack of adequate sampling leads to imprecise estimates. on the other hand, intensive sampling or over-sampling, while associated with highly informative data, may lead to unnecessary discomfort for the patient and cost to the investigator. optimal design is increasingly used to conceive studies 5 and provides cost-efficient designs. here we propose an optimised design to model vk in the case of influenza infection. we defined the number of participants, the number of samples to collect and their allocations. this design allows, at a minimum cost and discomfort, accurate vk curves and allows the natural history parameters to be well described. model a vk population model was proposed for influenza infection. this model describes with eight parameters the relations between free virus, uninfected target epithelial cells, infected epithelial cells, and early immune response. this model was built on a dataset of 44 volunteers from which nasal samples were collected once a day over 9 days. we call this dataset the ''original dataset''. three parameters, the induction of the early immune response, the virus production rate, and the virus clearance, did not show inter-individual variability and were precisely estimated (relative standard error below 10%). we considered them as fixed in this research work. five parameters were hence considered here: b the infection rate, d the infected cell mortality rate, w the effect of early immune response on virus production rate and v init the initial value of virus titre. in order to correctly estimate these parameters it is crucial to determine a design to collect informative data. optimal designs maximise the amount of information provided by the study. it involves the determination of the number and allocation of sample times per subject as well as the number of participants. 6 d-optimization is based on the maximization of the determinant of the fisher information matrix and thus minimizes the variance of the parameters. 7 we used the fedorov-wynn algorithm implemented in pfim 3.0 to maximize this determinant, 8 which implies to pre-define a set of possible sample times. with the hypothesis that the inoculation occurred at 8:00 am, we chose three possible hours (8:00, 14:00, and 20:00) for each day with respect of the sleep-time. 8 to validate the design, we simulated 100 datasets of 44 volunteers with the optimised design obtained. we then estimated the population parameters using monolix 3.1 for each of the datasets. we compared the estimated parameters obtained with the simulated datasets to the parameters used to build the optimal design. we computed the relative bias as: with n: number of successful estimations among the 100 simulated datasets. h i : parameter value obtained with the ith dataset. h: parameter value obtained with the original dataset. we also compared the observed rse from these simulations with the rse predicted by pfim and the rse obtained with the original dataset. the rse is proportional to ffiffiffi n p , where n is the population size. we can hence deduce the smallest number of participants necessary to obtain rse below 50%. where rse predicted is the highest predicted rse (here rse for w) with 44 participants and n predicted = 44 and rse min is equal to 0ae5. considering that 25% of the influenza-like illness cases are not due to influenza virus, the total number of participants should be multiplied by 1ae25. we found that the best design was when all the participants are sampled five times: three times during the second day post-inoculation at 8:00, 14:00, and 20:00 hours and twice on the third day post-inoculation at 8:00 and 20:00 ( figure 1 ). the comparison of the relative bias and rse predicted by pfim and those obtained after simulation and re-estimation of the parameters are shown in figure 2 . v init and d in a lesser extent present bias. fixed effect parameters are precisely estimated and accordingly to pfim except for v init . we found that 16 participants shedding virus or 20 participants with ili symptoms are necessary if 25% of them are not infected with influenza virus. we propose an optimised design to accurately study the vk of influenza virus with the minimal number of samples. this design is well balanced between the amount of necessary information and the precision of estimation. we found that 100 samples are necessary to precisely fit the vk curves, which is five times less than the number of samples collected in the original study. 4??? the samples should be collected during the second and third days after inoculation. yet we showed in a previous work that the incubation period lasted 1ae9 days. 4??? hence, the optimised sample times correspond to the two-first days of symptoms and this design could be applied to naturally acquired infections studies in which the inoculation time is unknown. an advantage of this design is its practicality and convenience. all samples are collected during the daytime and after the onset of symptoms. it can thus be used for studies with naturally acquired infections. the design was validated with several criteria concerning the accuracy of the estimation with the optimised design. the parameters estimates were generally satisfactory. the parameter describing the effect of the early immune response on the virus production rate was, however, less precisely estimated (predicted rse = 62%), and the initial value of the viral titre was very different of the one obtained with the original dataset (bias v init on figure 2 ). this is probably due to the fact that it was measured at day 2 post inoculation, and that the inter-individual variability is much higher than at day 0. furthermore, d (the infected cell mortality rate) seems also to be biased. this may be due to the fact that three parameters were fixed. the model used was developed from experimentally inoculated healthy volunteers with low serum haemagglutinin antibody titre and with virus inoculation time at 8:00 am. the applicability of the design to naturally acquired infection would depend on the pathogenicity of the virus as well as pre-existing immunity and the relevance of challenge method to natural influenza acquisition. 1 our design could be directly used to accurately study vk during influenza infections and would reduce the discomfort of patients and the cost of the experimentation. usefulness of a self-blown nasal discharge specimen for use with immunochromatography based influenza rapid antigen test introduction influenza rapid antigen tests (irat) have become very popular and are widely used for confirming suspected clinical diagnosis of influenza in japan. 1 most of the currently used irat that are based on immunochromatography (ic), nasopharyngeal swab, nasopharyngeal aspiration, and throat swab have been approved as specimens for japanese national health insurance purposes. but the specimen collection by these methods gives patients considerable discomfort, and sometimes appropriate specimens cannot be obtained due to patient resistance, especially by children. in the present studies, self-blown nasal discharge was used as the specimen for an irat, and the results were compared with the results of viral isolation and an identical kit primed with nasopharyngeal swab specimens for seasonal influenza viruses and pandemic (h1n1)2009 virus. patients who visited any of the 10 clinics that belong to the influenza study group of the japan physicians association in the 2006-2007 and the 2009-2010 influenza seasons with influenza-like illnesses exhibiting findings were registered after providing informed consent. a square plastic sheet of 20 · 20 cm was handed to the patient. nasal discharge was collected by blowing the nose into the plastic sheet as a specimen for irat, i.e. self-blown specimen. two nasopharyngeal swab specimens were also obtained at the same time for irat and virus isolation. self-blown specimens were obtained successfully by 152 (82ae2%) of 185 consecutive outpatients in the 2006-2007 season, as seen in table 1 the sensitivity and specificity of various influenza rapid antigen tests have been reported in various settings. [1] [2] [3] [4] direct comparison of the results is difficult because of differences in patient or influenza virus, characteristics such as age, study designs, and other features. in this study of the 2006-2007 influenza season, the sensitivity, specificity, and accuracy of the ic kit primed with nasopharyngeal swab specimens were 87ae0%, 94ae0%, and 90ae8%, respectively. these results were quite comparable to our results of the 2005-2006 season, 3 in which the overall results of other ic kits were 82ae9%, 81ae0%, and 82ae3%, respectively, indicating that the ic kit used is quite reliable. the sensitivity, specificity, and accuracy of an ic kit will vary by the method of specimen collection. in general, virus titer is considered to be highest with nasopharyngeal aspiration, lower with nasopharyngeal swabs, and lowest with throat swabs. practically, nasopharyngeal swab is the most popular. the sensitivity, specificity, and accuracy of the ic kit with self-blown discharge specimens compared well with those of an identical ic kit primed with nasopharyngeal swab specimens. for self-blown specimens, sensitivity and specificity were 87ae0% and 94ae6% for influenza a, 85ae7% and 99ae1% for influenza b, 88% and 88ae9% for pandemic (h1n1) 2009. self-blown specimens display sensitivity, specificity, and accuracy comparable to that of conventional nasopharyngeal swab specimens. there was no significant difference in sensitivity, specificity, or accuracy between self-blown specimens and nasopharyngeal swab for influenza a, influenza b, and pandemic (h1n1) 2009. these results suggest that selfblown specimens are as useful as nasal cavity swab specimens for the diagnosis of influenza in the clinical settings. nasal discharge, obviously, cannot be collected from infants incapable of blowing their own nose or patients who do not develop a nasal discharge. in this study, self-blown specimens were obtained from 82ae2% of the patients. the rate of successful collection was over 70% in the age groups of 0-6 and 7-15 years. these rates would seem to be sufficient for clinical use. the procedure of self-blown specimen collection using a plastic sheet is easy and causes no pain or discomfort. it seems to be more acceptable and safe than the other methods, especially for children. furthermore, this procedure reduces the risk of influenza transmission from patients to the medical staff members involved in sample collection. self-blown sample collection may be superior to other sample collection methods in these respects. we previously reported an inverse correlation between the amount of virus in a specimen and the time to a positive reaction. 4 in this study, there was no significant difference in the mean time to a positive between self-blown self-blown specimens enough to be examined were obtained 152 from 185 consecutive outpatients, and specimens showed a tendency to be obtained large amount from children rather than the aged. there were no statistically significant differences between the ic kit results primed with self-blown discharge and nasopharyngeal swab specimens for influenza a, influenza b and pandemic (h1n1) 2009. and nasal swab specimens, suggesting that the self-blown specimens contained sufficient viral antigen for the ic kits. the influence of the presence or absence of nasal congestion on the results of the kit was assessed. the sensitivity of selfblown specimens from patients with nasal congestion was significantly lower than that from patients without nasal congestion. it is possible that insufficient capability to blow the nose due to nasal congestion might tend to lead to false negatives. the observation that the time to positive is longer for patients with nasal congestion than for patients without nasal congestion is concordant. application of self-blown specimen collection only to appropriate patients would increase the sensitivity, which would be important in a clinical setting. we tested only two commercial antigen detection kit, the quick vue rapid sp influ kit and quicknaviô-flu (denka-seiken co., ltd). the resulting sensitivity, specificity, and accuracy of the ic kit primed with self-blown specimens were considered adequate for clinical use. to confirm the usefulness of self-blown nasal discharge specimens, further investigation is necessary using other kits and in different settings. the usefulness of a self-blown nasal discharge specimen for an influenza rapid antigen test based on immunochroma-tography was evaluated in the 2006-2007 and 2009-2010 influenza season. results suggest that self-blown nasal discharge specimens are useful as specimens for influenza rapid antigen tests based on immunochromatography for not only seasonal influenza viruses, but also pandemic (h1n1) 2009 virus. the specimen collection by the patients themselves will reduce the burden of other collection methods and the risk of infection to the medical staff. in april 2009, a mixed-origin h1n1 influenza virus was recognized as a new causative agent of influenza-like illnesses (ili) in humans. since its emergence, the virus has spread rapidly throughout the world and caused a pandemic. most commercial rapid antigen tests (rat) can detect influenza a or b viruses, but cannot specifically distinguish pandemic (h1n1) 2009 virus with seasonal influenza. recent studies have indicated that the poor performance of the rat approach and nonspecific detec-tion of the pandemic (h1n1) 2009 virus was the main obstacle to their widespread use in private clinics. 1,2 with the need for a new rapid kit with reasonable sensitivity and specificity for pandemic (h1n1) 2009 virus, we developed a new rat kit in collaboration with company, standard diagnostics, inc., (yongin-si, gyonggi, korea). monoclonal antibody (mab) against haemagglutinin (ha) of the pandemic (h1n1) 2009 virus was developed using korean isolate and applied to the new kit with the mab to seasonal influenza virus. we examined the detection limit of the kit using the serial dilution of korean pandemic virus isolate (a ⁄ korea ⁄ 01 ⁄ 2009). during december 2009, 432 clinical specimens from patients with ili were collected at 11 sentinel clinics of six provinces in korea. the specimens were tested by the new rat, and the results were compared with those of real-time reverse transcription polymerase chain reaction (rrt-pcr) by us cdc and virus isolation in mdck cell culture to determine the sensitivity and specificity for the diagnosis of pandemic (h1n1) 2009. the detection limit of the new kit against ha of a ⁄ korea ⁄ 01 ⁄ 2009 virus was confirmed to be 10 4 pfu ⁄ ml. by contrast, the detection limit against the np protein was 10 3 pfu. however, when the kit was applied to clinical specimens, no difference between the two targets was found. using rrt-pcr and viral culture as the references, the performance of the ridt is shown in table 1 . among 432 specimens, 178 were tested positive by rrt-pcr and 186 were tested positive by viral culture. among the 178 rrt-pcr confirmed cases, 122 were positive, and among the 186 viral culture confirmed cases, 120 were positive with the new rat. using rrt-pcr as the reference standard, the overall sensitivity of rat was 68ae5% (95% confidence interval (ci): 61ae7-75ae3%) and specificity was 98ae4% (ci: 96ae9-99ae9%). with viral culture as the reference, the rat sensitivity and specificity was 64ae5% (ci: 57ae6-71ae4%) and 97ae6% (ci: 95ae7-99ae5%), respectively. when analyzed by the regions tested, the sensitivity ranged between 57ae1% and 95ae5% for rrt-pcr and between 53ae3% and 87ae0% for viral culture as a reference. among 340 patients who had a record of their symptom onset and sample collection date, 86 (25ae3%) visited the clinic on the day of symptom onset, and 158 (46ae5%) visited 1 day later. when the rat performance was evaluated by day of onset, the sensitivity was lower at three or more days after the onset of symptoms; however, the sensitivity was highest at 2 days after onset and reasonable on the day of onset or at 1 day after ( table 2 ). we found that this new rat had reasonable sensitivity and high specificity compared with rrt-pcr and viral culture for detecting the pandemic (h1n1) 2009 virus. in one recent study, the sensitivity and specificity of the new rat kit was 77% and 100%, respectively, and the ha protein for pandemic (h1n1) 2009 was detected more sensitively than the np protein for influenza a virus. 3 the sensitivity and specificity of our new rat were lower than those of that study. we found that the test performance varied depending on the clinics in which the tests were performed, and this might be attributable to the persons who collected the specimens. although the clinicians were trained well for *ci, confidence interval. **ppv, positive predictive value. ***npv, negative predictive value. collecting specimens, there might be some differences in performance. the new rat kit could detect pandemic (h1n1) 2009 virus specifically. although the sensitivity was lower than those of rrt-pcr and virus culture, and negative rat results should be confirmed with more sensitive methods, this kit could be useful in sentinel clinics if used with caution. determination of infectious virus titres is central to many experiments designed to study the biology of influenza virus. assays based on the measurements of viral components, whether viral protein or nucleic acid, does not differentiate infectious virus from non-infectious or defective viral particles, which may have no infectivity or biological *three hundred and forty samples with a known date of onset and sample collection were analyzed. ª 2011 blackwell publishing ltd, influenza and other respiratory viruses, 5 (suppl. 1), 132-158 activity. therefore the ''gold standard'' of virus measurement requires bioassays that examine the ability of viral particles to replicate and further infect other cells. titration on madin-darby canine kidney (mdck) cells in a 96 well plate format is commonly used to measure influenza virus titre. this method is labour intensive, subjective in their read out of cytopathic effect, and takes several days to obtain a result. microneutralization tests that quantitate neutralizing antibody titres and assays of drugs for antiviral activity also require 96 well based assays of residual virus infectivity. therefore, technologies that improve on the titration of infectious virus will be of great benefit. this study utilized the xcelligence system (roche applied science), which adopts microelectronic biosensor technology to monitor dynamic, real-time label free and non-invasive analysis of cellular events. 1 the system measures electronic impedance using an array of microelectrodes located at the bottom of each culture well (e-plate 96). adherent cells are attached to the sensor surface of electrode arrays, and changes in impedance can be detected and recorded. the xcelligence system can monitor cell events induced by viral infection, such as changes in cell number, adhesion, viability, morphology, and motility. measured electrode impedance is expressed as dimensionless cell index and is graphically represented using software to show the phenotypic changes of a cell population over time. the aim of this study is to demonstrate that using this platform to measure real-time cell index has potential to circumvent many of the limitations of the currently established procedures of end point titration of virus infectivity and for microneutralization assays. madin-darby canine kidney cells were propagated in growth medium consisting of minimum eagle's medium (invitrogen) supplemented with 10% fetal bovine serum (invitrogen), 0ae6 mg ⁄ l penicillin (invitrogen), and 60 mg ⁄ l streptomycin (invitrogen), with incubation at 37°c in a 5% co 2 humidified atmosphere. influenza a ⁄ hong kong ⁄ 54 ⁄ 1998 (h1n1), a seasonal influenza virus from a patient who suffered from a mild febrile illness, was propagated in mdck cells maintained in virus medium consisting of minimum eagle's medium (invitrogen) supplemented with 0ae6 mg ⁄ l penicillin (invitrogen), 60 mg ⁄ l streptomycin (invitrogen), and 2 mg ⁄ l np-tosyl-l-phenylalaninechloromethyl ketone-treated trypsin (sigma, st louis, mo, usa), with incubation at 37°c in a 5% co 2 humidified atmosphere. virus stocks were aliquoted and stored at 70°c until use, and the 50% tissue culture infectious dose (tcid 50 ) of the virus stock was determined by titration in mdck cells according to standard procedures, 2 and the tcid 50 of the stock virus was calculated by the method of reed and muench. 3 to perform a microneutralization assay, mdck cells seeded at a density of 1000 cells ⁄ well in an e-plate 96 was removed from the xcelligence system after approximately 48 hour; growth medium was then removed, cells washed, and replaced with 100 ll virus-medium. a human serum, which is known to contain high titre antibody against the h1n1 virus was heat inactivated for 30 min at 56°c, and twofold serial dilutions were performed in virus medium. the diluted serum was mixed with an equal volume of virus medium containing influenza virus at 400 tcid 50 ⁄ 100 ll. after incubation for 2 h at 37°c in a 5% co 2 humidified atmosphere, 50 ll of virus-antibody mixture was added to the mdck cells to give each well an equivalent virus dose of 100 tcid 50 . a back titration of the virus challenge dose was performed, and a cell control (free of virus) was performed in quadruplicates. after incubation at room temperature for 30 minutes, the e-plate 96 was then placed back onto the xcelligence system in the incubator and maintain at 37°c with 5% co 2 , and the cell index values were measured every 15 minutes for at least a further 72 hour. the same procedures were performed with cells seeded in conventional 96 well cell culture plates for parallel comparison with the currently used standard method. in this case, cells were examined for cytopathic effect under an inverted microscope after 3 days of infection and the lowest virus dilution, which protected the cells from viral induced cytopathic effect taken as the neutralizing end point. after 48 hour of seeding mdck cells at 1000 cells ⁄ well, standard microneutralization assay for influenza virus was performed. integral to this assay, a serial titration of the input virus at 0ae5 log 10 increments was carried out. wells infected with the undiluted virus (100 tcid 50 ⁄ well), the cell index commenced dropping at a steeper gradient than the no-virus cell control after approximately 3 hour of infection ( figure 1 ). this drop in cell index continues at a consistent slope until it flattened out when approaching zero cell index. this steep decrease in cell index with constant gradient was also observed for virus dilutions up to and including 2 log 10 (100-folds), and the profile shifted with increased time in proportion to the dilution made to the virus. virus dilutions beyond 2 log 10 have cell index profiles similar to the no virus input control, and this corresponds to the absence of cytopathic effect as determined by microscopic observation at 72 hour after infection. hence, there was a correlation between the amount of virus used for infection, the onset of the influenza virus-mediated cytopathic effect, and the steep decline in cell index. a human serum with known microneutralization antibody titre to h1n1 virus was used in this study to investigate the real time cell index changes that occur during the assay ( figure 2 ). using influenza virus treated with serum dilutions up to and including a dilution of 1:160, the cell index profile remained essentially the same as the no virus cell control, which correlates with the lack of cytopathic effects under microscopic observation at 72 hour of infection. at a serum dilution of 1:320, the steep decrease in cell index, which is characteristic of cellular cytopathic effect induced by the virus, became evident at around 50 hour post infection, and this was reduced to 40 hour when serum dilution of 1:940 was used. in contrast, for the virus -no antibody control, the onset time for this steep decrease in cell index occurs at approximately 7 hour. for both serum dilutions of 1:320 and 1:940, full cytopathic effect was observed microscopically at 72 hour of infection. from microscopic observation of cytopathic effect, according to the current standard procedures, the neutralizing titre of the human serum used in this study is at 1:160 as it is the last dilution of the serum that prevented cytopathic effect from being detected. an essential part of the microneutralization assay is to confirm the titre of the input virus (normally 100 tcid 50 ⁄well) by performing a titration assay with decreasing serial dilutions of the virus. under normal procedures, cells are examined microscopically after 72 hour of infection for sign of cytopathic effects. in the case of mdck cells, the cytopathic effect is cell death, which is indicative of the presence of live influenza virus infecting and replicating in the cells. therefore, the titre of the virus is taken as the last dilution in which cytopathic effect is present. parallel realtime cell index measurements demonstrated that for wells with cytopathic effects, the profile exhibits a steep gradient linear decrease in cell index after infection with the virus, which can be termed the ''cpe plunge.'' the time in which the cpe plunge became evident appears to be inversely proportional to the amount of virus, therefore the opportunity exists to utilize this aspect to calculate or compare quantitatively different virus concentrations. for unequivocal assignment of cytopathic effect, it normally requires 2-3 days after infecting the cells, with 3 days after infection being the standard time to read virus titration and microneutralization assays. using the real-time cell index monitoring, it is found that apparent cytopathic effect can only be observed microscopically when the cell index has dropped to near zero. as the time of onset of the ''cpe plunge'' becomes evident many hours prior to observable cytopathic effect, it is possible that the time to results can be drastically reduced after some formulation of the method. we compared the current standard method in perfoming a microneutralization assay with one utilizing the real-time cell index measurement to investigate whether this approach is able to offer better performance over the existing one. the current standard neutralization assay is the microscopic observation of antibody mediated protection from virus cytopathic effect in mdck cells. this study showed that this may also be achieved by examining the profile generated from the real-time measurements of the cell index. using real-time cell index monitoring, it is possible to detect inhibitory activity at higher dilutions of the anti-serum than can be detected by the standard microscopic observation of cytopathic effect. therefore, the realtime cell index monitoring could potentially be developed to be a more sensitive method for measuring anti-viral activity. as drug resistant strains of influenza a viruses 4 including the 2009 pandemic h1n1 are being reported, the real-time cell based monitoring system may also have the potential to be developed for use as a diagnostic platform for drug resistance assays. this study suggests that real-time cell index monitoring has the potential to substantially reduce human resources in reading results, as well as reducing time-to-result of these assays from 3 days to two. the saving could be substantial for work involving bio-hazard level 3 ⁄ 4 pathogens such as h5n1 viruses as personnel working with these organisms are require to be highly trained and experienced. in addition, the reduction in transferring plates to and from the microscope in reading cytopathic effect will substantially reduce the possibility of accidents from occurring. furthermore, the system provides objective digital data to an otherwise subjective assay method, which can improve standardization, data exchange, and hence collaboration between different laboratories. with more detailed validation and development, real-time cell index monitoring could transform the way we study and diagnose infection with pathogens such as influenza viruses. the emergence of a novel h1n1 influenza a virus of swine origin, the pandemic a(h1n1) 2009, with transmissibility from human to human in april 2009 posed pandemic con-cern and required modifications to laboratory testing protocols. a new protocol for universal detection of influenza a and b viruses and simultaneous subtyping of influenza a (h1n1) 2009 virus, composed of two-one-step rt-pcrs, fast set infa ⁄ infb and fast set h1n1v (relab, italy), was evaluated and compared to the reference protocol recommended by who. fast set infa ⁄ infb was able to detect influenza a and b viruses circulating between 1995 and 2008 belonging to different subtypes and lineages, and no cross reactions were observed by either fast set infa ⁄ infb or fast set h1n1v. the who assay was found to have a slightly lower end-point detection limit (10 )5 dilution) in comparison to the new protocol (10 )6 ). specificity of the assays was 100% as assessed on a panel of stored clinical samples including adenovirus, respiratory syncytial virus, metapneumovirus, parainfluenza virus, s. pneumoniae, n. meningitidis, h. influenza, and human influenza viruses. the new assay panel allows the detection, typing, and subtyping of influenza viruses as requested for diagnostic and surveillance purposes. the high sensitivity of the protocol is coupled with capacity to detect viruses presenting significant heterogeneity by fast set infa ⁄ infb and with high discriminatory ability by fast set h1n1v. a rapid and sensitive assay for the detection of influenza virus in clinical samples from subjects with ili or low respiratory tract infections is a fundamental tool for epidemiological and virological surveillance, management of hospitalized patients, and control of virus nosocomial transmission. the emergence, in april 2009, of a novel h1n1 influenza a virus of swine origin, the pandemic (a(h1n1) 2009), with transmissibility from human to human poses pandemic concern and required modifications to the laboratory testing protocols. 1 molecular diagnosis of influenza is generally achieved through a twophase process: a screening phase for the detection of virus, and the subsequent strain characterization performed by either sub-type-specific rt-pcr or entire ⁄ partial genome sequencing. 2 during a pandemic, simultaneous implementation of both the detection of influenza a and b influenza viruses and identification of the new subtype is useful for clinical and epidemiological reasons. here, we describe a new protocol including two-one-step rt-pcrs, fast set infa ⁄ infb and fast set h1n1v (relab, italy) that allows universal detection of all influenza a viruses and, simultaneously, all subtypes that are influenza a(h1n1) 2009. specificity and clinical sensitivity of the two-one-step rt-pcrs (fast set infa ⁄ infb and fast set h1n1v; relab, italy) were evaluated by testing 306 selected specimens, including: • fifty samples collected from nasopharyngeal swabs representative of influenza viruses, belonging to differ-ent subtypes and lineages, and other respiratory viruses and bacteria circulating in italy between 1995 and 2008. • six purified a(h5n1), a(h7n7), and a(h9n2) strains, kindly supplied by alan hay, who influenza centre, london, uk. • two hundred-fifty influenza positive samples selected according to type, subtype, clade and viral concentration from >2500 specimens received by the liguria influenza reference laboratory between january 1st and december 31st, 2009. since 1995, nasopharyngeal swabs sampled from patients suspected of having contracted the influenza virus have been collected in viral transport medium, and upon arrival into the laboratory, the samples were divided in ‡3 aliquots. those not immediately processed were stored frozen at )80°c. stored samples were used for this evaluation, and all specimens were re-extracted for the study. samples collected between 1995 and 2008 included specimens positive for: no seasonal a(h1n1) have been detected since january 1st, 2009. furthermore, 1 weak positive sample using fast set infa ⁄ infb, but negative at block pcr and typing ⁄ subtyping assays was tested. the analytical sensitivity of the test under investigation was determined testing ten-fold serial dilutions of seasonal influenza a(h1n1), seasonal influenza a(h3n2), new pandemic influenza a(h1n1) 2009, and b cell culture-grown viruses. the intra-assay reproducibility was measured by testing the same a(h1n1) 2009 positive sample 15 times in the same experiment, while the inter-assay reproducibility was confirmed by testing the same samples in 3 independent experiments. to evaluate the performance of the protocol, all samples were tested using a block pcr confirmation test (seeplex ò rv12 ace detection), 3 and all specimens collected between january 1st and december 31st, 2009 and dilutions were also assayed using the recommended who ⁄ cdc protocol of real-time rtpcr for influenza a(h1n1). typing and sub typing were performed using the who protocol and ⁄ or sequencing. 4 viral rna was extracted from swabs using the qiaamp viral rna mini kit (qiagen) according to the manufacturer's protocol. fast set infa ⁄ infb and fast set h1n1v are two multiplex one-step real time pcr assays developed and evaluated by the liguria regional reference centre for diagnosis and surveillance of influenza in collaboration with relab diagnostics. both assays contain primers and a dual-labelled hydrolysis probe that targets two regions of the matrix gene (table 1) . amplification conditions were as follows: reverse-transcription 50°c for 15 minutes, denaturation 94°c for 2 minutes, then 40 cycles of 95°c for 15 seconds, 58°c for 30 seconds. the entire amplification process extended for 110 minutes. an internal control real-time assay was also incorporated in order to detect pcr inhibition, failed extraction ⁄ pcr and technical error. the cdc realtime rtpcr (rrtpcr) protocol for detection and characterization of swine influenza includes a panel of oligonucleotide primers and dual-labelled hydrolysis (taqman ò ) probes to be used in real-time rtpcr assays for the in vitro qualitative detection and characterization of swine influenza viruses in respiratory specimens and viral cultures. this protocol recommends three primer-and-probe sets: infa, amplifying a conserved region of the matrix gene from all influenza a viruses; sw infa, designed to specifically detect the nucleoprotein (np) gene segment from all swine influenza viruses and sw h1, designed to specifically detect the hemagglutinin gene segment from a(h1n1) 2009. 5 the seeplex ò rv12 ace detection for auto-capil-lary electrophoresis is a multiplex block rt-pcr that applies dpoô (dual priming oligonucleotide) technology and is designed to detect 12 major respiratory viruses, 11 respiratory rna (influenza a and b virus, parainfluenza virus type 1, 2 and 3, respiratory syncytial virus a and b, rhinovirus a ⁄ b, coronavirus oc43 and 229e ⁄ nl63) viruses and 1 dna (adenovirus) virus, from patients' samples including nasopharyngeal aspirates, nasopharyngeal swabs and bronchoalveolar lavage. 3 conventional viral culture was performed inoculating 0ae1 ml of each specimen into mdck-siat1 seeded into 24-well plates for influenza isolation. virus detection was performed by the hemagglutination test using 0ae75% guinea pig red blood cells (rbc). specificity and clinical sensitivity results of the new protocol are reported in table 2 . fast set infa ⁄ infb was able to detect influenza a and b virus circulating between 1995 and 2008 belonging to different subtypes and lineages, and no cross-reactions were observed by either fast set infa ⁄ infb or fast set h1n1v. among specimens collected between january 1st and december 31st, 2009, all 60 fast set infa ⁄ infb and fast set h1n1v high titre positive samples resulted positive using the who ⁄ cdc assay and showing reactivity using infa and sw infa primer-andprobe sets. among 90 low titre a(h1n1) 2009 positive samples at fast set infa ⁄ infb, 28 (31ae1%) were not detected by the who ⁄ cdc assay, but were positive using seeplex ò rv12. the who ⁄ cdc sw h1 primer-and-probe set works in 96ae7% (58 ⁄ 60) and 1ae1% (1 ⁄ 90) of high and low titre a(h1n1) 2009 positive samples, respectively. all a(h3n2) strains collected during 2009 and initially detected by fast set infa ⁄ infb were confirmed after rna re-extraction by seeplex ò rv12 and who ⁄ cdc assay showing reactivity using the infa primer-and-probe set. all infa ⁄ infb were confirmed after rna re-extraction by seeplex ò rv12. one influenza a case identified by the who ⁄ cdc kit (infa primer-and-probe set, ct values: 37ae5, sw infa primerand-probe set: negative) and new protocol (a primer-andprobe set, ct values: 34ae1, a(h1n1) 2009 primer-and-probe set, ct values: 34ae6) was not detected by either seeplex ò rv12 or by who subtyping protocol and ⁄ or sequencing, suggesting a very low viral load or unspecific results by real time assays. the analysis of serial dilutions of cell culturegrown a(h1n1) 2009 showed that the detection limit of fast set infa ⁄ infb, fast set h1n1v, and seeplex ò rv12 was identical (10 )6 ) and 1log 10 lower than that using the who ⁄ cdc protocol (10 )5 ). a similar analysis with respect to a(h1n1) and a(h3n2) strains indicated that fast set infa ⁄ infb sensitivity (10 )6 and 10 )7 , respectively) was 1log 10 lower than that showed by seeplex ò rv12 (10 )5 and 10 )6 , respectively). in comparison with the new protocol, the who ⁄ cdc assays, considering infa primer-and-probe set, was found to have a slightly lower end-point detection, detecting the 10 )5 a(h1n1) and a(h3n2) dilution. also in detecting influenza b virus, fast set infa ⁄ infb sensitivity (10 )6 and 10 )7 , respectively) was 1log 10 lower than that showed by seeplex ò rv12 and the who ⁄ cdc protocol. data on intra-assay and inter-assay precision, measured as cv% of ct showed that the dispersion indices observed had values of less than 3%. since 2009 samples were detected using the new protocol that resulted negative using the who ⁄ cdc assays. the unfortunately low quantity of low titre a(h3n2) samples collected during 2009 did not allow us to highlight differences between assays fast set infa ⁄ infb, and fast set h1n1v positivity was always confirmed by seeplex ò rv12, which demonstrated high sensitivity, showing a detection limit comparable or lower when compared with those observed using the who ⁄ cdc assays. the high analytical sensitivity of seeplex ò rv12 is reported by kim 3 who observed a detection limit of 10 copies per reaction for each type ⁄ subtype of influenza viruses. the high sensitivity of the new protocol is coupled with its capacity to detect viruses presenting a significant heterogeneity by fast set infa ⁄ infb and high discriminatory ability by fast set h1n1v. fast set infa ⁄ infb was able to identify representative influenza viruses of circulating strains during the last decade belonging to different subtypes, lineages, and clusters, and fast set h1n1v primerand-probe set reacted selectively with a(h1n1) 2009 target. a recent report demonstrated that the sw infa assay is not specific to a(h1n1) 2009 and is able to detect both human and avian (h5n1) influenza a viruses and so there is the potential for misidentification. 15 high titre (ct 8ae5 and 15ae6 at fast set infa ⁄ infb) a(h5n1) viruses did not react with fast set h1n1v primer-and-probe set (data not shown). available human a(h5n1) sequences are similar within the h1n1v primer-and-probe regions, but having 4-5 mismatches in the forward primer and, more notably, two of the mismatches occurred within 9 nucleotides of the 3 end, an important determinant for primer specificity. in conclusion, this protocol can be a powerful tool in the diagnostic laboratory setting for specific simultaneous analysis of several samples in minimal time, showing enhanced sensitivity in detecting influenza viruses, and high discriminatory ability in identifying the new pandemic a(h1n1) 2009. a university-corporate partnership to enhance vaccination rates among the elderly: an example of a corporate public health care delivery public health campaigns usually rely on governmental infrastructure and finance for vaccine implementation programs. however, there are many financial and physical barriers which preclude widespread and effective vaccine administration, especially among the elderly. on an international scale, both government agencies and citizen groups have a vested interest in searching for more resourceful methods of attaining significant immunization levels (>75% of the population). in fact, it seems to have become both a grassroots civic and governmental goal, especially among developing countries. we implemented the unique strategy of enlisting the assistance of a privately-owned food market chain to address the public health issue of mass vaccination for the elderly. in this context, publix pharmacy and the university of south florida (usf) recently developed both a handbook and a training program to facilitate the administration of vaccinations. between 2008 and 2009, the publix-usf partnership resulted in administration of over thirty thousand influenza a (h3n2) vaccinations, 76% of which were given to adults over 55 years of age. consequently, vaccine administration costs were decreased by using corporate resources and bypassing overly strained municipal resources. this unique university-corporate partnership successfully delivered h3n2 vaccine to a vulnerable cross-section of society at a lower cost and with minimal side effects and morbidity. it may be safely projected that university-corporate partnerships could result in an effective method for rendering a vital service to an aging and especially vulnerable segment of the population. government policy and funding are the foundation of immunization programs on an international scale. for example, in the united states, governmental programs account for over 50% of the monetary outlay used for immunization. 1 until 2006, the global alliance for vaccines and immunizations (gavi) acted as a catalyst for implementing vaccine and immunization programs in each targeted country. 2 under the auspices of gavi-collaborations between governments, charitable organizations, and multinational health agencies (such as uncief and the who)-many countries have increased their spending for vaccination programs. however, development of financially sustainable immunization programs geared toward reaching the majority of the population are still at a nascent level of evolution. 3 the development of more innovative and costeffective approaches has become imperative in order to reach a greater number of vaccination candidates. administering the influenza vaccine only to the subpopulation of over 65 year olds would save an estimated 220 000 quality-adjusted life years in a cohort of approximately half the world's population. 4 widespread public vaccination programs are made more complex by the continuing development of newer vaccines, concomitant specialized administration costs, and the logistical challenge of conveying recipients to vaccination points of service. 1, 5 in spite of the increasing complexity of mass vaccination, cost-benefit analyses clearly favor annual influenza vaccination in the elderly population on an international scale. 4, 6 recently, in 2008, influenza vaccine administration was reported to reach between 32% and 82% of the elderly population, which denotes varying degrees of success within each particular country. 7, 8 however, there was also a report of a uniform plateau effect at around 75% of the population, beyond which additional vaccination coverage was difficult to achieve. 9 physical limitations to vaccination seem to be more insurmountable for the elderly. unfortunately, this is the population segment which could experience the most significant vaccination-associated mortality reduction. 10 we employed the unique strategy of involving the resources of publix supermarkets, a corporate food market chain, to address the public health issue of widespread vaccination for the elderly. we took advantage of recent 2005 changes in the florida statutes, which expanded the scope of pharmacists' practice to include administration of vaccines. subsequently, publix pharmacy and the university of south florida (usf) developed a handbook and training program to facilitate and enhance vaccine administration by publix pharmacists. by using proprietary pharmacists and more practical supply storage, we were able to decrease the costs of vaccine administration. the consumer was charged $10 for administration costs plus the cost of the injection itself, regardless of insurance or eligibility for governmental subsidy. although patients were initially self-selected, they were ultimately excluded if they had demonstrated prior adverse effects to influenza vaccinations or to any of the components of such vaccinations. between 2008 and 2009, the publix-usf partnership vaccinated 30 063 people against influenza a (h3n2), of which 29 404 were florida residents. the age range was 1-105 years old with a median age of 65 years old. seventysix percent of the participants were over 50 years old (see figure 1 ). within the population surveyed, the reported side effects of the vaccine in this study were not serious, but included: vertigo, cold sweats, chills, vomiting, syncope, rash, nausea, stomach pain, elevated blood pressure, injection site reaction, inflamed bursa, and bilateral thigh discomfort. participants from all socioeconomic classes were vaccinated. an income-by-zip code analysis revealed 44% of those vaccinated resided in zip code areas where the average household income was <$50 000 per year. of those remaining, 25% had an average income of $50 000-$70 000 per year, and 31% had an income of >$70 000 per year. each person vaccinated was charged ten dollars for administration costs. this represents a decrease in the administration costs ranging from one dollar to ten dollars saved per vaccine. 11, 12 conclusion this unique university-corporate partnership successfully delivered h3n2 vaccine to a high-risk population with decreased vaccine administration costs. the influenza vaccine is well-tolerated, with minimal side effects when patients who have a history of adverse reactions are excluded. we can postulate that university-corporate partnerships may indeed be effective at reaching the aging population which is a challenge in most communities. this delivery model may prove to be another tool for improving the efficiency of mass immunization by facilitating accessibility, which results in wider coverage. this model also enhances delivery of healthcare by decreasing costs of immunization regardless of whether the payer is a government, insurance company, or self-pay consumer. the gavi initiative stressed three goals for accomplishing sustainability and independence in immunization programs. 3 the goals were to: (i) mobilize additional resources from governmental and non-governmental sources; (ii) improve program efficiency to minimize additional administration resources needed; and (iii) increase the reliability of funding. empowering privately owned corporations within the community, such as food markets or pharmacies, to administer vaccines mobilizes additional resources to readily achieve the first goal of gavi. mobilizing resources of non-healthcare, corporate vaccination locations enhances accessibility due to travel convenience. in our study, participants came from all socioeconomic classes, suggesting that ease of access is independently hindering mass vaccination, and that people of all incomes are more likely engaged when access issues are eliminated. the second and third goals were also accomplished by recruiting a corporation's resources for vaccine administration (refrigeration, storage, and employees). this minimizes the money spent from vaccine program funds to support the infrastructure of immunizations, thus improving financial efficiency and sustainability. financial efficiency implies that money is spent to safely reach as large a portion of the population as possible. by using corporate storage facilities instead of paying for independent facilities, money can be spent elsewhere. more vaccines can be purchased and more money can be spent on media communications to encourage vaccination. sustainability requires the ability to fund annual vaccination programs which reach 75% of the population or greater. key to the control of pandemic influenza are surveillance systems that raise alarms rapidly and sensitively. in addition, they must minimise false alarms during a normal influenza season. we develop a method that uses historical syndromic influenza data from the existing surveillance system 'servis' 1 monitoring seasonal ili activities in scotland. we develop an algorithm based on wcr of reported ili cases to generate an alarm for pandemic influenza. wcr is defined as the ratio of the number of reported cases in a week to the number of cases reported in the previous week. from the seasonal influenza data from 13 scottish health boards, we estimate the joint probability distribution ( figure 1) we compare our method, based on our simulation study, to the mov-avg cusum and ili rate threshold methods and find it to be more sensitive and rapid. the wcr method detects pandemics in larger fraction of total runs within the same early weeks of pandemic starting than does any of the other two methods ( figure 2 ). as shown in the table, for 1% pandemic case reporting rate and detection specificity of 95%, our method is 100% sensitive and has mdt of 4 weeks, while the mov-avg cusum and ili rate threshold methods are, respectively, 97% and 100% sensitive with mdt of 5 weeks. at 99% specificity, our method remains 100% sensitive with mdt of 5 weeks. although the threshold method maintains its sensitivity of 100% with mdt of 5 weeks, sensitivity of mov-avg cusum declines to 92% with increased mdt of 6 weeks. for a two-fold decrease in the case reporting rate (0ae5%) and 99% specificity, the wcr and threshold methods, respectively, have mdt of 5 and 6 weeks with both having sensitivity close to 100%, while the mov-avg cusum method can only manage sensitivity of 77% with mdt of 6 weeks. the first cases of the 2009 pandemic were reported in scotland in the 29th week of the season. the wcr algorithm as well as the mov-avg cusum method detects the pandemic 12 weeks later in week 41. the ili threshold method detects it 1 week later in week 42. both the wcr and mov-avg cusum methods therefore outperform the ili threshold method by 1 week in the retrospective detection of the 2009 pandemic in scotland. while computationally and statistically very simple to implement, the wcr method is capable of raising alarms rapidly and sensitively for influenza pandemics against a background of seasonal influenza. although the algorithm has been developed using the servis data, it has the capacity to be used at large scale and for different disease systems where buying some early extra time is critical. more generally, we suggest that a combination of different statistical methods should be employed in generating alarms for infectious disease outbreaks. different detection methods would provide cross-checks on one another, boosting confidence in the outputs of the surveillance system as a whole. real-time evidence being created worldwide will greatly contribute to the full understanding of influenza pandemics. here we report the real-time epidemiology and virology findings of the influenza a(h1n1)2009 pandemics in mongolia. the epidemiological and virological data collected through isss of nic, nccd, mongolia (real-time information on registered ili cases and virological laboratory results are available from the weekly updates in the nic, mongolia website: http://www.flu.mn/eng/index.php?option=com_ content&task=category&sectionid=5&id=36&itemid=51) were used for analysis in relation to the previous seasonal influenza activities in the country. influenza viruses were detected in naso-pharyngeal samples from ili patients by rt-rt-pcr with applied biosystems fast real time pcr system 7500, using primers and instructions supplied by cdc, usa. 1 influenza viruses were isolated by inoculation of rt-rt-pcr-positive samples of mdck cell culture according to the standard protocol. 2 ten representative strains of a(h1n1)pdm viruses were selected for sequencing of different gene segments, namely: a ⁄ ula, and a ⁄ dundgovi ⁄ 381 ⁄ 2010. sequencing of influenza virus gene segments was performed in applied biosystems 3130xl genetic analyzer using primers and instructions supplied by cdc, usa, 3 and bioinformatic analysis was performed with abi ⁄ seqscape v.2.5 and mega4 programs. the pandemic alert in mongolia was announced by the government on april 28, 2009, just after the who announcement of the pandemic alert phase, and planned containment measures were intensified. despite intensive surveillance, no a(h1n1)pdm virus was detected in mongolia until the beginning of october 2009. around 40 suspected cases, mostly arriving from the a(h1n1)pdm epidemic countries, tested zero by rt-rt-pcr for a(h1n1)pdm virus. the first a(h1n1)pdm case detected by the routine surveillance system in ulaanbaatar city, the capital of mongolia, was confirmed by rt-rt-pcr on october 12, 2009 (41st week of 2009). the reported ili cases escalated rapidly, reached the peak in the 43-44th week of 2009, and gradually decreased thereafter ( figure 1 ). week of 2010. however, the registered ili cases increased again from the 5th week of 2010, and peaked at the 8-9th weeks of 2010. the viruses isolated during this 2nd peak were influenza b strains ( figure 1 and table 1 ). for the genetic characterization of the mongolian pandemic isolates, 3 gene segments i (pb2), 2 gene segments ii (pb1), 3 gene segments iii (pa), 9 gene segments iv (ha), 3 gene segments v (np), 4 gene segments vi (na), 7 gene segments vii (m), and 3 gene segments viii (ns) of the representative a(h1n1)pdm mongolian strains were sequenced, and all sequences have been deposited in the genbank (accession numbers: cy050844, cy0533648, cy050846, cy050845, cy065987, cy065989, cy065990, cy 065988, cy065991, cy065985, cy065984, cy065986, cy052366 , cy053365, cy054546, cy054547, cy055169, cy 055170, cy055171, cy056363, cy055308, cy057191, cy057083, cy065995, cy065997, cy065998, cy065996, cy 065999, cy065993, cy065992, cy065994, cy054548, cy054549, cy073448). all 8 genes of mongolian strains were possessing 99ae4-99ae9% similarity with the genbank deposited gene sequences of the original pandemic strain a ⁄ california ⁄ 072009(h1n1). the who declared the pandemic alert phase (phase iv) on april 27, 2009, 4 and was prompted to announce the pandemic phase (phase v) two days later. 5 after 43 days, the who declared the beginning of the pandemic peak period (phase vi) on june 11, 2009. 6 however, in mongolia, the pandemic alert period continued for 168 days. mongolia was free of the pandemic virus during the whole first wave of the pandemics in the northern hemisphere. with the confirmation of the 1st influenza a(h1n1)pdm case on october 12, 2009 in ulaanbaatar, mongolia entered into the pandemic phase (phase v), and after just 2 weeks, the registered ili cases peaked, confirming mongolia shifted into the pandemic peak period (phase vi), which i i i v i i v i v v v i i i i iii iv v vi vii viii worldwide by who in mongolia coincided with the 2nd wave of pandemics in many countries of the northern hemisphere (see, picture 1 and table 1 ). despite the relatively milder clinical manifestations, the disease burden for the health service was enormous, while the morbidity per 10 000 population at the peak period was 5-6 times higher above the upper tolerant limit, and 3-4 times higher above the seasonal influenza outbreaks. in contrast to the seasonal influenza outbreaks where over 80% of the registered ili cases have been in the age group under 15, 2 it has been observed that over 50% of the registered cases in this pandemic peak period were in the age group of 16-60. on january 18, 2010, we regarded the pandemic had entered into the post-peak period (phase vii) when the registered ili cases became lower than the upper tolerant limit, during which time mongolia experienced an influenza b outbreak. on may 24, 2010 we determined that mongolia entered the post-pandemic period (phase viii) as the influenza virus isolations were almost stopped, and after no pandemic virus detected for 3 months. who announced pandemic vii and viii phases much later. 7, 8 this first ever real-time laboratory confirmed influenza pandemics in mongolia and confirmed some variations of pandemic spread in different parts of the world. the comparison of deduced amino-acid sequence changes have shown that the mongolian strains belong to the clade 7, according to the classification of a(h1n1)pdm influenza strains suggested by m. nelson, 9 which has circulated worldwide since july 2009. this is also evidence that the 1st wave of the pandemics did not hit mongolia. the who public health research agenda for influenza 1 is aimed to support the development of evidence needed to strengthen public health guidance and actions essential for limiting the impact of influenza on individuals and populations. each stream-specific group reviewed and discussed the proposed organization, content, rationale, and global health importance of their designated research stream. specific research recommendations were made for topics within each stream: background: a syndromic surveillance system using nonclassical data sources for detection and monitoring evolution of flu and flu-like illness (ili) in djibouti is reported here as part of the preliminary report of djibouti who-copanflu international study (wcis)**. methodology: clinical reports, over-the-counter drug sales, lab diagnosis report, and health communication trends were obtained for an integrated statistical analysis. results: transition to winter is concomitant with upsurge of ili cases and ili drug sales. in addition, more rural folks manage ili infections on self medicament than through clinical consultancy. inefficient and vague data collections were observed. a successful implementation of wcis will create a platform upon which challenges faced in djibouti health department in routine surveillance will be addressed to achieve a near-real time surveillance of flu pandemic. conclusion: innovations, prompt reporting, and instituting open source syndromic surveillance system software's in resource limited environment like djibouti will enhance early detection and evolution monitoring of pandemic flu. the spanish flu in 1918 ⁄ 1919 infected and killed millions of people, and threatened to wipe humanity off the face of planet. however, the recent scenarios of influenza h1n1 (2009) pandemics' worldwide occurrence fell short of most scientific prediction on its magnitude and intensity. this dampened their confidence; they cannot state precisely as to when, how, where, and which of the spanish flu-like pandemic will occur in the future. in support of scientific community and governments, the who hasn't gone to slumber, but is reminding its member states to up their post pandemic surveillance and monitoring of influenza virus in circulation for advance preparedness in case of an outbreak. 1 despite all uncertainty around the pandemic flu h1n1 2009, there remains a common knowledge and understanding that this flu has shown a great potential to evolve and cause huge morbidity and mortality. although its future magnitude may be unpredictable, its recurring events have severe consequences on human health and the economic well being of everyone. 1 and therefore, advance planning and preparedness is critical in protecting any population in the future, especially those located in resource limited environment without universal health cover and generous disaster emergency funds. 2 . two collapsed sets of a weekly and monthly mean data (of four years period) were clustered in five categories of ili cases, drug sales, lab results, vaccine consumption, and health promotion. this was followed by a descriptive statistics analysis of cumulative weekly and monthly data to establish presence or absence of trend. time series analysis was not done due to data limitation. copanflu program: as at the time of going to press, the cohort study is at the household recruitment and inclusion phase and the study covers the djibouti city. it is in our intention to use the cohort study findings to validate or improve the niph ministry of health djibouti ili surveillance effort for better preparedness. clinical service: 83% of all health facilities are in djibouti city. of the 39ae9% (326 445) of the population that seeks medical care on influenza and influenza like illness each year, 58ae2% (187 586) and 11ae9% (38 389) of them are attended to at the city's public and private clinics, respectively ( figure 1 ). the rest are attended from the regional health centers. the majority of ili incidence sharply rise with the onset of the winter season (october to april), affecting mostly the middle age group (1-24 years). pharmaco-surveillance: 2% (99 350) of total prescriptions were antipyretic and antiflu drugs, 91ae4% (90 765) of which were consumed by peripheral regions, the non djilab diagnostics: the annual ili lab diagnosis was negligible 0ae0072% (250), which can be attributed to less equipped virology laboratories to warrant routine service utility. documented cases were from previous bouts of avian influenza that had a human incidence from 2006 and 2009. 5 with support of egypt-based naval army medical research unit three (namru 3), clinicians were motivated to sample all ili patients and submit to collaborating international reference influenza lab in cairo, egypt. vaccination: influenza vaccinations were undocumented, but at least 0ae4% (3122) of population sought the service (for yellow fever and meningitis) as mandatory travel advisory or as childhood immunization need. at the time of going to the press, there were at least 80 000 vaccine doses of h1n1 (2009) virus donations yet to be administered. health promotion and hygiene: print and audiovisual risk communication remained favorite means of reaching out to urban dwellers (70ae6%). while to the rural and nomadic population, person to person communications was the preferable means. to increasing public awareness that will encourage reporting of ili cases and entrench risk aversion health behavior that limits flu spread, who-copanflu international study djibouti has incorporated basic training on ili infection and personal hygiene by interviewers during household inclusion. improving national epidemic surveillance capacity and response under new international health regulation 6 is important for any nation, including djibouti. our finding indicates the winter season predisposes one to ili infections; they therefore opt for medical services or self medication depending on their capability and ⁄ or understanding. in djibouti, almost no city dwellers favors self medication over clinical consultation, suggesting the presence of inhibitory factors like distance from the health centers and the cost of accessing consultancy. common in the absence of universal primary health care setting, it therefore calls for active innovativeness in outbreak detection, disease reporting, and preventive medicine on the part of health authority so as to achieve good population health. 7 in respond to these, niph has turned resource limitation to a motivation instead and is working towards institutionalizing a near-real-time syndromic surveillance system as a core functional unit. it capitalizes on three major aspects within its reach: prompt accurate data generation for analysis, ehesp wcic-study input, and information technology use. prompt accurate data generation for analysis: data used in our analysis suffered from un-timeliness (weekly instead of daily basis), incompleteness (vague over-counter drug sales records), entry errors (incidence case reports), and poor collection format (most of data collection forms). use of satellite handset phones for regional health centers and mobile phones for city sentinel clinics will reduce unnecessary data delivery delays. in addition, creating awareness to data entry personnel on the importance of careful and completeness of entries is important, as is the need to reformat data collection forms to capture exact aspects of surveillance needs for relevant executable analysis. 8 besides alerting for immediate impending epidemics, these data can also be adopted for projective predictive modeling of annual epidemics, including that for influenza. 9 ehesp wcic-study input: djibouti wcic-study is complementary to the existing syndromic surveillance system, but with emphasis on flu and flu-like illness. various innovations as suggested above are used in seeking to overcome the prevailing challenges. while every attempt is made to realize its (wcic-study) objective and for global comparison, lessons learned from successful implementation will form a platform for future refined syndromic surveillance protocol as equally reported elsewhere in asian countries. 7, 10 information technology: national institute of public health djibouti has an informatics department with sufficient working pcs and personnel to execute efficient data collection and management for epidemiological analysis. however, licensing cost of near-real time syndromic surveillance software is prohibitive, but the open access software with capacity to generate custom graphs, maps, plots, and temporal-spatial analysis output for specific syndromes should make implementation a lot easier. such output for conditions like flu (or gastroenteritis) will be essential to cause prompt response of the local public health office and international partners in saving lives and suffering of djibouti people. 11 pandemic flu surveillance and preparedness requires multifaceted, interdisciplinary, and international approach whose efficiency and efficacy can only be refined over time. building on the health care system's swot for preparedness, the ehesp wcic-study promises to refine surveillance system operation and knowledge on individual's risk determinants to swine flu (h1n1) 2009 virus infection at the household level in djibouti. these efforts are ultimately creating available control options at the time of need (pandemic occurrence), and at the same time exploring investment in quality data profiling and information technology, which will include syndrome surveillance software systems like essence, ewors, or other open sourced ones. the antibody efficacy -which compares the illness frequency between those with and those without a protective level of pre-epidemic hi antibodies ( ‡1:40) -has been proposed 1 ; however, this index has rarely been used due to practical difficulties in confirming the strain-spe-cific disease corresponding to each of the vaccine-induced antibodies. we followed 114 elderly individuals residing in a nursing home, whose serum specimens were obtained before and after undergoing trivalent influenza vaccination, in 2002 ⁄ 2003 influenza season (medium-scale mixed [a ⁄ h3n2 and b] epidemic in study area, and a ⁄ h3n2 was circulating at the nursing home). 2 the serum antibody titre to each strain of influenza virus was measured by the hi method, using the same antigens as those in the vaccine. all participants' body temperatures, respiratory symptoms, other general symptoms, hospitalization, discharge, and death were recorded daily from 1 november 2002 to 30 april 2003 in a prospective manner. when the participants suffered any influenzalike symptoms, such as sudden fever ‡37ae8°c, throat swabs were collected and tested using a rapid diagnosis kit for influenza, which utilizes an immunochromatographic method. the adjusted odds ratios (or adj ) for febrile illness and kit diagnosed influenza were evaluated using multiple logistic regression models adjusting for possible confounders (i.e., age, sex, coexisting conditions, and vaccine strains). after vaccination, the proportion of subjects achieving an hi antibody titre ‡1:40 (seroprotection level) were 61ae4% (51ae8-70ae4%) for a ⁄ h1n1, 79ae8% (71ae3-86ae8%) for a ⁄ h3n2, and 26ae3% (18ae5-35ae4%) for b. during the follow-up period, the a ⁄ h3n2 strain was isolated therein, and 44 subjects experienced sudden-onset fever ( ‡37ae8°c), and eight subjects were positive for rapid diagnosis kit. patients with a seroprotection level of the hi antibody titre ( ‡1:40) had lower incidences of febrile illness (or adj , 0ae35; 95% ci, 0ae09-1ae28) and rapid kit diagnosed influenza (or adj , 0ae35; 95% ci, 0ae03-4ae64) than those with a lower titre. thus antibody efficacy (1 ) or adj ) against fever related to a ⁄ h3n2 and kit diagnosed influenza were both estimated to be 65%. although statistical significance was not detected due to limited sample size, these results lend support for the usefulness of antibody efficacy. some data presented within this manuscript was also published in hara et al. asia via a regional network from which epidemics in the temperate regions were seeded. 1 the virus isolates obtained from nasopharyngeal swab specimens from outpatients were typed and subtyped by the hemagglutination (ha) inhibition assay. 2 the emergence of a ⁄ fujian ⁄ 411 ⁄ 2002 coincided with higher levels of influenza-like illness in korea than what is typically seen at the peak of a normal season. most of the intermediates and fujian-like strains were isolated from asian countries, and the mutational events associated with the fujian strains took place in asia. closely dated phylogeny from december 26, 2001 to august 11, 2002 showed that the antigenic evolution of the h3n2 fujian strains had periods of rapid antigenic changes, equivalent to 10 amino acid changes per year ( figure 1 ). the fujian-like influenza strains were disseminated with rapid sequence variation across the antigenic sites of the ha1 domain. the antigenic evolution of the fujian strains was initiated by exceptionally rapid antigenic change that occurred in asia, which was then followed by relatively modest changes. some of the data presented in this manuscript was previously published in kang et al. we compared reactivity to the novel virus strain using haemagglutination inhibition (hi) assays performed on discarded plasma specimens left over from routine testing. samples were taken from healthy adult blood donors (>16 years) before and after the ph1n1 influenza epidemic that occurred during the southern hemisphere winter of 2009, and again prior to onset of the 2010 southern hemisphere influenza season. reactivity to the novel h1n1 2009 strain of influenza was relatively uncommon among the healthy adult population during the first australian winter wave, rising from a baseline of 12% to 22%. a further increase in the seropositive proportion from 22% to 43% was observed over the summer months, most likely attributable to immunisation. this level of immunity appears to have been sufficient to constrain the 2010 winter epidemic. together with a final serum collection, planned for late 2010, these data will aid evaluation of the extent and severity of disease in this 'second wave' of ph1n1. assessment of the extent of disease due to novel influenza a(h1n1) virus (ph1n1) during the 2009 winter outbreaks in australia was made difficult by the generally mild nature of disease. the epidemic was experienced in a staggered fashion around the country, 1 reflecting the considerable geographical distances between state and territory capital cities ( figure 1 ). differences in the intensity of case-finding during the evolving pandemic response and between jurisdictions hindered comparisons of disease burden in distinct geographical regions. 2 rates of reported hospitalisations and deaths appeared fairly similar across states 1 but, without a consistent exposure denominator, assessment of relative severity was difficult. we conducted a national serosurvey of antibody to ph1n1 using residual plasma from healthy blood donors collected before and after the 2009 epidemic to estimate ph1n1 exposure. 3 here we report the findings of that first collection, together with new data on seroprevalence of ph1n1 antibody in specimens gathered in march-april 2010. these latter samples were collected prior to onset of seasonal influenza activity to assess the impact of a national ph1n1 vaccine program conducted in spring ⁄ summer 2009 ⁄ 10 on the proportion of individuals with antibody titres deemed protective. 4 findings informed estimates of population susceptibility to ph1n1 prior to the 2010 influenza season and provided a baseline for a subsequent serosurvey that will be collected at the end of 2010 to assess the extent of exposure during the 'second wave.' tralian red cross blood service (the blood service) for dengue fever surveillance studies. these samples were used to provide a baseline estimate of prevalence of cross-reactive antibody to ph1n1 in the australian population. discarded plasma specimens, taken for virologic testing from healthy adult blood service donors, were prospectively collected at two additional timepoints for measurement of antibody to ph1n1. collection periods were as follows: approximately 120 plasma samples were randomly selected from donors in each of brisbane, hobart, melbourne, newcastle, perth, sydney, and townsville on each occasion. up to 20 specimens were identified in each of the following age strata: 16-24, 25-34, 35-44, 45-54, 55-64, and >65 years. at the last collection timepoint, there was deliberate over-sampling of the oldest and youngest age strata in which approximately 40 specimens were collected (i.e., up to 160 specimens per site). in accordance with the provisions of the national health and medical research council's national statement on ethical conduct in human research, individual consent was not required for use of these specimens, given the granting of institutional approval by the blood service human research ethics committee. reactivity of plasma against ph1n1 was measured in haemagglutination inhibition (hi) assays using turkey red blood cells (rbc). 5 egg-grown a ⁄ california ⁄ 7 ⁄ 2009 virus was purified by sucrose gradient, concentrated and inactivated with b-propiolactone, to create an influenza zonal pool preparation (a gift from csl limited). plasma samples were pretreated with receptor destroying enzyme ii (denka seiken co. ltd), 1:5 (volume ⁄ volume) and tested as previously described. 6 following 1 hour incubation, 25 ll 1% (volume ⁄ volume) of rbc was added to each well. hi was read after 30 minutes. any samples that bound to the rbc in the absence of virus were adsorbed with rbc for 1 hour and reassayed. samples in which background activity could not be eliminated by these means were excluded from the analysis. titres were expressed as the reciprocal of the highest dilution of plasma where haemagglutination was prevented. a panel of control sera and plasma samples was included in all assays. it comprised paired ferret sera pre-and postinfection with the pandemic virus or seasonal influenza a(h1n1), a(h3n2), or influenza b viruses and paired human plasma and sera collected from donors before april 2009 or after known infection with the pandemic virus or after immunisation with the australian monovalent pandemic 2009 vaccine. all assays were performed by the who collaborating centre for reference and research on influenza. for each of the three study timepoints and within each age group, the proportion of seropositive individuals (hi titres ‡40) was calculated, with exact (clopper-pearson) confidence intervals. the contribution of individual variables (age, gender) and location to seropositive status was assessed in separate multivariate logistic regression models developed to assess the post-pandemic and pre-influenza season 2010 collections. all statistical analyses were conducted in stata 10. locations of specimen collection are shown in figure 1 , together with the number of samples tested from each centre. samples with high background hi titres or discrepancies between assays were excluded at each timepoint as follows: 5 at baseline, 27 from the post-pandemic collection, and 0 in early 2010. pared with baseline was 10% overall, rising from 12% to 22% (table 1 ). the only jurisdictions in which seropositive proportions were higher in october ⁄ november than in the baseline collection were hobart [31% (95% ci 22ae3, 39ae7)], perth [24% (16ae3, 31ae7)], and sydney [24% (16ae3, 31ae7)]. in the multivariate regression model, the only jurisdiction in which exposure appeared somewhat higher than the reference population of brisbane was hobart [or 1ae83 (95% ci 0ae99, 3ae4), p = 0ae06]. a marked age effect on antibody status was observed at this timepoint, with an increase in the proportion of seropositive individuals in relation to the baseline collection only noted for those aged between 16 and 34 years (table 1) . according to the multivariate model, the youngest and oldest cohorts had similar titres, with all other groups showing significantly lower seropositive proportions than the reference population of 16-24 years [e.g. 45-54 years or 0ae36 (95% ci 0ae21, 0ae64, p < 0ae0001)]. an overall increase in the seropositive proportion from 22% to 43% was observed between october 2009 and april 2010, distributed throughout all jurisdictions ( (17, 25) ]. antibody titres prior to the 2010 influenza season rose in all age groups, but remained significantly lower among [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] year olds than in the youngest age cohort (table 1) . adjusted ors for the seropositive proportion in the multivariate model in these age groups were: 35-44 years [or 0ae49 (95% ci 0ae31, 0ae77)]; 45-54 years [or 0ae48 (0ae31, 0ae75)]. the relatively low titres observed in these groups reflected small incremental increases in the seropositive proportion across each of the time points studied, suggestive of both low rates of infection and vaccination. the rise in immunity observed across the population was most likely attributable to immunisation in the majority, given the absence of observed outbreaks and very few notified cases of ph1n1 during the period between the two plasma collections. this study suggests that, while adult exposure to ph1n1 during the 2009 southern hemisphere winter was uncommon at around 10%, vaccine uptake in the australian population over the period november 2009-may 2010 was in the order of 20%. this latter estimate is in keeping with recently published figures for adult ph1n1 vaccine coverage from a national immunisation survey conducted by the australian institute of health and welfare. 4 in that survey, vaccine coverage was significantly higher in tasmania than in other states, but mostly in those over 65 years of age, possibly in a subgroup whose health status may have differed from that of the donor population. 4 no allowance has been made in this analysis for likely waning of natural or vaccine induced immunity, possibly resulting in lower estimates of natural and ⁄ or vaccine exposure than may have occurred over the period. regardless of such intervening processes, the seropositive proportion among australian adults at the start of the 2010 winter season appeared likely to be sufficient to constrain transmission of infection in the age groups tested. this assertion has been borne out in practice, with only modest levels of influenza reported during the late and protracted 2010 season. 7 a final serum collection is planned for the end of the 2010 influenza season in australia from which to assess the level of exposure in relation to the baseline observed here. the need for epidemiologic studies such as this has been highlighted by groups such as the european centre for disease control to aid evaluation of the extent and severity of the 'second wave,' 8 known to be variable from historical reports of past pandemics in disparate populations. 9 in 2009-2010, the first wave of the swine-origin novel h1n1 flu (h1n1) pandemic swept across the world, including japan. to examine the epidemiological nature of this novel infectious disease among school children within and among small regional communities, we have carried out a complete survey on the incidence of h1n1 among school children using absentee reports provided by school health teachers in two small administrative districts (population: about 140 000 in total) in japan. we then examined the epidemiological diversity on the inci-dence of h1n1 within and among small regional communities. we investigated seventeen elementary and ten junior high schools in moroyama-town and sakado-city located in the central part of saitama prefecture. populations are: all ages, 37 015 and 100 634; elementary schools, 1873 and 5389; junior high schools, 1131 and 2568, respectively. the number of school children in each school ranges from 137 to 876. the surveillance system was built on an apache-and mysql-based web server using html, php, and java-script. 1 school health teachers enter information on children absenteeism due to school infectious diseases via web browsers at each school infirmary on a daily basis. in addition to the trend graphs shown on the web browser, detailed analyses were reported to the schools and local educational boards weekly. the basic reproduction number (r 0 ) of h1n1 was estimated according to becker. 2 agentbased modeling and simulations were also performed using a multi-paradigm simulator anylogic version 6.5 (xj technologies, st. petersburg, russia). by the end of march 2010, cumulative incidence (ci) of h1n1 among school children in moroyama and sakado reached 30% and 34%, respectively. the overall r 0 among school children in this area was 1ae43. vaccination rate of children in this area during the surveillance period was reported to be very low (<10%). there was no considerable difference between the epidemic curves in this neighboring town and city. on the other hand, in the individual schools, the cis as of the end of march 2010 scattered from 16% to 51% ( figure 1 ) even though the schools are closely located. to examine the cause of this diversity, we built an agent-based community model 3 consisted of the same numbers of agents as those of children in the actual schools and people in moroyama and sakado to simulate the infection. the ratio of probability of infection in schools and the remaining places were assumed to be 1:1 or 10:1. using a heuristic optimization scheme, we estimated the parameters for the simulations to give the overall ci of 33% (the ci as of the end of march 2010). we then performed simulations repeatedly. the cis obtained with the repetitive simulations with the assumption of higher probability of infection in schools scattered from 23% to 44%, indicating that the cis of the small population communities may vary considerably, even though all the agents were assumed to have the same susceptibility to infection at the beginning, and the other conditions were the same. the policies for surveillance ⁄ analyses ⁄ prevention of communicable diseases in local communities have generally been decided on governmental-and ⁄ or each local administrative district-basis (populations: several hundred thousands to several millions) in japan. we found the considerable variations in the cis of h1n1 for children among much smaller areas, i.e., the school districts (populations: all ages, several thousands; school children, several hundreds). we thus conclude that the granularity of surveillance ⁄ analyses ⁄ prevention should be finer than in the past to achieve the most effective policies against influenza and similar communicable diseases in the local communities. the cause of this diversity can be explained in part by the stochastic nature of infection transmission processes in the small populations shown by the agent-based simulations. we have already conducted a complete questionnaire survey for the school children and their parents to clarify the relevance of the other issues including differences in environmental factors, preventive policies (e.g., vaccination, school closures), etc., in each school. the detailed analyses will be reported elsewhere. a www-based surveillance system for transmission of infectious diseases among school children within and among small regional communities. j epidemiol 2010; 20(s1):s151. this study confirms previous findings that age, pandemic influenza vaccination, and history of ili are associated with elevated post-seasonal gmt. this study also shows that seasonal influenza vaccination may have contributed to an increase of the hai titer, especially in the elderly. further analyses in this cohort are needed to confirm and explain these first results. the follow-up of subjects involved in the copanflu-france cohort will provide data to study the risk factors for infection by the influenza virus. the first cases of the 2009 a ⁄ h1n1v pandemic influenza were reported in mexico and the united states in april 2009. given the context of this new influenza virus and considering the likelihood of its pandemic spread, the cohorts for pandemic influenza (copanflu) international consortium was created in order to study individual and collective determinants of pandemic a ⁄ h1n1v influenza across different countries by setting up prospective cohorts of households, followed during 2 years. this study relies on the first available data from the copanflu-france project, which is part of the copanflu international consortium. we studied factors associated with elevated haemagglutination antibody titers against a/h1n1v at entry in the copanflu-france cohort. we focused in this primary analysis on the association between the titers and influenza vaccination (seasonal or pandemic) across age groups. the copanflu-france cohort was set up in fall 2009. inclusions began on december 4, 2009 and ended on july 23, 2010. households were sampled using a random telephonic design (mitofsky-waksberg method) 1 in a stratified geographical sampling scheme, aimed at including a sample of subjects representative of french general population. all household members were eligible to the cohort, without any age limit. the inclusion of a household required the participation of all members: the refusal of one or more member(s) prevented the inclusion of other members. the protocol was approved by a research ethics committee and written informed consent was obtained for all subjects. this study requires several visits to the households by nurses who collect written data with questionnaires and biological samples. during the inclusion visits, nurses collected from all subjects detailed data regarding medical history, including vaccination and preventive measures against influenza. blood samples were collected at entry and centralized. a standard hai technique was adapted to the detection and quantification of antibodies to the 2009 a ⁄ h1n1v virus. 2 the titration endpoint was the highest dilution that exhibited complete inhibition of haemagglutination in two independent readings. the lowest read dilution was 1 ⁄ 40. geometric mean titers were calculated for hai assays with the use of generalized estimating equations for interval-censored data, 3, 4 taking into account a within-household correlation. multivariate models were derived from this method to identify factors associated with elevated gmts. we defined the ''gmt ratio'' (gmtr) as the multiplicative factor applied to the gmt in presence of an explanatory variable. for qualitative explanatory variables, a gmtr of n means a predicted n-fold higher gmt for subjects exposed to the considered factor compared to others. for continuous explanatory variable, the same interpretation applies to a unit difference. the following variables were included in the multivariate models: age, history of pandemic or seasonal influenza vaccination, and history of ili. age was categorized in three groups: 0-15 years (reference group), 15-50 years and over 50 years. the definition of ili was that used by the cdc 5 : fever ‡37ae8°c and cough and ⁄ or sore throat without another known cause. history of ili was defined as an ili reported by the subject between september 12, 2009 (beginning of the influenza epidemic in france) and the date of inclusion. this preliminary analysis included 1304 subjects belonging to 544 households. results reported hereafter do not account for missing data. participating households were sized 1-7 subjects, mean size = 2ae4. in comparison, the mean size of french households is 2ae3 according to the latest national census. 6 the median age of subjects at entry was 40ae6 years [iqr: 16ae5; 58ae6] versus 34ae4 [15ae2; 52ae5] for french population. 7 the proportion of subjects reporting a history of ili since the beginning of the epidemic varied from 2ae9% for subjects over 50 years to 9ae1% for subjects below 15 years (table 1) . vaccination with the pandemic strain was the highest in subjects below 15 (16%) whereas vaccination with the seasonal strain was the highest in subjects over 50 (44ae9%). detailed data regarding vaccination is given in table 1 . this study confirms previous findings that age, pandemic influenza vaccination, and history of ili are associated with elevated post-seasonal gmt. [8] [9] [10] [11] [12] [13] among non-vaccinated subjects, elevated gmt in the elderly may be the result of exposure to similar viruses in early life, whereas children and young adults with elevated gmt are likely to have been infected by the 2009 a ⁄ h1n1v virus. [13] [14] [15] [16] interestingly, a significant drop in the hai titer is observed during the months following vaccination with the pandemic strain. this study also shows that seasonal influenza vaccination may have contributed to an increase of the hai titer, especially in the elderly. the reason for this association is not obvious: although we cannot discard the hypothesis of a higher incidence of a ⁄ h1n1v infections in seasonal vaccine recipients, as described by several other studies, 17-19 the main explanation may be a cross-reaction between pandemic and seasonal strains. 16, 20, 21 further analyses in this cohort are needed to confirm and explain these first results. the follow-up of subjects involved in the copanflu-france cohort will provide data to study the risk factors for infection by the influenza virus. in april 2009, the cdc alerted about the appearance of a new strain of ia h1n1 with unknown virulence. infants under 5 years old had higher risks of hospitalization, complications, and rate of death for sari. materials and methods: a cross-sectional study was executed from may to december in 2009. the sources were: mandatory reporting form of the province surveillance system, databases of the hospital management information system, clinical pictures reviews, and telephone daily medical reports. inclusion criteria: children under 5 years old with diagnostic of ili or sari and confirmed cases with epidemiological nexus or laboratory confirmation (rrt-pcr, ifi). the age specific mortality rates were calculated with an estimated population for the province according to the national statistics and census institution. results: the ili rate in infants under 5 years old was 1295ae19 ⁄ 100 000 people (95% ci 1236-1356) being higher in infants of 4 years old (1569 ⁄ 100 000 people of 4 years (95% ci 1426-1722) ( table 1) . infants had less risk of getting sick in relation to the rest of the population (rr 0ae40 [95% ci 0ae33-0ae47]) (p < 0ae05). the chance of sari in infants was 2ae89 (95% ci 2ae46-3ae39) compared to the rest of the population. the lethality rate was higher in infants under 1 year old (14 ⁄ 100 000 people [4 ⁄ 28 364]). discussion: the evidence suggests that the infants under 5 years old had lower risk of getting sick than the rest of the population, but had higher risk of sari if they had some past illness. the highest lethality rate was presented in infants under 1 year old. non-medical interventions had an important role in the epidemic containment for not having a specific vaccination available. as this age group had high risks of hospitalization, it would be advisable to prioritize their vaccination. in april 2009, the cdc alerted about the appearance of a new strain of ia h1n1 with unknown dissemination and virulence. in june, the world health organization declared the pandemic. 1, 2 the ili often presents an unspecific clinical picture in infants under 5 years old, from mild symptoms to sari, especially in the newborn babies. infants under 5 years old have higher risks of hospitalization, complications, and rate of death for sari. 3, 4 on may 7th, argentina declared the first imported case of ia h1n1, and by the end of the month, it announced the viral circulation in the country. the epidemiological surveillance system of the province arranged that all the patients with influenza diagnosis made by a doctor must be reported. from april 24th to november 14th, 20 212 suspected cases of ili in the province of tucumán were reported. the ili rate was 1353 ⁄ 100 000 people, and ia h1n1 comprised 40 ⁄ 100 000 people. the lethal rate of sari ia h1n1 was 306ae7 ⁄ 100 000 people (62 ⁄ 20 212). 5 the objective of this research was to determine the epidemiological characteristics of the pandemic ia h1n1 in infants under 5 years old in the province of tucumán between may and december in 2009. the province of tucumán is placed in the center of the northwest of the republic of argentina. it has a population of 1 493 488 inhabitants of which 138 821 are infants under 5 years old. the crude birth rate for 2008 was 19ae9&. the infant mortality rate was 13ae8&. respiratory pathologies in infants under 5 years old were the third cause of death in the province (12%). the public health system of the province is composed by three sectors: public, private, and welfare. with 91 health facilities as a total, the average of available beds is 3& per inhabitants and 6& per neonates. a cross-sectional study was executed from may to december in 2009 in the province of tucumán, argentina. the following sources were used: mandatory reporting form of the surveillance system of the province filled by a doctor, databases of the hospital management information system, clinical pictures reviews, and telephone daily medical reports (patients with sari). inclusion criteria: • suspected case of ili: sudden appearance of fever higher than 38°c, cough, or sore throat. it may or may not be accompanied by asthenia, myalgia or prostration, nausea or vomiting, rhinorrhea, conjunctivitis, adenopathy, or diarrhea. ) were used for the analysis. the odds rations, risk ratio and 95% confidence interval were calculated to compare ambulatory with hospitalized patients, confirmed and dismissed, <5 years old and the rest of the population. it was considered significant a rate of p < 0ae05. the age specific mortality rates were calculated with an estimated population for the province according to the national statistics and census institution. the epidemiological surveillance system of the province received 21 012 ili reports, 11ae77% (2474 ⁄ 21 012) were infants under 5 years old. twenty seven percent were dismissed (676 ⁄ 2474), and 34% (623 ⁄ 1798) of suspected cases were confirmed. the first ia h1n1 case was a child of 2 years from the province of buenos aires, in 22th epidemiological week, and the last suspected case was reported in october 26, 2009 ( figure 1 ). the ili rate in infants under 5 years old was 1295ae19 ⁄ 100 000 people (95% ci 1236-1356), being higher in infants of 4 years old (1569 ⁄ 100 000 people of 4 years, [95%ci 1426-1722]). the higher ili rates in confirmed the pandemic of ia h1n1 (2009) was detected for the first time in the province of tucumán. the evidence suggests that infants under 5 years old had lower risk of getting sick than the rest of the population (protective factor), but had higher risk of sari if they had some past illness. the highest lethality rate was presented in infants under 1 year old. towns with the highest demographic density had superior proportion of cases. non-medical interventions had an important role in the epidemic containment for not having a specific vaccination available. as this age group had high risks of hospitalization, it would be advisable to prioritize their vaccination. outbreak of h1n1 influenza -2009: behavior of influenza h1n1 in school children in the province of tucumá n, argentina criteria: patients treated with antiviral medication for prophylaxis, respiratory pathologies which did not justify specific medication, and incomplete forms. results: from all notifications, 6342 were cases of ili in the group aged 5-17 years old; 53% were males. the incidence rate in this group was 17ae4 per thousands of inhabitants. the 13% of laboratory samples were influenza a h1n1, 34% were confirmed as unspecific influenza, and 49% were dismissed. the school aged children group had a high risks of getting sick (r.r. 1ae78 [95% c.i. 1ae55-2ae04]), especially males. it appeared that school aged children had a protective factor for presenting sari (or 0ae66 [95% c.i. 0ae66-0ae89], p < 0ae05). the lethality rate in this group was 9ae32 ⁄ 10 thousands. headaches, myalgia, coryza, and sore throat were very common and significantly different (p < 0ae05) than the rest of the population. it was reported a decrease in the ew 28 coinciding with winter holidays (ew 27). the epidemic curve was different in males compared to females during the winter holidays. discussion: school aged children got sick more than the rest of the population, although they presented less proportions of sari. however, comorbidities were decisive in order to present sari or death. the epidemic curve was different in males compared to females. through its analysis, the beneficial effect of school closure was observed, as long as children meet the recommendation to stay home. in april 2009, different countries reported cases of influenza a h1n1; mexico reported a high mortality rate associates with this disease. 1 the world health organization (who) declared the phase 6 influenza pandemic alert on june 11. 2 several reports from different countries describe the behavior of the pandemic in school aged children. this group plays an important role in the transmission of influenza. in germany, during the summer peak, pandemic hardly spread within this group. this might be explained by the timing of the summer school holidays, which started between ew 27 and 31. since mid october, after the autumn holidays, the school-aged children began to be more affected, and the proportion increased from 16% in the initiation period to 43ae8% in the acceleration period. 3 in australia, 55% of h1n1 cases were school aged children (5-17 years), with a median age of 16 years (29% of cases were aged 13-17 years and, and 26% between 5-12 years). 4 in canada, the infection rate was highest in this group. 5 in chile, the incidence rate was 4500 ⁄ 100 000 inhabitants, although in general they had mild desease. 6 school closure can operate as a proactive measure, aimed at reducing transmission in the school and spread into the wider community, or reactive, when the high levels of absenteeism among students and staff make it impractical to continue classes. the main health benefit of proactive school closure comes from slowing down the spread of an outbreak within a given area and, thus, flattening the peak of infections. this benefit becomes especially important when the number of people requiring medical care threatens to saturate health care capacity. it has its greatest benefits when schools are closed very early in an outbreak, before 1% of the population falls ill. school closure can reduce the demand for health care by an estimated 30-50% at the peak of the pandemic under ideal conditions, but too late in the course of a community-wide outbreak, the resulting reduction in transmission is likely to be very limited. policies for school closure need to include measures that limit contact among students when they are not in school. 7 tucumán is placed in northwest argentina and has a total area of 22 524 km 2 . the population (2001 census, projection 2009) was 1 493 488 inhabitants; of wich 421 638 were 5-19 years old. the health system of the province is composed of 3 sectors: public, private, and welfare. it has a total of 91 health facilities with internement available and an average of 3 & inhabitants. influenza-like illness (ili) has seasonal and endemic behavior in this province, as evidenced by past records from the national health surveillance system and influenza sentinel surveillance unit of the province. an increase of ili was reported in 2009, with a peak in the ew 28. the objectives were: general objective to describe the behavior of the influenza a h1n1 2009 epidemic in school aged children from the province of tucumán, argentina. specific objectives • to explore the response to preventive measures by school aged population. • to assess the effect of the suspension of classes in this group. • to estimate the magnitude and severity of the disease. • to observe the effect of co-morbidities in this group. a cross-sectional study was executed from may to december 2010. data were gathered through mandatory reporting forms, wich were collected from all public and private health centers. inclusion criteria: patients with compatible symptoms with influenza a; school aged children 5-17 years old. exclusion criteria: patients treated with antiviral medication as prophylaxis, respiratory pathologies which did not justify specific antiviral medication, and incomplete forms. • suspected case of ili: cases considered by clinical criteria (fever higher than 38°c, cough or sore throat. it may or may not be accompanied by asthenia, myalgia or prostration, nauseas or vomiting, rhinorrhea, conjunctivitis, adenopathy, or diarrhea). • confirmed case: person with positive laboratory results for influenza a h1n1 or unspecificed influenza a (by laboratory results through rrt-pcr or immunofluorescence techniques). • dismissed case: by negative or different laboratory results, or different clinical evolution. • comorbidities: chronic illnesses like arterial hypertension, diabetes, asthma, recurrent obstructive bronchial syndrome (robs), smoking, chronic obstructive pulmonary disease (copd), immunosuppression, hiv ⁄ aids, cancer, nephropathy, obesity; pregnancy was also considered. data were analyzed using epi2000 software (epi infoô cdc, atlanta, eeuu). rates were calculated and rr was estimated with their respective confidence interval (ci). population data were taken from 2001 national census projections. an estimation based on the same census was used for the group between 5 and 17 years old. to observe the effects of other co-variables, the or and their ci were calculated. logistic regression was used to evaluate the influence of the comorbidities. x 2 was used to compare proportions. respiratory samples (nasopharyngeal and faryngeal swabs) were obtained. they were analyzed at influenza sentinel surveillance unit of tucumán, and ⁄ or sent to national reference laboratory dr. c. malbrán (rt-pcr). from all notifications (20 212), 6342 were cases of ili in the group aged between 5 and 17 years old, 53% (3340 ⁄ 6342) of which were males. the incidence rate was 17ae4, and it differed according to the sexes: 18ae0 males and 16ae7 females per thousands of inhabitants (p < 0ae05). of all laboratory samples (370) 13% were confirmed as influenza h1n1, 34% were confirmed as unspecificied influenza, and 49% were dismissed. the remaining percentage corresponded to the isolation of other viruses (parainfluenza, respiratory syncytial virus, and adenovirus). the school aged group had higher risk of getting sick, in relation to the rest of the population (rr 1ae78 [95% ci 1ae55-2ae04]), especially males (rr 1ae97) compared with females (rr 1ae66). the highest attack rate was observed in the capital of tucumán (42 ⁄ 1000 inhabitants). according to the rest of the population, it looked like being school aged children meant a protective factor for presenting sari (severe acute respiratory infection) (or 0ae66 [95% ci 0ae66-0ae89], p < 0ae05). the lethality rate was 9ae32 ⁄ 10 thousand. the risk of dying was low compared to other ages. persons with comorbidities had significantly higher risk of presenting sari (or 1ae8 [95% ci 1ae35-2ae58], p < 0ae05) and of dying (or 8ae1 [95% ci 19ae6-3ae34], p < 0ae05). respiratory comorbidities were the most frequent: asthma 2ae5% (162 ⁄ 6342) and 2% rors (122 ⁄ 6342). the symptoms headaches, myalgia, coryza, and sore throat were very common and significantly different (p < 0ae05) than the rest of the population. if we compared the group aged 5-17 years with 18-39 years old, the epidemic curve of the first group showed a decrease in the ew 28, coinciding with winter holidays (ew 27) (figure 1 ). there was a slight increase in the tendency when classes began, but it showed a clear declination afterwards. the analysis of rates in school aged children by ew showed a reduction of 22ae2% in males and 26ae6% in females (p < 0ae05) at ew 28. however, after the first week of winter holidays, the curve in males had a significant increased to 65ae6% compared to ew 27, reaching the highest weekly rate of the epidemic (18 ⁄ 10 000 inhabitants). the reopening of classes coincided with a significant decrease of the rate (34ae9%), from 18 to 11ae7 ⁄ 10 000 inhabitants in ew 31 (p < 0ae05). in females, the school closure coincided with a plateau-shaped curve, and the reopening with a significant decrease of 43ae4% of the rate, from 12ae1 to 6ae9 in ew 32 ( figure 2 ). the school children got sick a lot more than the rest of the population, although they presented less proportions of sari. however, comorbidities were determined in order to present sari or death. symptoms like headache, myalgia, coryza, and sore throat were considered more conducting for the definition of cases in this population in tucumán. the epidemic curve was different in males compared to females during the winter holidays. the beneficial effect of school closure was observed as long as persons met the recommendations. the difference between males compared to females during winter holidays could mean that women would have carried out social distance recommendations much better, for example, remained at home. the significant reduction after the opening of classes is a factor to be considered as an effective intervention in the declining stage of the curve. here, we report pdmh1n1 infection attack rate (iar) during the first wave of the pandemic. we used our iar estimates to infer the severity of the pandemic strain, including the age-specific proportion of infections that led to laboratory confirmation, hospitalization, intensive care unit (icu) admission, and death. [1] [2] [3] [4] part of these results are now available in ref. 5 subjects of a community study, 5-14 years old between 1 november 2008 and 31 october 2009, we conducted a cohort study of pediatric seasonal influenza vaccination and household transmission of influenza. one hundred fifty-one children aged 5-14 were recruited and provided baseline sera in november and december 2008. between september and december 2009 a further 766 children aged 5-14 were recruited and provided baseline sera for the second phase of the study. for this serologic survey, we tested the 151 sera collected before the first wave and the 766 sera collected after the first pandemic wave. written informed consent was obtained from all participants. parental consent was obtained for participants aged 15 or younger, and children between the ages of 8 and 15 gave written assent. all study protocols were approved by the institutional review board of the university of hong kong ⁄ hospital authority hong kong west cluster. age-stratified data on virologically confirmed outpatient consultations, hospitalizations, icu admissions, and deaths associated with pdmh1n1 from 29 april 2009 to 15 november 2009 were provided by the hong kong hospital authority (the e-flu database). 6 since may 2009, patients admitted with acute respiratory illnesses routinely underwent laboratory testing for pdmh1n1 virus by molecular methods. sera were tested for antibody responses to a ⁄ california ⁄ 4 ⁄ 2009 by viral microneutralization (mn). 7 most individuals infected with influenza develop antibody titers ‡1:40 by viral microneutralization after recovery. 8 we defined the pdmh1n1 seroprevalence rate as the proportion of individuals who had antibody titers ‡1:40. while mn antibody titers of ‡40 are not by themselves conclusive evidence for pdmh1n1 infection, we have assumed that the increase in cross-sectional seroprevalence between the pre-and post-first wave time periods are evidence of recent pdmhn1 infection. the iar was defined as the proportion of individuals infected by pdmh1n1 during the first wave. the case-confirmation rate (ccr), case-hospitalization rate (chr), case-icu-admission rate (cir), and case-fatality rate (cfr) were defined as the proportion of pdmh1n1 infections that led to laboratory-confirmation, hospitalization, icu admission, and death. due to containment efforts until june 29, 2009 all laboratory-confirmed cases were required to be hospitalized for isolation regardless of disease severity. as such, only surveillance data from june 30 onwards were used to estimate severity measures. we estimated the iar as the difference between the prefirst-wave and post-first-wave seroprevalence rate. we used the estimated iar as the denominator for calculating the ccr, chr, cir, and cfr. we used an age-structured sir model with 5 age classes (0-12, 13-19, 20-29, 30-59, and ‡60) to describe the transmission dynamics of pdmh1n1 in hong kong between 10 june and 15 november 2009. we assumed that the mean generation time was 2ae3 days. using the age-structured transmission model, we estimated the following transmission parameters from the serial cross-sectional serologic and hospitalization data: (i) r o , the basic reproductive number; (ii) p 1 and p 2 , the reduction in within-age-group transmission for 0-12 and 13-19 years old during summer vacation (compared to school days during september-december 2009); (iii) d r , the average time for neutralization antibodies titer to reach ‡1:40 after recovering from infection; (iv) h a , the age-specific relative susceptibility with 20-29 years old adults as the reference group. we assumed non-informative priors for all parameters and used monte carlo markov chain methods to obtain posterior distributions of the parameters. sources of specimens: [1] pediatric cohort study (2-29 april 2009 virological surveillance data suggested that the first wave of pdmh1n1 in hong kong occurred from august to october 2009. most of the laboratory-confirmed infections in this first wave occurred in individuals aged below 25 years old accounting for >72% of the lab-confirmed cases and hospitalizations, 32% of icu admissions, and 6% of deaths. taking into account a delay of 2-3 weeks for antibody titers to appear during convalescence, 8 we found that these virological surveillance data were consistent with our serial cross-sectional seroprevalence data, which indicated a sharp rise in seroprevalence among the 5-25 years old from september to november and a plateau thereafter (data not shown). among individuals aged 5-14 years, the seroprevalence rates were similar across time between pediatric outpatient subjects and pediatric cohort study subjects (data not shown). similarly, for older age groups, the seroprevalence rates were largely similar between blood donor subjects and hospital outpatient subjects (except for the 20-29 years old in november-december). this provided some evidence that despite biases in our convenience sampling scheme, the resulting serologic data provided a reasonably representative description of seroprevalence in the community. the estimated pre-and post-first-wave seroprevalence rates and the corresponding iar estimates are shown in table 1 . the severity estimates (ccr, chr, cir, and cfr) are shown in table 2 . in summary, we estimated the iar was 43ae4% among 5-14 years old, 15ae8% among 15-19 years old, 11ae8% among 20-29 years old, 4ae3% among 30-39 years old, 4ae6% among 40-49 years old, and 4ae0% among 50-59 years old. overall, we estimated a population-weighted iar of 10ae7% (9-12%) among individuals aged 5-59 years through the first wave in hong kong. ccr were around 2ae8-5ae4% among the 5-59 years old. chr were around 0ae47-0ae87% among the 5-59 years old. cir increased from 7ae9 (5ae2-12ae6) per 100 000 infections in 5-14 years old to 75 (32ae7-281) per 100 000 infections in 50-59 years old. cfr followed a similar trend with 0ae4 (0ae1-2ae3) death per 100 000 infections in 5-14 years old to 26ae5 (10ae4-109) deaths per 100 000 infections in 50-59 years old. compared to children aged 5-14, adults aged 50-59 were 9ae5 and 66 times more likely to be admitted to icu and die if infected. the best-fit age-structured transmission model gave the following parameter estimates: 1. the basic reproductive number was 1ae38 (95%ci, 1ae36-1ae41). 2. it took an average of 14 (9-21) days for recovered individuals to develop neutralization antibody titer ‡1:40. table 2 . estimated age-specific proportions of individuals with pdmh1n1 infections that were laboratory-confirmed, were hospitalized, were admitted to icu, and died. case-icu and case-fatality rates are expressed as number of episodes per 100 000 infections 3. compared to 20-29 years old, 0-12 years old children and 13-19 teenagers were 3ae7 (3ae2-4ae5) and 1ae6 (1ae3-2) times more susceptible to pdmh1n1 infection, respectively. 4. compared to 20-29 years old, 30-59 years old older adults and 60-79 years old elderly were only 0ae42 (0ae3-0ae6) and 0ae33 (0ae18-1ae5) times as susceptible as the 20-29 years old, respectively. 5. compared to the school period during september-december 2009, summer vacation reduced within-agegroup transmission by 61% (53-72%) among 0-12 years old, but only 12% (3-18%) among 13-19 years old. using computer simulations, we estimated that if preexisting seroprevalence is zero, real-time serologic monitoring with about 1000 specimens per week would allow accurate estimates of iar and severity as soon as the true iar has reached 2% (data not shown). we estimated that during the first wave in hong kong, 43ae4% of school-age children and 10ae7% of individuals aged 5-59 were infected by pdmh1n1. a serologic survey in england found similar iars in london and the west midlands. 8 both studies highlight the importance of including serologic surveys in pandemic surveillance. the geographically compact and well-mixed population in the urban environment of hong kong permits some degree of confidence in the validity of our iar and severity estimates. the completeness of the pdmh1n1 surveillance system, welldefined population denominator, and our large-scale serologic survey provide accurate numerators and denominators for the severity measures. we based severity estimates for pdmh1n1 on the iar as the denominator. in most previous studies of pdmh1n1 severity, the denominator was clinical illness attack rate, which depends on the probability of symptoms as well as medical care seeking behavior of the population. 2, 9 our estimated cirs and cfrs are broadly consistent with presanis et al.'s 2 'approach 2' severity estimates, but around 7-9 times lower than their 'approach 1' estimates. our estimates of chr are 2-10 times higher than their approach 2 estimates of symptomatic chr. however, the hospitalization-death ratio was 4253 ⁄ 27 = 164 as of november 15 in hong kong, but 996 ⁄ 53 = 19 as of june 14 in new york, 2 suggesting that the clinical threshold for admission in terms of disease severity at presentation may have been lower in hong kong. our study has a number of limitations. first, we have used antibody titers of ‡1:40 by viral microneutralization as an indicator of recent infection, correcting for pre-existing seroprevalence levels, but this may lead to underestima-tion of the iar if some infections led to antibody titers <1:40, or if some individuals with baseline titers ‡1:40 were infected. second, our estimates of the iar would be biased upwards if infection with other circulating influenza viruses led to cross-reactive antibody responses resulting in antibody titers ‡1:40. however between august and october 2009, 83% of influenza a viruses detected in hong kong were pdmh1n1, and only 3% of isolated viruses were seasonal h1n1 viruses. 10 third, a minority of severe illnesses associated with pdmh1n1 infection might not be identified by molecular detection methods, for example if admission occurred after viral shedding from the primary infection has ceased, in which case we may have underestimated the disease burden of pdmh1n1. finally, our analyses are primarily based on seroprevalence among blood donors to the hong kong red cross, who may not be representative of the whole population. we do not have detailed data on donors to compare their risk of infection with the general population, but we did observe very similar seroprevalence rates across the three groups of subjects in our study, i.e., blood donors, hospital outpatients and participants in a community cohort (data not shown). in conclusion, around 10ae7% of the population aged 5-59 and half of all school-age children in hong kong were infected during the first wave of pandemic h1n1. compared to school-children aged 5-14, older adults aged 50-59, though less likely to acquire infection, had 9ae5 and 66 times higher risk of icu-admission and death if infected. thus, although the iar of pdmh1n1 is similar to that of a seasonal epidemic, the apparently low morbidity and mortality of 2009 pandemic influenza (h1n1) appears to be due to low infection rates in older adults who had a much greater risk of severe illness if infected. the reasons why older adults appear relatively resistant to pdmh1n1 infection even though they appear to lack neutralizing antibody remains unclear. if antigenic drift or other adaptation of the pdmh1n1 virus allows these older age groups to be infected more efficiently, the morbidity and mortality of subsequent waves of the pandemic could yet become substantial. and the national institute of allergy and infectious diseases, national institutes of health (contract no. hhsn266200700005c; adb no. n01-ai-70005). the funding bodies had no role in study design, data collection and analysis, preparation of the manuscript, or the decision to publish. bjc reports receiving research funding from medimmune inc., a manufacturer of influenza vaccines. the authors report no other conflicts of interest. some data presented in this manuscript were previously published in wu et al. 5 it is well known that a primary goal of vaccination is to generate immunological memory against the targeted antigen to prevent disease in a vaccinated person. this ensures an accelerated immune response in the event of future contact with the pathogenic agent, such as a virus. therefore, it is very important to develop criteria for the assessment of vaccine immunogenicity by measuring both t and b memory cell levels from the vaccinated host. in contrast to inactivated influenza vaccines, live attenuated influenza vaccines (laivs) have been shown to provide primarily cellular and local immune responses. 1-3 to date, however, the hemagglutination-inhibition (hai) test (i.e. detection of serum antibodies) remains the method widely accepted for evaluation of an influenza vaccine's immunogenicity. improved understanding of the role of cellular and mucosal immunity and their contribution to protecting against severe illness caused by influenza infection has emphasized the need to reconsider methodologies used to evaluate the immunogenic impact of various influenza vaccines. such new assays need to include methods to measure local antibodies and virus-specific lymphocytes, especially in the case of live attenuated influenza vaccines, because of their potential to induce such broad-based immune responses. 4 the aim of this study was to assess the ability of new russian pandemic laivs a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92 (h5n2) ('ultragrivak,' registered 25ae03ae2009) and a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) ('influvir,' registered 13ae10ae2009) to induce memory t-cells in naïve human subjects and to compare results to levels of hai antibodies from each subject. a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92 (h5n2) laiv was generated by 7:1 genetic reassortment of low-pathogenic avian influenza virus a ⁄ duck ⁄ potsdam ⁄ 1406-86(h5n2) and master donor strain a ⁄ leningrad ⁄ 134 ⁄ 17 ⁄ 57 (h2n2). 5, 6 the vaccine strain contains ha gene from avian virus, as well as na and internal genes from the master donor virus. a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) laiv was generated by classical (6:2) reassortment of a ⁄ california ⁄ 07 ⁄ 2009 (h1n1) with the master donor virus. 7 the vaccine strain contains ha and na genes from a 'wild-type' h1n1 strain and internal genes from the master donor virus. participants were aged 18 to 20 years and were without contra-indication of laiv vaccination. immunogenicity of a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92 (h5n2) laiv was assessed in ten vaccinated persons and ten volunteers inoculated with a placebo (sterile physiological saline solution). immunogenicity of a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) laiv was estimated in 16 vaccinated volunteers and nine volunteers inoculated with placebo. viruses or placebo were administered intranasally twice with an interval period of 21 days at a dosage of 0ae25 ml per nostril for each vaccination. physical examination, venous blood and nasal swab samples were collected at four time points during the study: (i) before vaccination (day 0); (ii) 21 days after first vaccination (day 21); (iii) 21 days after the second vaccination (day 42); and (iv) 6 weeks after the second vaccination (day 63). serum hai antibodies were measured by standard hai assay 8 using 1% human red blood cells. test antigens for the assay were a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92 (h5n2) or a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) to match the appropriate vaccine antigen. local iga antibodies in nasal swabs were evaluated by elisa 9 using whole purified a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92 (h5n2) or a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) viruses at 16 hau per 0ae05 ml for absorption to elisa plates. endpoint elisa titers were expressed as the highest dilution of sera that gave an optical density (od) greater than twice the mean od of six negative controls in the same assay. percentages of virus-specific cd3 + cd8 + ifn-c + and cd3 + cd4 + ifn-c + peripheral blood memory cells were determined using a flow cytometry iccs assay performed by the published method. 3 pbmcs were prepared with standard histopaque-1077 gradient centrifugation from heparinized whole blood. wilcoxon matched pair test, mann-whitney u test and the students t-test were used for statistical data analysis. prior to the first vaccination (day 0), gmts of hai antibodies to a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92 (h5n2) and a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) laivs were 1 ⁄ 5ae4 and 1 ⁄ 5ae7, respectively. in addition, gmts of siga against these specific antigens from nasal swabs were 1 ⁄ 5ae7 and 1 ⁄ 9ae1, respectively. no hai antibody titers greater than 1:40 were observed prior to vaccination. background levels of virusspecific t-cells varied significantly within groups. mean levels of virus-specific cd8 + ifnc + cells were 0ae151% to a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92 (h5n2) and 0ae173% to a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1). for cd4 + ifnc + cells, initial levels were 0ae240% and 0ae336%, respectively. thus, background levels of virus-specific antibodies were low, but prior vaccination or virus exposure in some volunteers produced some pre-existing levels of t cells, thus they were not absolutely immunologically naïve in this sense. preexistence of h5n1-crossreactive antibodies and t-cells has been observed previously. [10] [11] [12] effect of vaccination antibody immune responses both influenza a (h5n2) and influenza a (h1n1) laivs stimulated production of serum hai antibodies and local iga antibodies in nasal swabs. following the first vaccination with influenza a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92 (h5n2) laiv, 20% percent of volunteers exhibited seroconversion of hai antibodies; after the second vaccination, 30% of volunteers exhibited seroconversion. after the first vaccination, a 10% conversion rate of siga was observed; after the second vaccination, 60% showed conversions in levels of siga. the first vaccination with a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) laiv showed 6ae3% of hai antibodies seroconversions vaccination, and 50% seroconversion after second vaccination. for local siga, those results were 18ae8% and 31ae3% following the first and second inoculation, respectively. figure 1 summarizes cellular immune responses observed in the vaccinated versus the placebo group. after the influenza a (h5n2) laiv inoculation, significant differences in both cd4 and cd8 ifnc-producing t-cells were observed at day 42 after the second vaccination (d63). these data indicate that healthy young people who never received such avian influenza vaccines and were not exposed to h5n1 wild-type viruses were able to respond to the live attenuated h5n2 influenza vaccine. after the first influenza a (h1n1) laiv vaccination, reliable increases were observed in cd8 + cells only. after the second vaccination, increases in both cd4 + and cd8 + fold changes were significantly higher in vaccinated volunteers compared to the placebo group. it is noteworthy that cellular immune responses (cd4 + and cd8 + cells) were more marked in the a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1). considering the long-term circulation of h1-subtype viruses among humans in contrast to the novelty of h5viruses, such a result would be expected. similar data were also observed following vaccination with the h5n2 laiv. after first vaccination, the percent of people with notable increases in virus-specific cd4 + and cd8 + t-cells was 20% and 10% to h5n2 and 38% and 75% to h1n1, respectively. after the second vaccination, these results were 40% and 30% to h5n2 and 69% and 69% to h1n1, respectively. importantly, a significant number of vaccinated volunteers without remarkable increases ( ‡4-fold) in hai antibodies had notable increases in cd4 + and ⁄ or cd8 + memory cells. the percent of people with notable increases in virus-specific t cells after the second vaccination among hai()) volunteers was 40% and 75% to h5n2 and h1n1, respectively. these results indicate that laivs were able to induce broadly responsive, key antiviral immune responses that would not have been detected by the hai assay alone. thus, it can be deduced that hai data alone fails to reveal important broad and specific immune responses to laiv. consequently, the hai test alone is not suitable for assessment of laiv immunogenicity. furthermore, vaccination with h5n2 laiv was able to induce cross-reactive memory t-cells to a seasonal vaccine strain, a ⁄ 17 ⁄ solomon islands ⁄ 06 ⁄ 9 (h1n1) ( table 1) . reliable increases to a (h1n1) were observed in up to 20% of volunteers. there was an inverse dependence between levels of memory t cells before and after vaccination. authors are thankful to path for the financial support of these studies. we are also thankful to jessica d'amico and dr. rick bright for their editorial review. options for the control of influenza vii background: increased susceptibility of older populations to secondary bacterial pneumonia-like infections following influenza infection has been well documented. 1 recent evidence in mouse models suggests that this increased risk from secondary bacterial infection occurs through a desensitization of the innate immune response. 2 this recent finding, however, does not account for potential differences in immune responsiveness due to age. materials and methods: to address this parameter, we used three age groups (aged, adult, and young mice) to evaluate the role of age in influenza-mediated vulnerability to secondary bacterial challenge with pseudomonas aeruginosa. all mice were evaluated for multiple parameters including: (i) survival; (ii) lung bacterial load; (iii) total lung protein content; (iv) immune cell infiltration; (v) cytokine ⁄ chemokine expression; and (vi) toll-like receptor (tlr) rna expression profiles. results: prior challenge with influenza contributed to aberrant cytokine ⁄ chemokine profiles and increased lung cellular infiltrate in response to secondary bacterial infection across all age groups, supporting a critical role for influenza infection in the alteration of immune responses to other pathogens. also similar to human influenza, these changes were exacerbated by age in mice as demonstrated by increased bacterial load, mortality, and total lung protein content (an indicator of lung damage) after p. aeruginosa challenge. conclusions: these data support a potential role for virus-mediated and age-mediated alteration of innate immune effectors in the pathogenesis of influenza and the increased susceptibility of influenza virus infected mice to secondary bacterial infection. the understanding of the complex interaction of host and pathogen -and the role of age -in human influenza is critical in the development of novel therapeutics and improved vaccine approaches for influenza. our results support further examination of influenza-mediated alterations in innate immune responses in aged and non-aged animals to allow elucidation of the molecular mechanisms of influenza pathogenesis in humans. there is considerable evidence in the clinical literature to support the role of influenza infections with an enhanced risk for secondary bacterial pneumonias. [3] [4] [5] given the increased pneumonia-related morbidity and mortality in both the young and elderly populations, there is rationale for gaining a deeper understanding as to the systemic changes in the pulmonary microenvironment. although there are some recent reports that account for some of the molecular mechanisms at work in this disease process, 2 there is a paucity of experimental evidence that considers the potential effects of age. developmental changes in the immune system that occur in the aged environment have been well documented with regard to senescence of the adaptive immunity, global changes in myeloid cell function, and the establishment of a general pro-inflammatory state. 6, 7 the aim of this work was to provide evidence for the contribution of the aged immune environment to the pathology of influenza mediated secondary bacterial infections. animals used in this study were housed under conditions approved by tulane university's institutional animal use and care committee. female balb ⁄ c mice used in these studies were divided into three age groups: aged (18 months old), adult (6 months old), and young (2 months old). each age group was subdivided into two groups: influenza infected and naïve (control). mice were infected by the intranasal route with 4 · 10 5 pfu of mouse-adapted influenza a ⁄ pr ⁄ 8 ⁄ 34. clinical disease was measured by body weight changes over a 6 week period post influenza challenge, and recovery was determined as return to pre-infection weight. all mice were subsequently challenged intransally with1 · 10 7 cfu pseudomonas aeruginosa strain pao1. twenty-four hours post-pseudomonas challenge, bal with sterile pbs was performed on all mice in all groups. total rna from the cellular fraction was pooled from three experimental animals from each group. tlr mrna was detected by qrt-pcr, where expression levels were determined as relative to b-actin mrna levels. cdna was synthesized from total cellular rna from bal samples using iscript cdna synthesis kit (biorad). pcr reactions were composed of 0ae1 lg cdna forward and reverse primers according to optimized conditions and 12ae5 ll of 2 · syber green icycler supermix (biorad), in a total vol-ume of 25 ll and were run using a biorad icycler utilizing melting point determination. primers and concentrations used in this study included: mus_tlr2f: tgctttcct-gctggagattt-600 nm, mus_tlr2r: tgtaacgcaac agcttcagg-900 nm, mus_tlr3f: atatgcgcttcaa tccgttc-300 nm, mus_tlr3r: caggagcatactggt gctga-600 nm, mus_tlr4f: ggcagcaggtggaattg tat-600 nm, mus_tlr4r: aggccccagagttttgttc t-900 nm, mus_tlr5f: ctggggacccagtatgctaa-600 nm, mus_tlr5r: acagccgaagttccaagaga-900 nm, mus_tlr7f: ggagctctgtccttgagtgg-900 nm, mus_tlr7r: caaggcatgtcctaggtggt-600 nm, mus_ b-actinf: agccatgtacgtagccatcc-600 nm, mus_b-actinr: ctctcagctgtggtggtgaa-900 nm. as a measure of protein leakage into the alveolar space, total protein content in each bal was measured by bca assay of each supernatant fraction according to manufacturer's instructions (pierce). cytokine and chemokines levels were measured by multiplexed bead array (bioplex, biorad). immune cell characterization of bal was estimated by flow cytometry. lymphocyte populations were gated by forward versus side scatter and characterized as b cells (f4 ⁄ 80 ) , cd19 + ) or t cells (cd11b ) , cd4 + ). the myeloid population that is composed of macrophages, neutrophils, dendritic cells, and natural killer cells was enumerated by gating all but those found in the lymphocyte gate using forward versus side scatter plots. flow cytometry data was analyzed using flojo software (treestar). statistical analysis, where appropriate, was performed using a two-way analysis of variance (age versus influenza infection status) supported by bonferonni's correction for multiple comparisons. 8 a recent finding by didierlaurent, et al., 2 described an influenza mediated desensitization of tlr function as a primary contributor to an increase in bacterial burden when challenged after resolution of the primary influenza infection. this finding, however, was obtained using animals that were 6-8 weeks of age, where our study included two cohorts of older mice (6 months and 18 months). using whole protein content of the bal as an estimate of protein leakage into the lumen of the lung, we found elevated protein content in aged mice as compared to young and adult mice. in aged mice, a slightly lower total lung protein when comparing influenza infected to protein in the bal from influenza naïve mice challenged with p. aeruginosa (table 1) . supporting previously published studies showing a generalized pro-inflammatory cytokine environment in the aged immune system, we provide evidence for significantly (p = 0ae0465) and an increase in ifnc (p = 0ae0323) was detected. the decrease in gm-csf correlates well with a previous report that gm-csf is less prevalent in influenza resolved animals (table 1 ). 2 we also report a noticeable change in the immune cell populations with respect to b-cells, cd4 + t-cells, and the myeloid cell populations. there is a trend of increased prevalence in cd4 t-cells in the post-influenza environment across all ages. b-cell numbers also trend toward increase in influenza treated animals in young and adult animals; however, there is a noticeable decrease in the bcells in aged animals. across all age groups, there is a general decrease in frequency of cells that would normally make up the myeloid cellular fraction of the bal (macrophages, neutrophils, dendritic cells, and natural killer cells) ( table 1 ). our study also shows, as cited by others, that toll-like receptor (tlr) gene expression in the post-influenza environment is decreased in cells found in the bal 2 after both influenza and pseudomonas infection. our data support the previous finding of a reduced expression of tlr mrna in influenza-cleared mice when we measured tlr 2, 3, 5, and 7. only tlr4 showed differences with respect to age with young mice showing little or no detectable change in tlr4 mrna expression. our results show an increase in the expression across all tlrs examined in the aged mice group (table 1) irrespective of influenza infection status. these data support earlier studies performed with adult mice that showed reduced tlr mrna expression in the post-influenza environment. this study also expands the current understanding of the potential role of age in influenza mediated bacterial infection-induced mortality. the impact of these alterations in the immune microenvironment across age groups and infection status is highlighted by the ability of bacterially challenged animals to clear infection. assessment of bacterial load in the lungs of p. aeruginosa challenged mice indicated a difference in young and adult mice if previously infected with influenza virus. in aged mice, both influenza challenged and influenza-naïve mice had higher bacterial loads and less variability when comparing within the age group, supporting the risk of age alone in susceptibility to bacterial pneumonia (table 1, figure 1 ). taken together, these data support the potential role for both virus-mediated and age-mediated alteration of innate immune effectors in the pathogenesis of influenza and increased the susceptibility to secondary bacterial infection that results from influenza infection in mice. these findings highlight distinct differences in the immune environment between age groups and thus reveal necessity for further examination as to the mechanisms of immunity across age with respect to current infection status. garnering a clearer understanding as to the complex interaction of host and pathogen with respect to age in influenza infections is central to the development of increased efficacy in vaccine and therapeutic strategies. prospective estimation of the effective reproduction background pandemic influenza a (h1n1) virus (ph1n1) emerged in early 2009 and rapidly spread to every continent. an urgent priority for international and national public health authorities was to estimate the transmissibility of the pandemic strain for situational awareness and to permit calibration of mitigation strategies. the basic reproductive number, r 0 , is defined as the average number of secondary cases that 1 index case generates in a completely susceptible population, and is a common measure of transmissibility. however, it is difficult to estimate r 0 without an understanding of the degree of any pre-existing immunity in the population. the effective reproductive number, r, is defined as the average number of secondary cases that 1 index case generates, and can be estimated over time (i.e. r t ). wallinga and teunis 1 described a method to estimate r t based on illness onset dates of the cases while assuming that all secondary cases would have been detected, and cauchemez et al. 2 extended the method to permit prospective estimation by adjusting for secondary cases that have not yet experienced illness onset at the time of analysis. we describe how the method can further be extended to account for reporting delays, allowing true real-time estimation of r t during an epidemic, and we illustrate the methodology on notifications of ph1n1 and associated hospitalizations in hong kong. we obtained data on all laboratory-confirmed ph1n1 infections ('cases') reported between may 1 and november 15, 2009 to the hospital authority and center for health protection in hong kong collated in the eflu database. a subset of the cases was hospitalised. the database also included information on age, sex, illness onset date, laboratory confirmation date, and contact history (for the early cases). laboratory-confirmed ph1n1 infection was a notifiable condition throughout our study period. we extended existing methods for estimating r t over time to allow for reporting delays between illness onset and notification, and between illness onset, notification, and hospitalisation for those cases that were hospitalised, where the reporting delay distribution were estimated empirically from the data. 3 we further extended the methodology to allow for imported cases (infected outside hong kong) contributing to the estimation of r t as infectors but not infectees. we used multiple imputation to allow for missing data on some symptom onset dates to make best use of all available data. 4 we used a serial interval with mean (standard deviation) of 3ae2 (1ae3) days, 5 and in sensitivity analyses, we used serial intervals with mean 2ae6 days 6 and 3ae6 days. 7 statistical analyses were performed in r version 2.9.2 (r development core team, vienna, austria). in late april 2009 following the who global alert, hong kong initiated containment protocols to attempt to delay local transmission of ph1n1 for as long as possible. these measures included screening at ports, airports, and border crossings, and enhanced surveillance for people with influenza-like illness, particularly for those who had recently returned from abroad. laboratory testing capacity was substantial due to heavy investment in local infrastructure following previous experiences with avian influenza a ⁄ h5n1 in 1997 and severe acute respiratory syndrome in 2003. laboratory-confirmed ph1n1 cases were isolated until recovery, and their close contacts were placed under quarantine for 7 days. imported cases were identified sporadically through may and early june 2009. the first case of ph1n1 not traceable to importation (i.e. a local case) was identified on june 11 and triggered a change to mitigation phase measures. some containment measures, including isolation of cases, were continued until the end of june to allow a soft transition between containment and mitigation phases. as an immediate measure to try to reduce community transmission of ph1n1, all childcare centres, kindergartens, and primary schools were proactively closed for 14 days (subsequently extended for another 7-14 days to summer vacation in early july). 8 any secondary schools in which one or more confirmed ph1n1 case was identified were reactively closed for 7 days. on june 13 the government opened eight designated flu clinics across the territory to provide free medical consultation for outpatients with influenza-like illness and free laboratory testing for ph1n1. these clinics resumed regular chronic disease services in mid-august, and laboratory testing and antiviral treatment was restricted to high risk groups in september. the various interventions are highlighted in figure 1 (a), superimposed on the epidemic curve of laboratory-confirmed ph1n1 cases and ph1n1-associated hospitalizations. around 15% of the cases were hospitalised, and this proportion increased somewhat towards the end of the epidemic. 3 figure 1(b) shows the estimates of r t based on laboratory-confirmed ph1n1 cases. the estimated r t peaked at 1ae5 on june 12, and fell below 1 between 20 june and 3 july (which was within the school closure period). r t fluctuated between 0ae8 and 1ae3 through the school summer vacations in july and august, it subsequently increased to around 1ae2-1ae3 after schools reopened in september until the epidemic peaked in late september, and then fluctuated below 1 as the epidemic declined. the trends in r t based on h1n1-associated hospitalizations were similar, although with wider confidence intervals due to the smaller number of events ( figure 1c ). the extension of the methods to allow for reporting delays avoided substantial bias in realtime estimates of r during the epidemic for the most recent 7 days, and closely tracked the final estimates of r t . 3 our results suggest that ph1n1 may have had slightly lower transmissibility in hong kong than elsewhere. for example, estimates of r t were around 1ae5-2ae0 in new zealand 9 and australia. 10 lower transmissibility in hong kong has been associated with school closures in june and july followed by summer vacations from july through august. 8 furthermore, in hong kong the influenza virus usually does not circulate after august, 11 and therefore seasonality could also be a cause for the lower r t . on the other hand, the interventions applied during the mitigation phase, such as the widespread use of antiviral treatment in hong kong and the pre-existing immunity in the ageing population in hong kong, may also be associated with lower transmissibility. there are some limitations to our work. first, we only used aggregated data, and we did not consider the heterogeneity among the cases in terms of sex and age or other factors. therefore our estimates can only provide a snapshot of the overall trend, but limited information for any specific subset of population. secondly, we did not consider the possibility that cases might be infected in hong kong and exported to other countries, which could lead to slight underestimation of the transmissibility. one has to be careful in translating the estimated r t to the effectiveness of any specific interventions, as interventions may not be the only factor influencing the transmissibility; for example, a depletion of the susceptible population during an epidemic can also be a factor for the decline in r t . 12 in conclusion, real-time monitoring of the effective reproduction number is feasible and can provide useful information to public health authorities for situational awareness and planning. in affected regions, laboratory capacity was typically focused on more severe cases, and changes in laboratory testing and notification rates meant that that case counts may not necessarily reflect the underlying epidemic. a useful alternative to case-based surveillance is surveillance of the subset of severe infections, for example hospital admissions, or icu admissions, 13 and our results show that it was feasible to monitor ph1n1-associated admissions in real-time to estimate transmissibility. influenza antigenic cartography projects influenza antigens into a two or three dimensional map based on immunological datasets, such as hemagglutination inhibition and microneutralization assays. a robust antigenic cartography can facilitate influenza vaccine strain selection since the antigenic map can simplify data interpretation through intuitive antigenic map. however, antigenic cartography construction is not trivial due to the challenging features embedded in the immunological data, such as data incom-pleteness, high noises, and low reactors. to overcome these challenges, we developed a computational method, temporal matrix completion-multidimensional scaling (mc-mds), by adapting the low rank mc concept from the movie recommendation system in netflix and the mds method from geographic cartography construction. the application on h3n2 and 2009 pandemic h1n1 influenza a viruses demonstrates that temporal mc-mds is effective and efficient in constructing influenza antigenic cartography. the web sever is available at http://sysbio.cvm. msstate.edu/antigenmap. as a segmented, negative stranded rna virus, influenza virus is notorious for rapid mutations and reassortments. the mutations on the surface glycoproteins (ha and na) of influenza viruses are called antigenic drifts, and these antigenic drift events allow the virus to evade the accumulating immunity from previous infection or vaccination and lead to seasonal influenza epidemics. a reassortment event with a novel influenza antigen may result in antigenic shift and cause influenza pandemic. for instance, the 2009 h1n1 pandemic virus is a reassortant with a swine origin ha antigen. vaccination is the primary option for reducing the effect of influenza, and identification of the right vaccine strains is the key to development of an effective vaccination program. the antigenicity of an optimal vaccine strain should match that of the epidemic strain. in influenza surveillance program, the influenza antigenic variants are generally identified by the immunological tests, such as hemagglutination inhibition (hi) assay, microneutralization (mn) assay, or elisa. these immunological assays measure the antigenic diversity between influenza viruses by comparing the reaction titers among the test antigens and reference antisera. however, data interpretation of the data from these assays is not trivial due to the embedded challenges such as data incompleteness, high noises, and low reactors. by mimicking geographic cartography, influenza antigenic cartography projects influenza antigens into a two or three dimensional map using immunological datasets. 1 antigenic cartography can simplify the data interpretation, and thus, facilitate influenza antigenic variant identification. recently, we developed a novel computational method, temporal matrix completion-multidimensional scaling (mc-mds), in antigenic cartography construction. 2 in this paper, we described the details of temporal mc-mds, especially the original concepts introduced in this method, and how they can achieve the robustness in antigenic cartography construction. our method included two integrative steps: it first reconstructs the hi matrices using low rank mc method, and then generates antigenic cartography using mds with a temporal regularization. the mc concept was adapted from the movie recommendation system in netflix and the cartography concept from geographic cartography. in 2006, netflix, an online dvd and blu-ray disc rentalby-mail and video streaming company, held a 3-year netflix prize contest (http://www.netflixprize.com/) on computational methods for improving its recommendation system. 3 in its recommendation system, netflix collected the rating data from the individuals. based on his or her renting history and the ratings in the systems (e.g., from evaluators and other renters), netflix recommendation system suggests certain movies to a renter. apparently, no individuals would be feasible to provide ratings for all of the movies, as it will take hundreds of years for a single person to rate over 50 000 movies available from netflix. thus, the resulting rating data is an incomplete matrix, and it can be as sparse as less as 1%. 4 the challenge in netflix recommendation system is a classic mc problem. [4] [5] [6] [7] [8] as the inspiration of netflix prize contest, many efficient low rank mc algorithms were developed, for instance, opt-space, 7 svt, 5 cf, 9 bellkor, 10 pf, 11 and fwls. 12 eventually, the team bellkor's pragmatic chaos won this contest. their methods combines nonlinear probe blending and linear quiz blending to come up with a predictor bigchaos. 13 matrix completion estimates the unobserved values based on the observed values. the users can refill the missing data without repeating the experiments. furthermore, mc will help reduce the noises in the data, for instance, those biases by different individuals performing experiments. in influenza antigenic characterization, hi assay is a commonly used assay for antigenic analysis, since hi assay is relatively economic and easy to perform. however, hi is labor intensive, and it is almost impossible for any individual lab to complete the hi assays for all pairs of antigens and antisera during influenza surveillance. in addition, both testing antigens and the reference antisera are dynamic. for instance, in seasonal influenza surveillance, generally only contemporary antisera are used in experiments. thus, we will have to integrate multiple hi tables in order to evaluate the overall antigenic changes for influenza vaccine strain selection. the resulting hi tables will be incomplete, and the observed entries in the integrated hi data can be as less as 3%. the completion of this matrix can be formulated as a typical mc. briefly, given the combination of hi matrix with m antigens and n antisera, the hi matrix can be represented as m m·n = (m ij ) m·n , where m ij denotes the hi values from the reaction between testing antigen i and antiserum j. the low rank mc assumes that both antigen and antiserum can be embedded into a low rank space. to be specific, the low rank mc method is to seek matrix u m·r , v n·r and a diagonal matrix r r·r , where m = u m·r r r·r (v n·r ) t . in order to achieve this goal, the optimization formulation has been employed, which can be represent as following, where e denotes the observed entries in hi matrix and g(x) is a regularization function. the eqn (1) is the standard format of a low rank mc formulation. the geographic cartography is a common technique to display the cities and their geographic distances in a map. this cartography can be generated using mds based on a geographic distance matrix. figure 1(a) shows the antigenic cartography generated using a distance matrix with seven cities, and figure 1 (b) is a map for comparison. as an analog of geographic cartography, the influenza antigenic cartography maps the influenza antigens into a two or three dimensional map based on the distance matrix generated using immunological data. this incomplete matrix can be filled through mc algorithm discussed in section mc and netflix. low reactors, non-random date incompleteness, and temporal model generally, three types of data are present in a combined hi matrix: high reactor, low reactor, and missing values. among these three data types, high reactors are the most reliable data points. the low reactors are those values present in the hi matrix as ''equal to or less than a threshold h'', where h can be 5, 10, 20, or 40. low reactors have similar values in the affinity dataset but could be from different binding settings. these low reactors are present due to the detection limits of biotechnology, and they are not reliable. both these missing values and low reactors make it very difficult to analyze and interpret antigenic correlations amongst tested antigens and reference antigens. to our best knowledge, none of the existing mc method can handle the threshold values. in addition, the non-random incompleteness of influenza immunological datasets generates an additional challenge in traditional mc methods, which are based on the assumption that the observed values are randomly distributed among the matrix. in a typical combined antigenic hi data, most of the off-diagonal entries are missing values or low reactor values. 1 in order to overcome the above issues, we incorporated a regularization function into the eqn (1), where this indicator function is only valid for those entries with low reactor values. an alternating gradient decent method is applied to solve the optimization problem in eqn (2) . in addition, a temporal mds method is proposed to project the antigens into a 2 or 3 dimensional map. x where d ij is the average distance between virus i and virus j, t i is the isolation year of virus i, d ij is the distance between virus i and virus j in cartography, d ac i is the distance between virus a and center of group i, and d c i c j is the distance between the centers of group i and group j. all the parameters are tuned by cross validation. we named this method as temporal mc-mds. by applying temporal mc-mds method in an h3n2 dataset, 2 low reactors. figure 2 (a) is a three-dimensional influenza antigenic map based on this data by using mc-mds method. the reported 11 clusters (hk68, en72, vi75, tx77, bk79, si87, be89, be92, wu95, sy97, and fu02) were displayed in the core of a spiral s-shape, and bk79 and be92 are located at the turning point of this s-shape. however, the antigenic distances between some viruses are incorrect. for example, the distance between hk68 and fu02 in the projection is 7ae1223 units, which is close to the distance between hk68 and bk79 (6ae5113 units). the main reason leading to those inaccurate distances is the unique distribution of hi datasets described in section 2.3. in comparison, with the temporal model, not only the viruses in 11 clusters have been clearly separated, but also the antigenic distances between each cluster are proportional to their isolation time interval. in this updated cartography ( figure 2b ), the antigenic distance between hk68 and fu02 is 15ae0633 units, where the distance between hk68 and fu02 is 6ae3984 units. this result suggested that the temporal information is critical for antigenic cartography construction for immunological datasets spanning a long time period. the hi data from seasonal influenza surveillance belong to this category. for seasonal influenza virus ⁄ pandemic influenza viruses within a short time span, the temporal model is probably not necessary, as there is lack of long-term immunological pressure present in the population. figure 2 (c) is an antigenic cartography generated using a hi dataset with 2009 h1n1 influenza viruses spanning from april of 2009 to june of 2009. this map demonstrates that there is lack of antigenic drifts during the first wave of this pandemic influenza as all of these viruses are mixed altogether. our limited studies on h5 and h7 avian influenza viruses suggested the temporal model is not needed for avian influenza viruses. however, extensive studies are required to investigate whether there is any special data structure present in this type of data. in this study, we described in details the concepts and applications of new computational method, temporal mc-mds for influenza antigenic cartography construction. we formulate the influenza cartography as two integrative steps: low rank mc problem from the concept of netflix movie recommendation system and mds from geographic cartography construction. in order to handle two additional challenges, including low reactor and non random distribution of antigenic data, a temporal model is incorporated into mc-mds as temporal mc-mds. our applications demonstrated that temporal mc-mds is effective in constructing influenza antigenic cartography. the three dimensional antigenic cartography for a ⁄ h3n2 seasonal influenza virus without temporal model, and the antigenic clusters were defined in ref. [2] ; (b) the three dimensional antigenic cartography for a ⁄ h3n2 seasonal influenza virus with temporal model; (c) the two dimensional antigenic cartography for 2009 a ⁄ h1n1 pandemic influenza without temporal model, and these viruses were labeled in shape by the corresponding month for them to be detected. one grid is corresponding to a twofold change in hemagglutination inhibition experiment. the mechanisms driving the three waves of infection and mortality in the uk in 1918-1919 are uncertain. although the circulation of three distinct viruses could have generated three waves of infection, 1 the virological evidence required to prove or disprove this hypothesis is lacking. social distancing, an alternate mechanism for generating fluctuations in the effective susceptible pool and therefore explaining multiple waves of infection, 2, 3 was not generally imposed in the uk as it was in the us and australia. we are therefore motivated to explore the possible role of continual population-level changes in the average protective response against the circulating virus in generating a multi-wave pandemic, within a biologically motivated deterministic model for influenza transmission. the nature and duration of protection against further infection following recovery from influenza is uncertain and depends on the mode and tempo of viral evolution, as well as the response of the cellular and humoral arms of the adaptive immune system. 4 for a given seasonal ⁄ pandemic strain, memory b-cells may generate a specific antibody response in a portion of the adult ⁄ elderly population, depending on the exposure to related antigenic sub-types. 5 however neutralising antibodies are unlikely to be a widespread immunological response to a novel (pandemic) strain. memory t-cells which recognise conserved internal viral proteins may be a more common mechanism for protection; the generation of very high levels of cytotoxic cd8 + t-cells potentially facilitates rapid viral clearance, 6, 7 and lower levels of cd8 + t-cells perhaps provide partial protection. 8 in this work we explore key drivers of multi-wave pandemics within phenomenological models that incorporate different immune response mechanisms building on existing models 9,10 incorporating the role of evolving population-level protection in multi-wave pandemics. we use weekly reports of influenza mortality rates 11 for five administrative units in the uk (blackburn, leicester, newcastle, manchester and wigan) where records from block censuses instigated by local medical officers to record the cumulative incidence of reported symptoms in each wave in a sample of 1000 or more households are also available. 10 the symptom reporting data allows us to estimate the case fatality rate and thus use the mortality time series to constrain our transmission model. furthermore, the incidence of individuals reporting symptoms in multiple waves provides information about the acquisition and loss of immunity. we extract the death rate and symptomatic (re)infection rates predicted by our model prevalence for a given set of parameters and estimate a likelihood-based on a comparison to all the death and cumulative reported incidence data assuming a negative binomial error distribution. we utilise monte carlo markov chain (mcmc) methods with parallel tempering algorithms to maximise this likelihood and obtain parameter estimates. parallel tempering -which concurrently searches for maximal likelihood parameter solutions on a set of scaled likelihood surfaces -allows for relatively rapid exploration of the parameter space. we use bayesian information criteria (combined with qualitative assessment of biological plausibility) to aid model selection. we have implemented a deterministic compartmental transmission model, which allows for a variety of phenomenological modes of protection against the pandemic virus. to facilitate this, we stratify the population into two groups; the 'experienced' population (stratum 1) who have had been exposed to an influenza virus and the 'naive' population (stratum 2) who have not. in each stratum, i hosts may be classified as either susceptible s i , exposed e1 i and e2 i , having (recovered from) a symptomatic i i (r i ), or asymptomatic a i (ra i ) infection. note that the states tq i , tq2 i , e2 i , t i , and t2 i are included so that the hosts move between the key epidemiological states with a peaked (rather than exponential) distribution of waiting times. hosts in the experienced stratum may exhibit reduced susceptibility, infectiousness, and symptomatic proportion compared to naive hosts, parameterised by e i , e s , and e a , respectively; however note that depending on the model parameters, there may be fully susceptible hosts within the experienced stratum. in addition, we assume homogeneous population mixing and a constant basic reproduction number r 0 with the force of infection: modulated by a sinusoidal seasonal term with amplitude b 1 with phase chosen to maximise transmission in the winter season. here n is the total population size, and x e is the initial fraction in the experienced strata. the proportion of symptomatic cases a and the case fatality rate l are permitted to vary from wave to wave (and given indices 1, 2 or 3 accordingly). the transmission dynamics is described by the following set of coupled ordinary differential equations. where s in,1 = p utq2 i and s in,2 = 0 in order to divert recovered infectious hosts from the naive stratum into the experienced stratum. the probabilities of gaining permanent protection are q 1 = q and q 2 = 0. the latent exposed period is fixed to be c = 1 ⁄ 1ae3 days, and the rate of recovery is parameterised by m = 1 ⁄ t inf , where t inf is the infectious period. hosts with prior sterilising protection begin in q 1 and move into s 1 at rate u q = 3 ⁄ t wq . recovered hosts (r i ) migrate back to s 1 at a rate u = 3 ⁄ t w . the state p 1 contains hosts with permanent protection. the modes of protection captured in this model are: i. permanent prior protection (beginning in state p 1 ), ii. waning prior protection (beginning in state q 1 ), iii. permanent acquired protection with probability q (moving into state p 1 ), iv. waning acquired protection with probability 1 ) q, and, v. partial prior protection (beginning in state s 1 ) resulting in reduced infectiousness (e i ), susceptibility (e s ), and symptomatic proportion (e a ). in the context of this model, 'permanent' protection refers to protection which lasts for the duration of the epidemic. here we explore the results of parameter fitting to two models which differ in the nature of the assumed pre-existing protection in the community at the beginning of the pandemic. protection hypothesis 1 assumes that the prior protection is sterilising but temporary, whilst protection hypothesis 2 assumes that the prior protection is partial but permanent and may act on susceptibility, infectiousness, and ⁄ or asymptomatic proportion. each model allows waning acquired protection and for a proportion q of the experienced population to gain permanent protection following infection. fitted parameters common to each model are t inf , b 1 , q, t w , a, l and the proportion beginning in p x i . prior protection hypothesis 1: sterilising, waning prior protection we fix x e = 1 and fit for q 1 (t = 0) ⁄ n and t wq so that protective modes i, ii, iii, and iv are enabled ( figure 1 ). it is important to note that due to the slow convergence of the mcmc chains, we cannot guarantee that our parameter estimates correspond to the global minimum. furthermore, parameter estimates can only be meaningfully interpreted for good fits to the data. due to the prediction of a fourth (unobserved) wave for the model fit to blackburn, we do not report these parameter estimates here. the fits to the leicester data are generated with the parameter set r 0 = 5ae7, a 1 = 0ae25, a 2 = 0ae65, a 3 = 0ae65, t w = 0ae28years, t wq = 0ae45 years, we fix q 1 (t = 0) ⁄ n = 0 and fit for x e , e a , e i , and e s so that protective modes i, iii, iv, and v are enabled (figure 2) . the parameters corresponding to the fit in figure 2 for leicester are r 0 = 7ae4, a 1 = 0ae09, a 2 = 0ae50, a 3 = 0ae61, t w = 0ae22 years, p 1 (t = 0) ⁄ n = 0ae01, s 2 (t = 0) ⁄ n = 0ae48, b 1 = 0ae021, t inf = 0ae94 days, q = 0ae561, e a = 0ae9966, e i = 0ae946, and e s = 0ae594. our model with protection hypothesis 1 -which, similarly to the model discussed in ref. [9] , assumes that a sub-population has waning sterilising prior protection -is able to generate multiple waves of infection via the continual replenishment of s 1 from an initially large proportion (over 40%) of hosts with prior protection in q combined with the waning of acquired immunity in around 23% of cases on a time-scale of 3 months. disease severity as measured by symptomatic proportion increases from 25% in the first wave to above 60% for the second and third waves. over a quarter of the population are initially permanently immune, and a large r 0 value of 5ae7 drives transmission in the remaining population. protection hypothesis 2 -which assumes that prior protection offers partial susceptibility and ⁄ or reduced infectiousness or symptomatic disease -performs slightly more poorly; the fit to the leicester data has an inferior likelihood (although the mortality data only likelihood is a little larger), despite the higher dimensionality of the model. nevertheless, the model fit still mirrors many characteristics of the data, particularly for leicester. we note that for this model, a 1 is very near the lower limit, corresponding to ubiquitous exposure in the first wave. in this scenario, refuelling of the susceptible pool to generate secondary and tertiary waves is still possible due to a shorter waning time of acquired protection (well within 3 months) and a lower probability of gaining permanent protection following infection, when compared with the parameter estimate for hypothesis 1. the parameter estimates suggest that approximately 50% of the population initially experiences reduced disease severity (e a $ 0ae66), but similar susceptibility and infectiousness. a larger value for r 0 $ 7ae4 is required to drive transmission despite low numbers beginning in p 1 , due to the large number of hosts who acquire temporary or permanent immunity early on in the pandemic. it is clear that, at least mathematically and perhaps biologically, there are multiple possibilities for the structure of population-level protection which are compatible with the generation of multiple pandemic waves. however, whilst the models considered here are able to explain the observed mortality and reinfection data for some patterns of infection and mortality (e.g. leicester), they are not consistently able to reproduce a pandemic which dies out after three waves across the connected populations we are studying (e.g. for blackburn). it is challenging to construct a deterministic model for the spread of disease within multiple locations in the uk in 1918, which assumes homogeneous mixing without modulation of the transmission rate by social distancing. an improved model working with these assumptions likely requires a richer structure for the host protection response than the structures we have explored thus far. we are currently seeking improved fits to the data by implementing a number of biologically defensible exten-sions to our model, including incremental immunity whereby t w increases by a factor v after each exposure to the pandemic flu, and incremental loss of prior protection whereby a increases as hosts lose their sterilising prior protection. it is important to note that the mechanism(s) generating differences in the pandemic experience recorded in geographically connected locations is an open question; true differences in demography, varying degrees of reactive social distancing, inhomogeneities in the circulation (or circulation history, i.e. prior immunity) of viral strains, stochastic variations, and ⁄ or unique socio-cultural ⁄ behavioural conditions may all contribute to this effect. the 2009 h1n1 experience in australia and elsewhere highlighted the difficulties faced by public health authorities in diagnosing infections and delivering antiviral agents (e.g. oseltamivir) as treatment for cases and prophylaxis for contacts in a timely manner. consequently, forecasts from mathematical models of the possible benefits of widespread antiviral interventions were largely unmet. we summarise results from a recently developed model that includes realworld constraints, such as finite diagnostic and antiviral distribution capacities. we find that use of antiviral agents might be capable of containing or substantially mitigating an epidemic in only a small proportion of epidemic scenarios given australia's existing public health capacities. we then introduce a statistical model that, based on just three characteristics of a hypothetical outbreak [(i) the basic reproduction number, (ii) the reduction in infectiousness of cases governments and public health agencies worldwide, spurred by outbreaks of sars and h5n1, have developed preparedness strategies to mitigate the impact of emerging infectious diseases, including pandemic influenza. pandemic response plans are presently being revised in light of the 2009 h1n1 experience. [1] [2] [3] many developed countries amassed large stockpiles of neuraminidase inhibitors (nais) with the expectation that they could be used to not only treat the most severely ill, but curb transmission in the community. without relevant field experience indicating how nais should be distributed, mathematical and computational modelling has been used to inform optimal deployment policy in a pandemic scenario. 4-10 models of population transmission were used to infer likely effects on epidemic dynamics, using data from human and animal studies of experimental infection and nai efficacy trials. in the australian (and wider) context, models indicated the potential for substantial benefit at the population level if nais were distributed in a liberal manner, targeting close contacts of indentified cases. 11 furthermore, results indicated that use of limited nai resources in this way may improve the impact of case treatment due to the effects on epidemic dynamics. 11 however, these models did not take into account logistic and other real-world constraints, such as finite diagnostic and antiviral distribution capacities, which were identified as limiting factors during the australian 2009 h1n1 pandemic response. [12] [13] [14] in particular, if using positive pcr diagnosis as a 'decision to treat' test, delays to confirmation of diagnosis, particularly once total laboratory capacity was exceeded, prevented timely delivery of nais to both cases and contacts of cases. 12 in previous work, 15 we have extended our existing models to examine how diagnostic strategies [e.g. using pcr confirmation versus syndromic influenza-like illness (ili) presentation as a decision to treat], diagnostic-capacity, and nai distribution capacity each impact on the ability to deliver an effective intervention. the model uses case severity (the proportion of infections deemed severe) to determine the overall presentation proportion, and so the ability to identify individuals eligible for nai treatment and contact prophylaxis. figure 1 (a) shows a key result from the model. for each curve shown, we simulated thousands of epidemics, sam-pling across plausible ranges of parameters describing virus, population, and intervention characteristics using a latin hypercube sampling (lhs) approach. without intervention, the proportion of the population infected either symptomatically or subclinically by the end of the epidemic is around 50%. if a syndromic strategy (ili presentation) is used to determine provision of nais as treatment and prophylaxis, excessive distribution of drug to individuals who are not infected with influenza occurs early in the epidemic. early stockpile expiry accounts for a marginal impact of the antiviral intervention on the final outbreak size, in the order of a few percent. the second strategy modelled (pcr ⁄ syndromic) is one where pcr confirmation of diagnosis is required early in the epidemic to make treatment decisions until such time as laboratory capacity is exceeded. from this point, individuals are treated on the basis of symptoms alone -during an epidemic phase in which a substantial proportion of ili presentations will be attributable to influenza. under this strategy, the intervention is able to control the outbreak in approximately 10% of the simulated epidemics given the 'base case' constraints on diagnosis and delivery assumed in the model. the results highlight that a successful antiviral intervention requires a highly sensitive diagnostic strategy in the initial stages of the epidemic and comprehensive distribution of post-exposure prophylaxis. a pcr ⁄ syndromic strategy for decision to treat and provide contacts with prophylaxis is thus optimal. the surface in figure 1(b) shows the percentage of simulation runs for the pcr ⁄ syndromic strategy that have a final population attack rate of <10% (a substantial reduction from the no intervention case of approximately 50%) as a function of pcr capacity and nai daily distribution capacity. as indicated by the arrow, the estimated australian pcr laboratory capacity appears to be sufficient, while significant benefits for the public health outcome may be achieved if logistical delivery constraints for nai distribution can be ameliorated. however, the probability that such an interventioneven with substantial increases in pcr and nai distribution capacity -would successfully mitigate an epidemic is low (12-25%), and consequently it is difficult to universally recommend an antiviral intervention. 15 in this study, we introduce a statistical model that predicts whether or not an nai distribution strategy based on a pcr ⁄ syndromic antiviral distribution policy will be successful in mitigating an epidemic. we thereby provide proof-of-principle for the design of a decision support tool that may be used by public health policy makers during an epidemic when faced with formulation of context specific nai distribution policy. synthetic data of hypothetical outbreaks and interventions were generated using the lhs simulations developed in ref. [15] . we selected a random sample of 100 outbreaks from a total of 2000 simulated epidemics (5% of model simulations). using these data, we identified independent model parameters that were most highly rank-correlated with the final attack rate. these parameters were included in a logistic regression model to assess their ability to predict whether an influenza epidemic would be successfully mitigated by an antiviral intervention (ar < 10%). model predictions were then validated against the full simulated dataset. full details of the simulation model, its structure, parameterisation and parameter distributions are available in ref. [15] . use of the lhs simulation approach, and the method of model analysis and evaluation was similar to that previously described. 16 matlab 2010a (mathworks, natick, ma, usa) was used for the analysis and statistical model fitting. table 1 shows results from our logistic regression model. key parameters sufficient to predict whether or not an outbreak may be controlled by the deployment of av agents are: 1. r 0 , the basic reproductive number of the outbreak (assigned values between 1ae35 and 1ae45 for this example). as the value of r 0 increases, the epidemic progresses more rapidly and is more difficult to control, explaining the negative correlation coefficient. 2. e t , the relative infectiousness of treated individuals (assigned values between 0ae8 and 1ae0). higher values for this parameter indicate only modest drug effects on transmission, explaining the negative correlation coefficient. 3. g, the proportion of infections that are severe (assigned values between 0ae001 and 0ae1), and which in turn determines the presenting proportion (derived values between 0ae11 and 0ae56). as the presenting proportion increases, the ability to identify and treat cases and deliver prophylaxis to contacts also rises, increasing the impact of the antiviral intervention. the roc curve (1-specificity versus sensitivity, not shown) for the logistic regression model specified in table 1 has an area under the curve of 0ae937, demonstrating that the model predicts the success of an antiviral intervention extremely well. for example, with a sensitivity of 95% we still have a specificity of approximately 80%. evaluation of the 2009 pandemic response has emphasised the need for early informed decision-making to implement proportionate disease control measures. our model identifies a low probability of successful epidemic mitigation using targeted antivirals alone (figure 1 and ref. 15 ), in distinction to results from models that fail to account for the diagnosis and delivery constraints inherent in any public health response. the decision support tool (table 1) highlights key epidemic characteristics that are predictive of a high likelihood of effective mitigation. the reproduction number was one of the earliest parameters estimated from early outbreak data during the 2009 h1n1 outbreak. 17, 18 our findings reinforce the importance of characterising epidemic severity as early and as accurately as possible, in order to inform a proportionate pandemic response. critically, a typically mild pandemic (low g), such as that experienced in 2009, is predictably difficult to contain using a targeted antiviral strategy due to the low proportion of infectious cases that present to health authorities. the relative infectiousness of treated individuals, e t , is strongly negatively correlated with successful mitigation, perhaps a surprising result given the model's underlying assumption (based on available epidemiological and human clinical trials data) that e t lies in the range [0ae8, 1]. that is, nais provided as treatment have a maximum impact of just a 20% reduction in infectiousness. however, our previous results 4 show a strong synergistic effect of treatment when overlayed on a contact prophylaxis strategy, explaining the observation here that e t is critical in determining likely success of an intervention. despite the limited impact of treatment at the individual-level, the model outcomes are highly sensitive to the value of the relative infectiousness of treated cases. it follows that determination of e t is important for predicting the population-level outcome of a control effort. a 'small' reduction (of the order approximately 10%) may be extremely valuable in terms of success of a public health control strategy, and so should not be discounted. using a mathematical model which takes into account some of the key logistic constraints that are inherent to healthcare responses, we have derived a logistic regression model for estimating the probability that an antiviral intervention based on liberal distribution of nais as treatment and prophylaxis could successfully mitigate an influenza epidemic. the model demonstrates an excellent degree of accuracy when applied to synthetic data. the choice of parameters for the regression model was restricted to those that were both highly correlated with the success of the intervention and hopefully feasible to measure during the early stages of an emerging epidemic. the model could therefore be a useful near real-time decision support tool for public health policy in the face of an influenza epidemic, although further validation on a range of synthetic data (and real-world data where available) is required. influenza to seasonal flu status to avoid overstretching the demands on healthcare services. a great deal of information has emerged as the result of the pandemic response exercises conducted by affected countries. however, uncertainties remain regarding the effectiveness of intervention measures, as well as the feasibility and the timing of their implementation. mathematical and computational models [2] [3] [4] have been used to project the outcomes of influenza outbreaks under various scenarios and epidemiological hypotheses. motivated by the events of 2009 and public health measures adopted by the taiwan cdc, we use a stochastic, individual-based simulation model 5 to study the spatio-temporal transmission characteristics of the h1n1 virus, so as to quantitatively assess the effects of early intervention strategies. our stochastic disease simulation model 5 builds upon a highly connected network of individuals interacting with each other via social contact groups. to represent the daily interactions of approximately 23 million people living in taiwan, we constructed a computer-generated mock population based on national demographic and employment statistics (to derive daily commute patterns) from the 2000 taiwan census (http://www.stat.gov.tw/). each individual is created with a set of attributes, including age, sex, residence, family structure, and social standing (employment status, etc.). based on their attributes and the time of day, each individual is assigned to miscellaneous contact groups, where the potential of interactions between any two individuals resulting in flu virus transmission occurs. such epidemiological properties are defined by empirically parameterized attributes such as basic reproduction number r 0 , transmission probability, contact probability and associated probability distributions outlining the disease's natural history. additionally, intervention measures are implemented as scheduled events that could alter control parameters during the course of a simulation run. the targeted basic reproduction number (r 0 ) in all our simulations is 1ae6, following the suggested range by who of 1ae2-1ae7. 6 as the latent ⁄ incubation and infectious periods for h1n1 have not yet been reliably ascertained, we adopt the natural history of the 1957 and 1968 pandemic influenza viruses. 2, 7 here, the latent period ranged from 1 to 3 days, with a median value of 1ae9 days. the infectious periods begin 1 day prior to symptom onset and can continue for 3-6 days, with a median value of 4ae1 days. twothirds of the infected individuals will develop clinical symptoms, and the asymptomatic cases will have half the infectious strength. the efficacy of antiviral drugs (oseltamivir) and vaccines are based on these studies. 8, 9 for the source region of the infected cases, we use the north american continent (canada, mexico and united states) with an estimated total population of 450 527 697 and an average 16 hours of flight time to taiwan. the average daily passenger number is 2489 based on the 2009 annual statistical report on tourism, tourism bureau, taiwan (http://admin.taiwan.net.tw/english/statistics/year.asp? relno=61). each simulation lasts 365 days and starts with a baseline simulation of r 0 % 1ae6 h1n1pdm outbreak at the source region. the outbreak was adjusted to approximate clinical attack rate (car) in the united states, april 2009-march 13, 2010. 10 we estimate the daily number of imported cases according to average daily passenger numbers and their probability of holding a disease status. we then apply airport exit ⁄ entry screening per corresponding success rates, by subtracting the number of identified symptomatic cases. we also consider latently infected passengers with inflight disease progression, by fitting a gamma distribution to the cumulative distribution of time to onset data with 16 hours average flight-time, as presented by pitman et al. 11 the daily imported cases are seeded according to the traveling patterns of foreign tourists and residents returning home. from the disease's natural history, we derive that roughly 50% of the infected travelers present no symptoms; the percentage increases if most symptomatic individuals elect not to travel in their condition, or are stopped by airport screening. we use the official epidemic data provided by the taiwan cdc to calibrate the simulation model and perform regression analysis on scenario parameters. this data is a close estimation of the weekly new clinical cases of h1n1pdm patients. it consists of weekly opd (outpatient department) icd-9 code 487 (influenza) tallies collected by the bureau of national health insurance, taiwanadjusted to exclude seasonal flu patients and to account for uninsured patients. we formulate our scenario settings according to 2009 events in taiwan, and establish settings to approximate the actual events. with domestic events and intervention schedules fixed in time, the start date determines the simulation outcomes and the data range for selected indicators, such as the mean car, the epidemic peak, and several significant dates for the incoming index case events. we plot the 2009 taiwan weekly h1n1 opd487 cases alongside the weekly new clinical cases from our simulation results in figure 1 . our simulations not only capture the epidemic trend, but also pick out the most likely date, may 20, for identifying the first symptomatic case at airport screening based on practical assumptions. we further analyze the effectiveness of various mitigation measures with february 6, 2009 as the empirical start date for h1n1pdm in north america. the simulation result confirms that by the time we identified the first symptomatic case at the border screening, infected cases had already made their way to the public. by our calculation, roughly four such cases had passed in each of our scenario settings, with the first case happening as early as 3 weeks before detection. figure 1 also highlights the importance of the timing for the implementation of mitigation measures; for example, a 6-day-delay of the identical intervention plan results in nearly an additional 1% of the population being infected. therefore, the rule of thumb for healthcare officials is to implement intervention measures as early as possible. in our study, we have ignored the possibility of inflight transmission and any false positive results by airport screening procedures. to assess the effectiveness of each mitigation strategy of interest and their combinations, we take the calibrated simulation model and perform 100 simulation realizations for groups of scenarios containing only those intended mitigation measures, and analyze the averaged results. for example, in the airport exit screening policy only scenario, the first imported symptomatic case can be delayed up to 2 months, and the epidemic peak can be delayed up to 13 days. as the data suggests, the exit screening policy alone has very little impact on car. combining various screening success rates for both exit and entry screening allows us to quantitatively assess their beneficial ramifications on the epidemic. for example, there is very little additional benefit between 100% and 80% suc-cess rates for entry screening policies when exit screening policies are adequate, as the enhanced border screening only delayed the epidemic peak by 1 day, and reduced car by <0ae01%. base on this result, the government should not attempt to exhaust all its resources in securing the border during a pandemic event, because the return of such a policy will be disappointing. instead, a response plan with a shifting focus on health resource allocation and the capacity of adjusting intervention strategies in line with the developing epidemic will be most effective. based on the same principle, we perform experiments with assorted scenarios, including relaxing entry screening policies after identifying the first imported symptomatic case, mass vaccination based on the actual vaccination schedule of h1n1pdm in taiwan, and altering the start dates of the vaccination schedule. our results show that with a reasonable reduction in the airport entry screening success rate, we conserve valuable healthcare resources, but loose a few days for the strategic planning and preparation of subsequent response measures. in other simulation scenarios, a national vaccination campaign has very little impact on the outcome, due to the late start of the vaccination schedule. we then explore the effect of a national vaccination campaign with various starting dates. the simulation results are illustrated in figure 2 , where the benefit of an early start date for mass vaccination is clearly demonstrated. considering a scenario with an 80% airport exit screening success rate, 80% airport entry screening success rate and 80% symptomatic case tracing success rate, the combined intervention strategy results in: a 2% reduction in car if the vaccination campaign starts in mid-november; 11% reduction if the campaign starts in mid-october; 26% reduction if the campaign starts in mid-september; and 37% reduction if the campaign starts in mid-august. in retrospect, the taiwanese government's response to h1n1pdm proved to be effective. first and foremost, it initiated enhanced border monitoring and on-board quarantine inspection as soon as the threat of a flu pandemic became clear. at the same time, the domestic preparations towards h1n1pdm were escalated, such as antiviral drug stockpiling and distribution, and vaccine acquisition. as the h1n1 cases increased worldwide, various revised plans were adopted and implemented; such as the shift from labor-extensive on-board quarantine inspection to the notifiable infectious disease reporting system and realtime outbreak and disease surveillance system in order to effectively track down symptomatic and exposed passengers, apply prophylaxis treatment and mandatory in-home quarantine. as a result, all h1n1pdm related statistics are well below the international average. in modern society, countries rely heavily on the global economy for their own prosperity. shutting down the border for any length of time is not only costly, but could have disastrous economic effects that linger long after the event is over. moreover, with nearly 50% of the infected passengers presenting no symptoms whatsoever, they are not detectable by any port authority's screening procedures, and the importation of the novel flu virus is therefore inevitable. many studies conclude that entry screening is unlikely to be effective in preventing or delaying the importation of influenza, and has negligible impact on the course of subsequent epidemic. however, these studies are based on the assumption that effective exit screening is in place. our study shows that as the exit screening success rate decreases, the sensitivity of the entry screening policy becomes more pronounced. with the same methodology, we can also study the effects of varying the length of flight time, or the disease's incubation time. lastly, the benefit of entry screening is even more crucial for a small island country such as taiwan, since all incoming traffic must go through the port authority where entry screening can be enforced. in england and wales, three waves of the pandemic struck in summer, autumn, and winter seasons of 1918-1919. although the proportion of people reporting symptoms was often greater in the first wave, 1-3 a puzzling feature was the much higher mortality in the second wave, in which 0.27% of the population died, compared with 0.03% in the out-of-season first wave and 0.10% in the third wave. 4 an obvious hypothesis to explain the changes in mortality from wave to wave would be that the 1918 virus mutated to higher virulence after the (lower mortality) first wave. although pandemic virus reconstituted from the high mortality waves has proven to have high virulence in animals, 5 it has not been possible to recover virus from the first wave in 1918 for comparative purposes. indeed it is questionable whether virulence mutation(s) occurring between wave 1 and wave 2 could have spread to so many different populations in the time-frames observed. furthermore, in all three pandemic waves, there was the same agedistribution of mortality, with more deaths occurring amongst younger adults than older adults. [1] [2] [3] this 'pandemic signature', arguably due to immune protection of older adults who were exposed to a similar virus in the years before 1890, 6, 7 suggests that the 1918-1919 viruses were at least immunologically similar in all three waves. a second hypothesis would be that the higher case fatality in the later waves was due to higher rates of complicating bacterial pneumonia, 8 to increased transmission of influenza virus in the cooler months of the year, or to other seasonal effects. 9 we have considered a third (immunological) hypothesis to explain the greatly increased mortality in waves 2 and 3. the underlying idea is that the mortality rate in the first wave was lower than in later waves because most persons were protected by prior immunity in the first wave, and that the mortality was higher in later waves because of waning of that short-lived immunity. this hypothesis builds on our earlier modelling papers suggesting that even before the first wave in 1918, military, 10 school, and urban 11 populations in england and wales apparently had (short-lived) immune protection, presumably induced by recent prior exposure to seasonal influenza. [10] [11] [12] we suggest that this short-lived strain-transcending protection was in addition to the longer-lasting immunity, presumably induced by exposures to a similar virus circulating prior to 1890, that arguably reduced pandemic mortality for older adults in 1918-9. 6,7 cumulative mortality rates attributed to pandemic influenza were available for each of the three waves in 1918-1919 for 330 populations in england and wales. 13 we have built immunological models to potentially explain the variation in mortality rates across waves and populations. to show proof of principle, we have fitted these models to mortality data from a randomly selected sub-set of twenty populations. our key assumption was that the risk of a fatal infection would be limited to persons with inadequate immunity who were being exposed to the pandemic virus for the first time. persons who were exposed and who survived an earlier wave were assumed to be protected against death in a later wave. model a and assumptions (see figure 1) before the first wave, we assumed that people could be fully susceptible (s 0 ), or partially protected (q 0 ), or fully protected (p) by prior immunity which was not necessarily specific for the new virus. we assumed that exposure to the new pandemic virus would be fatal (m) in a proportion h of fully susceptible persons who were actually exposed (e) in the relevant wave. for those surviving that first exposure, it was assumed that they would be permanently protected against death in later waves by an immune assumed that viral exposure and multiplication would induce an immune response specific for the pandemic virus that would protect them against death in that wave and in subsequent waves. in contrast, for persons with strong prior immune protection, p, the virus would not be able to multiply to induce pandemic-specific immune protection. between waves, it is assumed that due to the waning of non-specific prior immunity, persons in the p state can move to the q state, and persons in the q state can move to an s state before the next wave. the proportion (e) of susceptible persons exposed to productive infection in each population was estimated by applying the following version of the final size equation 11 to the proportion susceptible (s & q) in each wave, for each population: note: in both figures 1 and 2 , we have omitted the flows out of the q and e states that removed persons from the risk of death. parameters: s 0 = proportion fully susceptible to infection and death before wave 1; q 0 = proportion susceptible to immunising infection, but not to death from exposure in wave 1; p 0 = proportion temporarily protected against both immunising infection and death from exposure in wave 1; n 0 = proportion even more protected against both immunising infection and death from exposure in wave 1 (model b only); r 0 = basic reproduction number (the average number of secondary cases for each primary case) in a fully susceptible population; f = proportion moving from q to s between waves; g = proportion moving from p to q between waves; d = proportion moving from n to p between waves (model b); h = proportion of e that actually move to m and die. model a could provide a very good fit for the summer, autumn, and winter waves of the 1918-1919 pandemic (results not shown). however, because of the replenishment of the pool of susceptible persons over time, model a also predicted a fourth wave of influenza in the spring season of 1919. as no such wave was seen, and as we could not find parameters values for model a that did not predict a fourth wave, we must regard model a as inadequate. model b was similar to model a, but with an additional stage of prior immunity (n), which could wane to p. model b allowed us to not only fit the three observed waves, but also to fit the imputed data (zero cases) corresponding to the absent fourth wave. following earlier work, 10,11 we used a bayesian approach with markov chain monte carlo (mcmc) procedures to estimate model parameters, and we used hyper-parameters to allow for parameter variation between populations. 11 the initial conditions were specified by the parameters: p 0 , q 0 , s 0 and n 0 . from these and the other parameters, it was possible to simulate the behaviour of model a over three waves, and of model b over four waves, and to estimate the expected numbers dying in each wave in each population. we calculated the log likelihood of the observed numbers of deaths given the parameter estimates, and we used mcmc simulation to generate the posterior distributions of parameters. although we obtained an excellent fit between observed and expected numbers of deaths in each of the three waves for the 20 populations for model a, we could not find parameter values for model a that would fit the three observed waves without giving rise to a fourth wave in the spring of 1919. accordingly, in the modified model b, we allowed for an additional stage of prior immunity (figure 2) , and we fitted the model to the same data, plus imputed data corresponding to 'the absent fourth wave'. we obtained a very good fit to the three observed waves and the absent fourth wave in each population. the 95% credibility intervals for parameter estimates, derived from the posterior distributions of the hyper-parameters were: h = 0.05-0.08, s 0 = 0.015-0.032, q 0 = 0.40-0.47; n 0 = 0.2 (fixed); p 0 = 1 ) s 0 ) q 0 ) n 0 ; f = 0.33-0.44; g = 0.87-0.94; d = 0.94-0.97 and r 0 = 2.6-3.6. this analysis had allowed all parameters to vary from population to population under the constraints of the hyper-parameters. however, several of the biologically determined parameters might be expected to be more constant from population to population, whereas those dependent on mixing history and other social characteristics which vary more widely from population to population. to test this possibility, we fixed the mean values for the more biological parameters (f = 0.385; d = 0.955; g = 0.905) and estimated the 95% credibility intervals for the others as: h = 0.05-0.10; s 0 = 0.012-0.032, q 0 = 0.36-0.44; and as before n 0 = 0.2 (fixed); in a subsequent paper we will be able to provide more details of the method, the robustness of the assumptions, and the results from fitting to many more populations. this short report suggests that the observed patterns of mortality in england and wales over the three waves of the 1918-1919 influenza pandemic 4,13 can be explained by an immunological model. in particular, the lower mortality in wave one can be explained by the assumption of protective immunity antedating the first wave, arguably induced by prior exposure to seasonal influenza. 10, 11 the much greater mortality in wave two can be explained by the waning, between wave one and wave two, of that short-lived and less-specific immune protection. the somewhat lesser mortality in wave three and the 'absent fourth wave' can be explained in terms of the progressive acquisition of immunity specific to the pandemic virus. the credibility estimates for parameters are of potential interest. for example, r 0 estimates of 3.1-4.5 across different populations are consistent with our earlier findings. 10, 11 if all persons had been susceptible, such r 0 values imply that the virus would have infected most people in all populations. however, even in the first wave, the proportion susceptible, s 0 + q 0 , was <50% in all populations, so that a considerable number of persons escaped productive infection in that wave; as their immunity waned, they became susceptible to infection in the later waves. it is likely that the variation in r 0 between populations is due to different rates of population mixing. estimates for h indicate that between 5% and 10% of infections in the most susceptible persons were fatal; the higher values of h could reflect higher rates of secondary bacterial infection in the most socially disadvantaged and overcrowded populations. 4 although we have shown the plausibility of an immunological explanation for wave to wave changes in pandemic mortality, we cannot assume that our particular model is even approximately correct. nor can we exclude the possibility that the higher mortality in the later pandemic waves in 1918-1919 was because of genetic change in the virus in later waves, or because of changing rates of secondary bacterial infection 8 or seasonal effects. 9 nevertheless, there is growing evidence that the population spread of pandemic influenza, whether in 1918-1919 10, 11 or in 2009, 12, 14 can be constrained by significant prior immunity, even for viruses that are ostensibly novel. previous reports, reviewed in ref. [10, 11] , support the idea of strain-transcending immune protection, which can wane over periods of a few months. this form of protection, probably induced by recent exposure to seasonal influenza, may not be mediated by hi or neutralizing antibody. 11 in contrast, strain-specific immunity, most often mediated by hi or neutralizing antibodies can be so long-lasting that after several decades it will still provide significant protection against any closely-related virus that re-appears in the population. 14 it has not escaped our notice that although attack-rates in the h1n1 2009 pandemic were low in many countries, with generally mild symptoms, the virus did cause lifethreatening illness in a small proportion of younger affected persons. 15 it seems likely that those who were most severely affected in 2009 were doubly unlucky: they had missed out on seasonal influenza infection or vaccination in the preceding season(s), and they were born too late to have been protected by the closely-related viruses that are thought to have circulated before 1950. 14 during the early phases of the 2009 influenza pandemic in italy, real-time modeling analysis were conducted in order to estimate the impact of the pandemic. in order to evaluate the results obtained by the model we compared simulated epidemics to the estimated number of influenza-like illness (ili) collected by the italian sentinel surveillance system (influnet), showing a good agreement with the timing of the observed epidemic. by assuming in the model mitigation measures implemented in italy, the peak was expected on week 44 (95% ci: 44, 45). results were consistent with the influnet data showing that the peak in italy was reached in week 46. these predictions have proved to be a valuable support for public health policy makers for planning interventions for mitigating the spread of the pandemic. mathematical models have recently become a useful tool to analyse disease dynamics of pandemic influenza virus can-didates. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] as of april 2009, after the pandemic threat emerged worldwide, 12 it was crucial for policy makers to have early predictions on the possible spread of the 2009 pandemic influenza virus in order to support, with quantitative insight into epidemic, policy decisions. thus, after the first pandemic alert was announced by the world health organization (who) in late april 2009, a national crisis management committee headed by the minister of health was established in italy in order to provide weekly advice to the italian ministry of health. real-time analyses using an individual based model were undertaken. the transmission model was previously used for evaluating the effectiveness of the control measures adopted in the national pandemic preparedness plan 3 and for assessing the age-prioritized distribution of antiviral doses during an influenza pandemic. 2 to parameterize the transmission model, we used data derived from the national surveillance system until 17 june 2009 and estimates of key epidemiological parameters as available at that time. in order to provide a preliminary assessment of the model predictions performed during the early stages of the epidemic, we compare model predictions with surveillance data of influenza-like illness (ili) available since august 2009. after the first pandemic alert was announced by the who in late april 2009, a national active surveillance system for the 2009 pandemic influenza was set up from 28 april to july 2009. 13 however, over the period from april to october 2009, surveillance systems, laboratory testing, and diagnostic strategies have varied considerably in italy. since end of july 2009, following who recommendations, 14 the focus of surveillance activities has changed in reporting requirements, as active case-finding became unsustainable and unnecessary. for this reason, the ministry of health (ministry of health, available in italian at the website: http://www.normativasanitaria.it) requested regional health authorities to report the weekly aggregated ili cases according to a new case definition (sudden onset of acute respiratory symptoms and fever >38°c plus at least one of the following systemic symptoms: headache, malaise, chills, sweats, fatigue; plus at least one of the following respiratory symptoms: cough, sore throat, nasal obstruction). by october 2009, following the increasing number of cases, the sentinel influenza surveillance system (influ-net available at: http://www.flu.iss.it) became the official surveillance system for ili cases in italy (ministry of health, available in italian at the website: http://www. normativasanitaria.it). since 1999, influnet is routinely based on a nation-wide, voluntary sentinel network of sentinel community based physicians in the 21 regions and autonomous provinces of the country. incidence rates are, therefore, not based on consultations, but on the served population of each reporting physician each week. influ-net usually consists of an average of 830 (range 648-902) general practitioners (including physicians and pediatricians) per year, covering about 1ae5-2% of the general population, (representative for age, geographic distribution, and urbanization level) reporting ili cases (according with a specific case definition). italian influnet surveillance system is part of the european influenza surveillance scheme (eiss). a stochastic, spatially explicit, individual-based simulation model 3 was used. individuals are explicitly represented and can transmit the infection to household members, to school ⁄ work colleagues, and in the general population (where the force of infection is assumed to depend explicitly on the geographic distance). the national transmission model was coupled with a global homogeneous mixing susceptible-exposed-infectious-removed (seir) model accounting for the worldwide epidemic, which is used for determining the number of cases imported over time. regarding the epidemiological assumptions (e.g., length and shape of the infectivity period, which lead to an effective generation time of 3ae2 days), this study is consistent with refs [3, 4, 11, 15] , but for the proportion of symptomatic individuals, which is assumed to be 66ae7%. 16 the basic reproductive number of the national transmission model was set to 1ae4, according to the early estimates as obtained during the initial phase of the epidemic in mexico in a community setting. 17, 18 we initialized our simulations through the global homogeneous mixing model in such a way that 158 imported cases were generated until 7 june 2009. this gives a reliable way for fixing the time in the simulations and thus determining the timing of school closure and vaccination in the simulations. the model accounts for school closure for both summer and christmas holidays: we assumed that in these periods contacts among students decrease, while contacts in the general community increase, as in ref. [19] . we also considered scenarios accounting for partial immunity in the population. 20 in order to investigate the effects of recommendations of the ministry of health (confirmed cases coming from affected areas were isolated for 7-10 days, either in hospital or at home) established in the early phase of the pandemic (april-july 2009), we assumed that a fraction of the imported symptomatic cases were isolated on the first day after the symptoms onset. this recommendation was in place until 27 july 2009. we also assumed, according to the italian school calendar, that schools were closed from 10 june 2009 to 10 september 2009 for the summer holidays, and from 22 december 2009 to 6 january 2010 for christmas holidays. the effects of prolonged school closure were also investigated. when considering vaccination, we assumed 6 weeks for the logistical distribution of doses of pandemic vaccine. since at the time of simulation specific recommendations regarding the administration of a single dose of pandemic vaccine from ema were not available yet, we considered the administration of 2 vaccine doses 1 month apart). the pandemic vaccine was considered effective after the administration of the second dose with a vaccine efficacy of 70%. we assumed the vaccine to be administered by priority, vaccinating first the target population accounting for essential services workers (including health care workers and blood donors), pregnant women at the second or third trimester, and at risk patients (with chronic underlying conditions) younger than 65 years old. the vaccination coverage was assumed 90%. regarding antiviral treatment and prophylaxis, recommendations of the ministry of health in the initial phase of the epidemic were to administer antivirals to all confirmed cases and to their close contacts. we assumed that the surveillance system would be able to detect 90% of symptomatic cases. after 8 july 2009, recommendations changed and antiviral treatment was considered only for cases with severe complications and in case of local clusters. since it was difficult to establish the proportion of treated cases, we considered different scenarios: antiviral treatment from 0% to 30% of the symptomatic cases. consistently with ref. [3] , both treatment and prophylaxis were assumed to start 1 day after the clinical onset of symptoms in the index case. treatment was assumed to reduce infectiousness by 70%, whereas antiviral prophylaxis was assumed to reduce susceptibility to infection by 30%, infectiousness by 70%, and the occurrence of symptomatic disease by 60%. as of 26 july 2009, approximately 1238 confirmed cases have been reported to the italian surveillance system for pandemic influenza. during july, the sudden increase of ili confirmed cases suggests for sustained autochthonous transmission in italy. by analyzing the number of ili cases reported to the surveillance during the weeks from 39 to 44, we found that the exponential growth rate was 0ae832 ⁄ week and thus we estimated the national reproductive number to be r 0 = 1ae38. this estimate of the basic reproductive number supports the choice of the value adopted in the model simulations (r 0 = 1ae4). in the absence of intervention measures, the predicted cumulative attack rate was 30ae6% (95% ci: 30ae6, 30ae7), and the peak was expected on week 42 (95% ci: 41, 42) with a peak day incidence of 0ae35% (95% ci: 0ae329%, 0ae37%). by assuming case isolation, antiviral treatment, and prophylaxis to 90% of symptomatic cases until 8 july 2009, the peak was expected on week 43 (95% ci: 43, 44). when considering 33ae3% of natural immunity in the population aged more than 59 years, the peak was expected 1 week later than in the previous scenario, i.e., on week 44 (95% ci: 44, 45). to validate the model, we compared model predictions (which are based only on the available information on the early phases of the epidemic) with ili data (figure 1 ). based on model predictions, we estimated the underreporting factor of influnet ranging from 3ae3 to 3ae7, considering different scenarios. by aligning the simulations with the ili data adjusted by the underreporting factor, we can observe that almost all the points in the increasing phase of the epidemic lie within the 95% ci of the model results (both considering or not natural immunity). the decay phase of the simulated epidemics shows a small delay with respect to the ili data. when introducing single and combined mitigation measures, such as case isolation, antiviral treatment, prophylaxis, and vaccination in the model, results showed that even a low proportion of symptomatic cases treated with antiviral drugs could have led to a relevant reduction in the epidemic size (table 1) . we simulated the planned italian vaccination strategy (begun on 15 october 2009), obtaining a limited but not negligible reduction in the attack rate with respect to the scenarios accounting only for antiviral treatment. moreover, the effect of vaccination would be higher if coupled with antiviral treatment; vaccination would have no effect on delaying the peak incidence. model predictions produced in italy during the early phase of the 2009 pandemic influenza are in excellent agreement with italian surveillance data on the beginning of the epidemic (when case isolation, antiviral treatment of index cases, and antiviral prophylaxis to close contacts were implemented by the italian regional public health authorities) and are basically consistent with the influnet data during the course of the epidemic. the model has been useful for predicting the timing of the epidemic, while it has overestimated the impact of the 2009 influenza pandemic for adult and elderly individuals. however, the disalignment is probably due to the model parameterization. based on literature values, 4,5 we assumed a similar fraction of cases in the different social contexts considered in the model (namely 1 ⁄ 3 in households, 1 ⁄ 3 in schools ⁄ workplaces, and 1 ⁄ 3 in the general community), since analysis on the relative transmissibility of the virus was not carried out for any country yet. we were also able to estimate an underreporting factor for the influnet data in the range 3ae3-3ae7. if we focus our attention on the reporting factor computed by considering the total number of cases (instead of symptomatic cases), the resulting value lies in the range 18-20ae2%, which is in excellent agreement with the range estimated in ref. [21] on previous a ⁄ h1n1 influenza seasons, namely 16ae2%-21ae6%. moreover, based on our results showing that vaccinating 40% of the italian population was more than adequate to mitigate the pandemic, the ministry of health decided to stockpile a limited number of vaccines. we have also shown that starting the vaccination program in october (or later) could have had only a limited effect on reducing the impact of the epidemic, although it may have been useful to prevent a possible second wave and to protect essential workers and at-risk patients. finally, our results have shown that antiviral treatment would have been the most efficient strategy to reduce the impact of the influenza pandemic, even with a limited antiviral stockpile. a population-wide passive immunotherapy program in this paper, we assume that convalescent plasma (cp) is efficacious in treating severe cases of pandemic influenza. under this premise, we test the hypothesis that a population-wide passive immunotherapy program that collects plasma from a small percentage of convalescent individuals can harvest sufficient cp to treat a substantial percentage of severe cases during the first wave of the pandemic. the proposed program involves recruiting adults (individuals age 20-55 years) to donate blood if they have experienced influenza-like symptoms more than 2 weeks ago (to account for the time needed for neutralizing antibodies to build up). the blood samples would be screened for infectious diseases (including hiv, hbv, hcv, htlv, and syphilis, etc., as in routine blood donation screening) and neutralizing antibodies against the pandemic virus. donors whose blood samples are free of known infectious agents and contain a sufficiently high titer of neutralizing antibodies would then be invited to donate plasma by plasmapheresis or routine whole blood donation. qualified donors with higher titers may be given higher priority for plasma donation. in this paper, we use the demographic and logistical parameters of hong kong as a case study. see figure 1 for a schematic of the proposed passive immunotherapy program. we examine the following questions regarding the logistical feasibility and potential benefits of the proposed passive immunotherapy program: (i) what percentage of convalescent individuals (donor percentage) is needed in order for the program to significantly reduce pandemic mortality? (ii) how many severe cases can be offered passive immunotherapy? (iii) what are the ratelimiting factors in the supply of passive immunotherapy? (iv) what are the epidemiologic and logistical factors that determine the demand-supply balance of passive immunotherapy? a more detailed presentation of our results is now available in ref. [ 5 ] . transmission and natural history model for pandemic influenza we use an age-structured disease transmission model to simulate the spread of pandemic influenza. the natural history model is similar to that used by basta et al. 6, 7 the most important parameter in characterizing the growth of an epidemic is the basic reproductive number r 0 , which is defined as the average number of secondary cases generated by a typically infectious individual in a completely susceptible population. we consider values of r 0 between 1ae2 and 2, which is consistent with recent estimates. 6, [8] [9] [10] logistical model for the passive immunotherapy program we assume that q d (%) of 20 to 55 year-old individuals who have recovered from symptomatic infections of pandemic influenza donate their blood for screening t r = 14 days after cessation of symptoms. follow-ups of convalescent individuals infected with h1n1pdm in an ongoing clinical trial of passive immunotherapy suggested that neutralizing antibodies level reaches maximal level around 14-21 days after recovery and stays at that level for months after. 11 we assume that q s (%) of these donors are qualified for plasma donation of which q r (%) are recurrent donors who return to donate plasma every t w = 14 days. screening involves both detection of infectious agents and neutralizing antibodies against the pandemic virus. the latter is the rate-limiting step because neutralization tests of pandemic viruses can only be done in a bsl3 setting. we assume that five bsl3-trained technicians are available to test the blood specimens, each running 150 viral neutralization tests in 3 days. therefore, the capacity and turnaround time of blood screening are u s = 750 and t s = 3 days, respectively. hong kong currently has nine plasmapheresis machines which allow a maximal throughput of 162 plasma donations per day (assuming 12-hour daily operation with each donation taking 40 minutes). therefore, the capacity and turnaround time of plasmapheresis are u p = 9 and t p = 1 ⁄ 18 days, respectively. collected cp are ready for use in transfusion after final quality check, which takes t q = 2 days. we assume that r t plasma donations are required to treat one severe case on average. the expert panel of the abovementioned study of passive immunotherapy for h1n1pdm in hong kong suggested that r t < 10. we assume that p h (%) of symptomatic cases will be severe cases for whom passive immunotherapy is suitable. although p h will be smaller than the case-hospitalization rate (passive immunotherapy may not be suitable for some hospitalized cases), we assume that the two have similar ranges and consider p h ranging from 0ae1% to 1%. because each severe case requires r t plasma donations on average, demand for cp is simply r t p h times the number of symptomatic cases. therefore, r t p h can be regarded as a single parameter, which we refer to as the lumped demand parameter. we define the outcome as the percentage of severe cases that can be offered passive immunotherapy by the proposed program during the first wave of the local epidemic. we refer to this outcome as treatment coverage and denote it by q. we consider the base case scenarios assuming q r = 20% and q s = 80%. in general, the treatment coverage q increases sharply as the basic reproductive number r 0 and the lumped demand parameter r t p h decrease (figure 2a ). in particular, when r 0 is large and r t p h is small, q is very sensitive to r t p h , but insensitive to r 0 . similarly, when r 0 and r t p h are small, q is very sensitive to both. with a donor percentage of q d = 15%, the proposed program can supply passive immunotherapy to more than 82% of severe cases (q > 82%) if r 0 < 1ae4 and r t p h < 1ae5%, but <35% if r 0 > 1ae8 and r t p h > 1ae5%. in general, the treatment coverage q increases sharply as the donor percentage q d rises from 0%, but with rapidly decreasing marginal increase ( figure 2b ). when r 0 < 1ae4 and r t p h < 1ae5%, q > 67% even if q d is as low as 5%, which is comparable to the current average blood donation rate of 38ae1 donations per 1000 population in developed countries. 12 when q d is >15%, q becomes largely insensitive to further increase in q d in most scenarios. the treatment coverage q for q d = 15% is more than 81% that for q d = 50% across all values of r 0 and r t p h considered in the base case. therefore, increasing the donor percentage q d beyond 15% has a relatively small impact on cp supply. this is because increasing q d can boost supply only when plasmapheresis is not yet the supply bottleneck. for the same reason, once the donor percentage q d has reached 15%, the treatment coverage q is insensitive to further increase in q d even when the plasmapheresis and screening capacity are doubled ( figure 2b , lower panel). we conduct an extensive multivariate sensitivity analysis to test the robustness of our base case observations against uncertainties in parameter values. we generate 15 000 epidemic scenarios by randomly selecting parameter values from their plausible ranges using latin-hypercube sampling. although there are numerous model parameters, the treatment coverage q is mainly determined by three lumped parameters: (i) r t p h , which indicates the magnitude of demand; (ii) q s q d , which indicates the magnitude of supply; (iii) the initial growth rate of the epidemic r (results not shown). while the dependence of q on r t p h and q s q d is readily comprehensible, it is not obvious a priori that q depends on the natural history and transmission dynamics of the disease via only the initial epidemic growth rate. when the plasmapheresis and screening capacity are very large, the supply-demand dynamics is further simplified: the treatment coverage q depends on lumped demand parameter r t p h and the lumped supply parameter q s q d only via their ratio. finally, q becomes insensitive to q s q d when the latter increases beyond 15-20%, which is consistent with our base case observations. our results suggest that with plasmapheresis capacity similar to that in hong kong, the proposed passive immunotherapy program can supply cp transfusion to treat 67-82% of severe cases in a moderate pandemic (basic reproductive number r 0 < 1ae4, lumped demand parameter r t p h < 1ae5%) when the donor percentage is 5-15%. increasing the donor percentage beyond 15% has little additional benefit because cp supply is constrained by the capacity of plasmapheresis during most stages of the epidemic. increasing plasmapheresis capacity could significantly boost cp supply, especially when there is a substantial pool of recurrent donors to alleviate the dependence of cp supply on donor percentage. in an ongoing clinical trial of passive immunotherapy for h1n1pdm virus infection in hong kong, 20% of convalescent individuals agreed to donate their plasma for the study. therefore, the donor percentage required by the proposed passive immunotherapy program (5-15%) is likely to be feasible. in view of the logistical feasibility of such program, we recommend that further clinical studies are conducted to evaluate the safety and efficacy of passive immunotherapy as a treatment for severe cases of pandemic influenza virus infection. our study is based on the premise that cp will be efficacious in reducing morbidity and mortality associated with pandemic influenza. in theory, the polyclonal nature of neutralizing antibodies in cp would lower the probability of an escape mutant emerging in treated patients. further, besides providing neutralizing antibodies against the pandemic virus, cp also might carry antibodies to other bacterial pathogens, which might decrease the severity of coexisting bacterial infections. 4 as such, cp not only might reduce the case fatality rate but might also increase the recovery rate and shorten duration of hospitalization of severe cases. the proposed passive immunotherapy program can thus significantly reduce the burden on the healthcare system, especially the intensive care unit, which will likely be stressed, if not overloaded, at the peak of an influenza pandemic wave, hence benefiting the general public and not only those receiving passive immunotherapy. although the hypothesized efficacy of cp has yet to be proven in clinical trials, our modeling results show that a public health system similar to that in hong kong has the capacity to support a population-wide passive immunotherapy program that can supply cp treatment to a substantial percentage of the severe cases in a moderately severe pandemic. we estimate that compared to other developed countries, hong kong has a relatively low plasmapheresis capacity. our conclusions regarding donor percentage needed and rate-limiting factors remain valid for plasmapheresis capacity ranging from 50% to 400% of what we have assumed in the base case (results not shown). our conclusions are robust against uncertainties in the natural history and transmission dynamics of pandemic influenza. our sensitivity analysis shows that the outcome depends on these epidemiological characteristics only via the initial growth rate of the epidemic. as such, our results are applicable not only to pandemic influenza, but also to other emerging infectious diseases for which the time-scales of disease transmission and antibody response are similar to that for influenza virus. the three determinants of treatment coverage (the initial epidemic growth rate, the lumped demand parameter r t p h , and the lumped supply parameter q d q s ) are all readily measurable in real-time during an epidemic. therefore, our methods and results can be used as a general reference for estimating the treatment coverage of the proposed passive immunotherapy program for a given plasmapheresis capacity. background highly pathogenic h5n1 virus continues to pose a serious threat to human health and appears to have the capacity to cause severe disease in previously healthy young children and adults. at present, antiviral therapy by oseltamivir remains the mainstay for managing h5n1 patients. while early treatment improves survival, approximately 50% of patients treated within 4 days of illness still succumb to the disease. in addition to the role of viral replication, there is good evidence that the host proinflammatory responses contributes to h5n1 pathogenesis. this suggests that both antiviral and immune-modulatory drugs may have a role in therapy. we previously demonstrated that cyclooxygenase 2 (cox-2) plays a regulatory role in h5n1 hyperinduced pro-inflammatory responses, and its inhibitor has potent effects at modulating this host response. now we demonstrate that, in addition to its immune-modulatory effect, a selective cox-2 inhibitor, ns-398 has a direct antiviral effect against h5n1 infection. materials and methods human primary monocytederived macrophages or alveolar epithelial cells (a549) were pre-treated with ns-398 or drug-vehicle for 1 hour before h5n1 virus infection. h5n1 viruses at multipicity of infection (moi) of 2 was used to infect the cells. following virus adsorption for 30 mins, the virus inoculum was removed, and the cells were washed and incubated in corresponding medium with ns-398 or drug-vehicle as controls for 3, 6, 24, 48, and 72 hours post-infection. cells were harvested for rna isolation at 6 hours post-infection to study viral matrix (m) gene expression. supernatants were collected for 50% tissue culture infection dose (tcid 50 ) assay to determine the virus titers at 3, 24, 48, and 72 hours after h5n1 infection. results ns-398 was found to suppress virus gene transcription and infectious virus yield in h5n1-infected human cells. conclusion we demonstrate that a selective cox-2 inhibitor, ns-398, shows an inhibitory effect on h5n1 viral replication in addition to its immune-modulatory effect that could counter the detrimental effects of excessive proinflammatory cytokine production. the findings suggest that selective cox-2 inhibitors may be a therapeutic target for treating h5n1 disease in combination with appropriate antiviral therapy. the emergence and spread of the highly pathogenic avain influenza viruses (h5n1) in poultry and wild birds with repeated zoonotic transmission to humans has raised pandemic concern. at the time of writing, 507 human cases have been reported with 302 fatalities, an overall case fatality rate of around 60% (cumulative number of confirmed human cases of avian influenza a ⁄ (h5n1) reported to world health organization updated to 18 october 2010). our previous data demonstrated that cox-2 was markedly up-regulated in h5n1-infected primary human macrophages, and that it played a regulatory role in the h5n1hyperinduced host pro-inflammatory responses. 1 such cytokine dysregulation is proposed to be a major contributor to the pathogenesis of h5n1 disease in humans. 2 with the use of selective cox-2 inhibitors, we found that the h5n1-hyperinduced cytokine response was significantly suppressed by the drug in a dose-dependent manner. 1 selective cox-2 inhibitor is a form of a non-steroidal anti-inflammatory drug that selectively targets cox-2, and it is an inducible enzyme responsible for inflammatory process and immune response. here, we report a novel finding of a direct antiviral effect of a selective cox-2 inhibitor, ns-398, against h5n1 infection in human primary macrophages and alveolar epithelial cells. taken together with our previous findings that suggest an immuno-modulatory effect that can modulate virus driven cytokine dysregula-tion, these findings highlight a role for cox-2 and its downstream signaling as potential novel targets for adjunctive therapy of severe viral pneumonia, such as that caused by h5n1. such therapy may be combined with conventional antiviral drugs. the h5n1 virus used was a ⁄ vietnam ⁄ 3212 ⁄ 04 (3212 ⁄ 04) (h5n1), a virus from a patient with h5n1 disease in vietnam during 2004. the viruses were grown and titrated in madin-darby canine kidney cells cells as described elsewhere. 3 virus infectivity was expressed as tcid 50 . all experiments were performed in a biosafety level 3 facility. monocyte-derived macrophages: peripheral-blood leucocytes were separated from buffy coats of healthy blood donors (provided by the hong kong red cross blood transfusion service) by centrifugation on a ficoll-paque density gradient (pharmacia biotech) and purified by adherence as reported previously. 3 the research protocol was approved by the ethics committee of the university of hong kong. macrophages were seeded onto tissue culture plates in rpmi 1640 medium supplemented with 5% heat-inactivated autologous plasma. the cells were allowed to differentiate for 14 days in vitro before use in the infectious experiments. alveolar epithelial cells: a549 cells were obtained from atcc and maintained in culture using dulbecco's modified eagle medium supplemented with 10% fetal calf serum, 0.6 mg ⁄ l penicillin, and 60 mg ⁄ l streptomycin. differentiated macrophages or a549 cells were pre-treated with a selective cox-2 inhibitor, ns-398 (cayman), at concentrations as indicated or drug-vehicle for 1 hour before infection. cells were infected with h5n1 viruses at moi of 2. following virus adsorption for 30 min, the virus inoculum was removed, the cells were washed and incubated in corresponding medium with ns-398 or drug-vehicle as controls throughout the experiments. cells were harvested for rna isolation at 6 hours post-infection to study viral m gene expression. supernatants were collected for tcid 50 assay to determine the virus titers at 3, 24, 48, and 72 hours after h5n1 infection. total rna was isolated using the rneasy mini kit (qiagen) according to the manufacturer's instructions. the cdna was synthesized from mrna with poly(dt) primers and superscript iii reverse transcriptase (invitrogen). transcript expression was monitored by real-time pcr using power sybr ò green pcr master mix kit (applied biosystems) with specific primers. the fluorescence signals were measured using the 7500 real-time pcr system (applied biosystems). the specificity of the sybr ò green pcr signal was confirmed by melting curve analysis. the threshold cycle (ct) was defined as the fractional cycle number at which the fluorescence reached 10 times the standard deviation of the base-line (from cycle 2 to 10). the ratio change in target gene relative to the b-actin control gene was determined by the 2 )ddct method as described elsewhere. 4 ns-398 reduced the viral m gene expression in h5n1infected human macrophages in a dose-dependent manner ( figure 1) . similarly, production of infectious virus yield in h5n1 infected macrophages was found to be suppressed in the presence of ns-398 at 100 lm compared to vehicletreated cells (figure 2a) . a comparable effect of ns-398 was observed in h5n1-infected human alveolar epithelial cells ( figure 2b ). we have previously demonstrated that cox-2 expression was dramatically upregulated following h5n1 infection in human macrophages in vitro and in epithelial cells of lung tissue samples obtained from autopsy of patients who died of h5n1 disease. 1 this suggests that cox-2 may be an important host factor involved in h5n1 pathogenesis and also provide a possible explanation on why h5n1 virus replication is susceptible to a selective cox-2 inhibitor. cox-2 was previously reported to play an important role in the pathogenesis of other influenza a viruses. 5 an in vivo study has highlighted the importance of cox-2 in h3n2infected mice. findings showed that infection induced less severe illness and reduced mortality in cox-2 knock-out mice than in wild-type mice. on the other hand, cox-1 knock-out mice had enhanced inflammation and earlier appearance of proinflammatory cytokines in the bal fluid, whereas the inflammatory and cytokine responses were dampened in cox-2 knock-out mice. these data suggests that cox-1 and cox-2 may lead to opposite totally contrasting effects on influenza h3n2 infected mice. cox-1 deficiency is detrimental, whereas cox-2 deficiency is beneficial to the host during influenza viral infection. therefore in the present study, instead of blocking cox enzymes in general as reported by others, 5 we have chosen ns-398 that selectively block cox-2 but preserve cox-1 activity and showed that this drug significantly reduced h5n1 virus replication in a dose-dependent manner. taken together with our previous report suggesting its immuno-modulatory effects, 1 we believe that selective cox-2 inhibitors and cox-2 signaling pathways deserve investigation as a promising approach for targeting therapy in h5n1 diseases. however, a few reports have suggested the importance of cox-2 in the late stage of inflammation for the resulution of inflammation, [6] [7] [8] and this raises concern whether inhibition of cox-2 may be harmful in treating diseases related to dysregulation of host inflammatory response such as acute lung injury, 9 which is a leading cause of death in h5n1 patients. we previously looked at the autopsy samples of lung tissues from h5n1 patients and found that cox-2 expression was markedly up-regulated compared with that from persons who died of non-respiratory causes. 1 moreover, data also demonstrated that pro-inflammatory cytokines, such as tnf-a, was markedly elevated in the h5n1 infected lung autopsies. 10 taken together, with the histo-pathological findings, which showed predominant features of exudative inflammatory phase in autopsy lung samples from h5n1 patients, 11, 12 we may therefore speculate that people who had fatal h5n1 infection died during acute inflammation phase, and before the resolution could occur, especially for the cases with a short disease duration (<10-12 days). 13 in conclusion, the roles of cox-2 in both pro-inflammation and pro-resolution phases deserves detailed investi-gation. the timing of selective cox-2 inhibitor therapy in h5n1 infected patients may be extremely critical. therefore a time-dependent study using selective cox-2 inhibitors on h5n1-infected animal models will be particularly important in order to address the effectiveness of this drug in treating h5n1 disease. avian antibodies to combat potential h5n1 pandemic and seasonal influenza highly pathogenic avian influenza a virus (hpaiv) strain a ⁄ h5n1 with unprecedented spread through much of asia and parts of europe in poultry remains a serious threat to human health. passive immunization (transfer of protective immunoglobulins) offers an alternative and ⁄ or additional strategy to prevent and cure influenza. here, we report that virus-specific immunoglobulin y (igy) isolated from eggs of immunized hens provide protection in mice against lethal h5n1 virus infection by neutralization of the viruses in the lungs upon intranasal administration. importantly, chicken eggs obtained from randomly selected supermarkets and farms in vietnam, where mass poultry vaccination against a ⁄ h5n1 is mandatory, contain high levels of igy specific for a ⁄ h5n1 virus. when administered before or after the infection, igy prevented and significantly reduced replication and spread of hpaiv h5n1 and related h5n2 strains. thus, the consumable eggs readily available in markets of countries that impose poultry vaccination against a ⁄ h5n1 could offer an enormous source of valuable biological material that provides protection against a ⁄ h5n1 virus with pandemic potential. the approach could be used to control seasonal influenza. since 2004, hpaiv of the h5n1 subtype has resulted in more than 430 cases of laboratory-confirmed human infection in 15 countries with a death rate of more than 50% (http://www.who.int/csr/disease/avian_influenza/). h5n1 influenza virus remains a global threat because of its continued transmission among domestic poultry and wild birds. passive immunization (the transfer of antigen-specific antibodies (abs) to a previously non-immune recipient host) offers an alternative and ⁄ or additional countermeasure against influenza. 1 development of human monoclonal antibodies (mabs) against h5n1 influenza haemagglutinin (ha) using epstein-barr virus (ebv) immortalization of b cells isolated from patients infected with h5n1, 2 phage display, 3 humanized mabs, 4 and human recombinant abs 5 has been attempted. chickens produce a unique immunoglobulin molecule called igy that is functionally equivalent to mammalian igg. 6 igy is found in the sera of chickens and is passed from hens to the embryo via the egg yolk. 7 egg igy has been used to prevent bacterial and viral infections (see review 8 ) of the gastrointestinal tract and recently for protection against pseudomonas aeruginosa infection of the respiratory tract of patients with cystic fibrosis (cf). 9 the epidemic of hpaiv h5n1 virus has resulted in serious economic losses to the poultry industry, mostly in southeast asia. therefore, many countries including china, indonesia, thailand, and vietnam have introduced mass vaccination of poultry with h5n1 virus vaccines that controls the h5n1 epidemic to some extent. 10 chickens immunized with recombinant h5 and ⁄ or inactivated h5n1 reassortant vaccines produced a high level of virus-specific serum antibodies (abs) and were protected from h5n1 virus challenge. 11 theoretically, these abs could be found in egg yolk and separated for use in humans to prevent and cure h5n1 hpaiv infection and disease, respectively. here, we examined the possibility that igy isolated from consumable eggs available in supermarkets in vietnam, where mandatory h5n1 vaccination has been implemented, provide prophylaxis and therapy of hpaiv h5n1 infection in mice. six-to 8-week-old female balb ⁄ canncrl (h-2d) mice (charles river and jackson laboratory) and hy-line . igy abs were extracted from egg yolks as previously described. 12 the 50% egg infectious dose (eid 50 ) was determined by serial titration of virus stock in eggs, and eid 50 ⁄ ml values were calculated according to the method of reed and muench. 13 human virus stocks were grown in mdck cells as described previously ,14 with viral titers determined by standard plaque assay. the 50% tissue culture infectious dose (tcid 50 ) of virus was determined by titration in mdck cells. 15 the standard elisa was performed for detection of anti-igy in the sera of igy-immunized mice. fifty percent lethal dose (ld 50 ) titers were determined by inoculating groups of eight mice i.n. with serial 10-fold dilutions of virus as previously described. 16 for infection, ketamine-anesthetized mice were inoculated intranasally with a lethal dose with 250 pfu (5 · ld 50 ) of a ⁄ pr ⁄ 8 ⁄ 34 (h1n1) virus as previously described, 17 10 · ld 50 of vn ⁄ 1203 (h5n1) or 5 · ld 50 a ⁄ aquatic bird ⁄ korea ⁄ w81 ⁄ 2005 (h5n2) resuspended in 50 ll pbs per animal. ketamine-anesthetized mice were treated intranasally with 50 ll of igy before or after infection. mice were observed for weight loss and mortality. subsets of animals were scarified for virus titre. we found comparable hai titers in the sera and egg yolks obtained from a farm in vietnam that was participating in a national mass vaccination program. furthermore we found 90% of eggs purchased in randomly selected supermarkets in hanoi, vietnam containing h5-specific igy. the hai and vn titers of pooled egg yolk igy are comparable with those of sera obtained from hens selected randomly from the farm that underwent supervised h5n1 vaccination. in contrast, igy separated from eggs purchased in korean markets where poultry are not vaccinated against avian influenza h5n1 has no detectable h5-specific hai or vn activity. we first treated naïve mice intranasally with h5n1-specific igy before infection with hpaiv h5n1 strain, a ⁄ vietnam ⁄ 1203 ⁄ 2004, isolated from a fatal case. such treated mice displayed mild weight loss and recovered completely by the end of the first week after inoculation ( figure 1a ). when animals were treated once with h5n1specific igy after h5n1 inoculation they exhibited minimal weight loss during the first week after inoculation, and virus titers in the lungs were substantial reduced at day 3 after infection; however, 50% of treated mice succumbed to infection during the second week after inoculation ( figure 1b) . it is possible that not all the hpaiv a ⁄ h5n1 viruses were neutralized upon the single treatment with igy, and escaping viruses can spread systemically to organs outside of the lungs. these viruses may reappear in lung tissue later when specific igy is absent. indeed, vn ⁄ 1203 virus injected intravenously or into the brain can spread to the lungs. 18 to circumvent the virus escape, we administered multiple treatments with h5n1specific igy after the infection. as a result, all infected mice recovered completely by the second week post-infection ( figure 1c) , and virus titers in the lungs were substantially reduced to the level that seen in protected mice that received single prior-infection treatment ( figure 1d) . similarly, the protective efficacy of h5n1-specific igy was observed in mice infected with lethal dose of mouseadapted avian influenza virus strain a ⁄ aquatic bird ⁄ korea ⁄ w81 ⁄ 2005 (h5n2). this virus shares 94.4% nucleotide sequence homology with ha (h5) but has different na (n2) from the one used for mass immunization in vietnam (reassortant avian h5n1 influenza virus a ⁄ goose ⁄ gd ⁄ 96-derived, strain re-1). the results indicate that h5n1-specific igy isolated from eggs purchased in markets have preventive and therapeutic effects against infection with hpaiv h5n1 and the related strain h5n2. the findings suggest that while a single treatment with igy prior to lethal infection was sufficient to protect the animals from the infection, multiple treatment is required for complete therapeutic effect after infection with hpaiv such as vn ⁄ 1203 strain. we further examined the protective efficacy of igy isolated from eggs laid by hens immunized in the laboratory with heat-inactivated human influenza a ⁄ h1n1 virus, a ⁄ pr ⁄ 8 ⁄ 34. we found substantial levels of hai and vn abs in the sera and yolks derived from immunized hens. when naïve mice were administered intranasally with such anti-pr ⁄ 8 igy at 6-8 hours before or after infection with lethal dose of pr ⁄ 8 virus, they were protected from the infection or lethal disease, respectively. the virus titers in the lungs of a ⁄ pr8 specific igy-treated mice at day 3 after infection were also significantly lower than those seen in untreated mice or mice receiving normal igy. intranasal administration is the most effective route as compared to oral or peritoneal or intravenous administration for protection against lethal challenge, and the presence of virus-specific igy in bronchoalveolar lavage (bal) is required for the protection. the results provide a proof-of-concept that intranasal administration of virus-specific igy prevents influenza virus infection and cures the disease. the concept could be applied to control influenza outbreaks including seasonal and pandemic influenza. the protection was correlated with hai and vn activities of the igy and reduced virus titers in the lungs after treatments, suggesting that the protection is mediated by vn. we asked if administration of igy in the respiratory tract induces anti-igy ab response in mice. if this is the case, the next question is whether pre-existing anti-igy abs block igy-mediated protection. indeed, significant levels of anti-igy were observed in animals that received single or multiple administration of igy. when igy-immune mice were treated with virus-specific igy before or after lethal challenge, the results were identical to those obtained from treated naive mice, indicating that pre-existing anti-igy abs do not interfere with the protection mediated by virus-specific igy. consistently, incubation with anti-igy serum did not interfere with hai and vn activity of the virus-specific igy, indicating that anti-igy abs do not block virus binding by virus-specific igy (figure 2 ). the finding suggests that the igy treatment could be applied to persons who have developed anti-igy during the individuals' life, and such treatment strategy could be repeated if multiple treatment is required and ⁄ or necessary later on to protect infections with other pathogens. the approach using specific igy for prevention and therapy of hpaiv h5n1 infection offers a practical alternative to immunotherapy using convalescent plasma 19 and an additional therapeutic option to antiviral drugs since widespread drug resistance has been recently reported among influenza virus strains. igy is relatively stable. we found no change in protective activity after at least 12 months storage at 4°c, and lyophilization does not affect the activity, making production of igy practical. the use of igy immunotherapy has many advantages, since igy does not activate the human complement system or human fc-receptors, which all are well-known cell activators and mediators of inflammation. 20 we chose the water dilution method for preparation of igy. the method is simple, efficient and does not require any toxic compounds or any additives. such igy preparations by this method have been used in other human study. 9, 21 eggs are normal dietary components, so there is minimal risk of toxic side effects, except for those with egg allergy. thus, our study demonstrated that influenza virus-specific igy can be used in passive immunization that provides great help for immunocompromised patients and elderly who have weaken immune response to influenza vaccines. importantly, the consumable eggs readily available in the markets of countries that impose mandatory h5n1 vaccination offer an enormous source of valuable, affordable, and safe biological material for prevention and protection against potential h5n1 pandemic influenza. parts of the information and data presented in this manuscript were previously published in http://www.plosone.org/ article/info:doi%2f10.1371%2fjournal.pone.0010152. the polyphenol rich plant extract cystus052 is highly introduction the 2009 ⁄ 2010 h1n1influenza a virus pandemic clearly demonstrates that influenza is still a major risk for the public health. although the pandemic swine origin influenza a virus (soiv) caused only mild symptoms, the control of the outbreak still remains difficult. even as vaccine is available against this virus, the possibility of reassortment between the pandemic and a seasonal or avian a ⁄ h5n1 influenza virus strain is indeed a frightening, but a likely event. this reassortant strain might be able to transmit easily between humans causing fatal infections, and the current soiv vaccine might no longer be sufficient to protect against the reassorted virus. in such a case, we can only rely on effective antiviral drugs. today, neuraminidaseinhibitors, such as oseltamivir, represent the most common clinically approved medication against influenza a viruses. unfortunately, the frequency of reports describing the appearance of drug-resistant seasonal h1n1 and also h5n1 influenza a viruses dramatically increased in the recent past. [1] [2] [3] [4] drug resistance to the known antivirals highlights the urgent need for alternative antiviral compounds with novel defense mechanisms. recently, we have reported that a polyphenol rich plant extract, cystus052, which showed antiviral activity against influenza a viruses in cell culture and in mice. 5, 6 moreover, the antiviral activity of cy-stus052 against seasonal influenza virus and common colds was also demonstrated in humans. 7 however, the efficiency of cystus052 against soiv and a ⁄ h5n1 isolates was unknown so far. therefore, we investigated cy-stus052 effectiveness against the pandemic strain and seven natural influenza a ⁄ h5n1 isolates detected in several avian species during 2006 ⁄ 2007 avian influenza outbreak. additionally, the potency of the most common neuraminidase inhibitor oseltamivir was also investigated against these isolates. here, we show that cystus052 treatment was effective in in vitro studies against soiv and a ⁄ h5n1 influenza virus. viruses avian h5n1 isolates were originally obtained from the bavarian health and food safety authority, oberschleissheim, germany. the soiv a ⁄ hamburg ⁄ 4 ⁄ 2009 was obtained from the robert-koch-institut, berlin, germany. all h5n1 viruses were further propagated in embryonated chicken eggs or mdck ii (h1n1v) cells at the friedrich-loeffler-institut, tübingen, germany. for the cytopathological effect (cpe) inhibition screening, in accordance with sidwell, 8 mdck ii cells were infected with different viruses at moi of 0ae005. virus-infected cells were then treated with antiviral compounds cystus052 from 0ae1 to 1000 lg ⁄ ml or oseltamivir from 0ae01 nm to 1 mm. after incubation for 48 hours at 37°c and 5% co 2 , cells were fixed, and viable cells were stained with crystal violet. after extraction of crystal violet from viable cells with 100% methanol, the extinction was measured with an elisa reader. immediately before infection, mdck ii cells (8 · 10 4 cells ⁄ well) were washed with pbs and subsequently incubated with virus diluted in pbs ⁄ ba (0ae2% ba) 1 mm mgcl 2 , 0ae9 mm cacl 2 , penicillin and streptomycin to a multiplicity of infection (moi) of 0ae001 for 30 minutes at 37°c. cystus052 was added in a concentration of 50 lg ⁄ ml directly to the virus-stock and on the cell monolayer simultaneously with the infection. after 30 minutes incubation period, the inoculums were aspirated and cells were incubated with either mem or mem containing 1 lm oseltamivir. at indicated time points, supernatants were collected. infectious particles (plaque titers) in the supernatants were assessed by a plaque assay under avicel as described previously. 9 in order to investigate the antiviral potential of cy-stus052, ec 50 values based on the inhibition of the cpe on mdck ii cells were determined for cystus052 and in addition for oseltamivir. the ec 50 values for cystus052 ranged from 1ae53 to 18ae88 lg ⁄ ml. cystus052 demonstrated the highest sensitivity against the soiv, sn1 and mb1 isolates with ec 50 values below 5 lg ⁄ ml. compared to these virus strains, cystus052 showed a slightly increased ec 50 value for gsb1 (18ae88 lg ⁄ ml). in contrast the ec 50 values for bb1 and bb2 were notably elevated (65ae68 and 76ae22 lg ⁄ ml). thus, the weakest antiviral effect of cystus052 was observed against these two isolates. the ec 50 values evaluated for oseltamivir ranged from 0ae07 to 512ae76 lm ( table 1 ), indicating that bb2 (512ae76) and gsb1 (356ae92 lm) can be considered resistant against oseltamivir. to confirm these results we investigated the ability of cystus052 to block virus replication as published before. 1 as a control, virus infected cells were treated with oseltamivir as described earlier. 6 in the absence of the drugs all influenza strains showed similar growth properties (figure 1, black squares) . first progeny viruses were detectable between 8 and 20 hours post infection (figure 1, black squares) . treatment with cystus052 resulted in reduction of virus titers of all influenza virus strains (fig. 1a-h, open triangles) . surprisingly, oseltamivir failed to inhibit the replication of two h5n1 influenza virus strains (gsb1 and bb2), supporting the data of ec 50 values ( figure 1d+h , grey rhombes). we assessed the antiviral activity of cystus052 against the newly emerged soiv and seven avian h5n1 influenza viruses. cystus052 showed efficient antiviral activity against the pandemic h1n1v strain and was effective to a wide range of h5n1 viruses. furthermore, cystus052 demonstrated a broader and more efficient antiviral potential than oseltamivir. cystus052 treatment leads to a stronger reduction of progeny virus titers, and more importantly, cystus052 was effective against all tested viruses, while oseltamivir was unresponsive against two of seven a ⁄ h5n1 viruses. even though the pandemic strain in general is still sensitive to oseltamivir treatment, there are increasing numbers of reports of emerging resistant variants. the treatment with cystus052 does not result in the emergence of viral drug resistance since the mode of action is an unspecific physical binding of the virus particle that is also beneficial to reduce opportunistic bacterial infections. 5,7,10 cystus052 is an extract from a special variety of the plant cistus incanus, and it is very rich in polymeric polyphenols. 11 it is well known that polyphenols exhibit protein-binding capacity. 12 however, cystus052 exhibited no neuraminidase inhibiting activity. therefore, ingredients of cystus052 may act in a rather unspecific physical manner by interfering with the viral hemagglutinin at the surface of the virus particle as demonstrated before. 5 while this prevents binding of the virion to cellular receptors, it does not block accessibility and action of the viral neuraminidase. since, infections with influenza a viruses are still a major health burden and the options for control and treatment of the disease are limited, plant extracts such as cystus052 should be considered as a new candidate drug for a save prophylactic and therapeutic use against influenza viruses. attenuation of respiratory immune responses by antiviral neuraminidase inhibitor treatment and boost of mucosal immunoglobulin a response by co-administration of immuno-modulator clarithromycin in paediatric influenza the antiviral neuraminidase inhibitor osv and zanamivir are widely used treatment options for influenza infection and are being stockpiled in many countries. although mucosal immunity is the frontline of defense against pathogens, the effects of neuraminidase inhibitor treatment on airway mucosal immunity have not been reported. the suppression of viral rna replication and viral antigenic production by these drugs may result in a limited immune response against influenza virus. 1 macrolides, such as cam and azithromycin, have anti-inflammatory and immunomodulatory properties that are separate from their antibacterial effects. [2] [3] [4] this study examined the impact of osv treatment on immune responses in the airway mucosa and plasma in mice infected with iav and pediatric influenza patients. we also assessed the immuno-modulatory effects of cam in influenza patients who were treated with or without osv. female 3ae5-week-old weanling balb ⁄ c mice were nasally inoculated with 25 pfu of iav ⁄ pr8 ⁄ 34 h1n1 at day 0. immediately after infection, mice were given 50 lg of osv orally or vehicle at 12-hours intervals for 11 days. the levels of virus-specific siga in nws and bronchoalveolar fluids (balf) and igg in plasma were measured by elisa as reported previously. 5 a retrospective clinical study was conducted. for the study, 40 children with acute influenza were recruited and grouped according to the treatment received: 5 days treatment with osv (n = 14), cam (n = 8), osv + cam (n = 12), and untreated (n = 6). since parents in japan are well aware of the adverse effects of osv especially the neuropsychiatric complications, 6 the decision on whether to administer osv or not and to prescribe cam was made by the parents and the attending paediatricians, based on their anti-viral and immuno-modulatory activity. 3, 6 comparisons were made of the levels of siga against iav ⁄ h3n2 and iav ⁄ h1n1, total siga, in nws and disease symptoms before and after treatment. anti-ha siga and total siga in nws of patients were determined from the standard regression curves with human iga of known concentration in a human iga quantitation kit (bethy laboratories). because an affinity purified human anti-ha-specific siga standard of each influenza a subtype is not available, the relative value of anti-haspecific siga amount was expressed as unit (u). one unit was defined as the amount of one lg of human iga detected in the assay system as reported previously. 6 the concentrations of siga in individual nws were normalized by the levels of total siga (lg ⁄ ml). oseltamivir suppresses viral rna replication and viral antigenic protein production. to investigate the influence of daily treatment with osv on ha-specific mucosal and systemic immune responses, we analyzed ha-specific siga levels in nws and balf as well as igg levels in plasma at days 8 and 12 post-infection in mice treated orally with osv or methylcellulose (mc) as vehicle. the osv treated mice showed lower antibody responses in nws and balf than control mice treated with mc solution (table 1) . significantly reduced ha-specific siga responses were particularly noted in the osv group at day 12, the period of maximal mucosal siga induction. the airway secretions and plasma from mice at day 0 did not contain detectable levels of ha-specific antibodies. these findings were supported by other data whereby mice treated with osv displayed significantly lower numbers of ha-specific iga antibody-forming cells (afcs) in the nasal lamina propria, mediastinal lymph nodes, and lungs compared with mc-treated mice. these results clearly indicate that oral administration of osv downregulates ha-specific siga responses in mucosa. on the other hand, there were no significant differences in the elevated levels of ha-specific plasma iga and igg antibodies or the increased numbers of ha-specific iga and igg afcs in the spleen between osv-and mc-treated mice. taken together, these results implicated the oral administration of osv in a suppressed induction of haspecific siga responses in respiratory lymphoid tissues, although systemic ha-specific antibody responses were not significantly affected by osv. since cam up-regulates il-12, a mucosal adjuvant cytokine in the airways, and promotes the induction of siga and igg in the airway fluids of mice infected with iav, 4, 7 we assessed the impact of treatment with osv and ⁄ or cam on the levels of anti-influenza siga in nws and clinical status of influenza patients. the concentration ratio of table 1 . anti-ha-specific siga to total siga in nws was expressed as titer: anti-ha-specific siga (u ⁄ mg) ⁄ total siga (lg ⁄ mg) · 100. figure 1 shows changes in the anti-ha(h3n2) siga ratio (titer) and fold of increase in siga titer in each patient during the 5-days' treatment for the four different treatment groups. it is noteworthy that, upon admission to the hospital, the siga titers were <3 in 93% of patients. during the 5 days of treatment, rapid increases in the titers were observed in almost all patients in cam, osv + cam, and no treatment groups. in contrast, in the osv group, the anti-ha-specific siga titers remained unchanged or decreased in the majority of patients. the finding of significant low induction of anti-viral siga in the osv group was supported by the results of animal experiments. however, the addition of cam to osv augmented siga production and restored mucosal siga levels; 75% of patients treated with osv + cam showed >5-fold increase in the titers during treatment. these observations suggest that cam stimulated the local mucosal immunoresponse in the nasopharyngeal region of patients treated with osv. the prevalence of disease manifestations was also analyzed. 6 among the symptoms listed, a significant decrease in the prevalence of cough was recorded between the no treatment group and the osv + cam group and between the osv group and the osv + cam group (**p < 0ae01), despite the limited number of patients in each group. the duration of the febrile period was significantly shorter in the osv and osv + cam groups than the no treatment group. however, no significant difference was observed between the osv group and osv + cam group. it has been reported that osv does not affect the cellular immune responses, such as cytotoxic t lymphocytes and natural killer cells. 8 however, the effects of osv on mucosal immunity have not been studied so far. the present study showed that osv treatment of mice infected with iav induced insufficient protective mucosal siga responses in the respiratory tract, although treated mice showed the similar levels of systemic igg and iga antibody responses in plasma to those in mice treated with vehicle (table 1) . 1 the observed effect of osv on mucosal immunity was probably due to a suppression of viral replication and viral antigen production in the mucosal layer. these observations in mice are further supported by our clinical reports of siga in nws and balf of osv treated influenza patients. 6 the 14 membered-and 15 membered-ring macrolides have been found to possess a wide range of anti-inflammatory and immuno-modulatory properties, 2,3 and to be effective in the treatment of respiratory syncytia and iav infection. 9,10 the efficacy of low doses administered on the long term against pathogens that are insensitive to macrolides indicates a mode of action that is separate from their antibacterial activity. 2, 3, 9, 11 in the present study, we evaluated the immunomodulatory effects of cam on mucosal immune responses in pediatric influenza. a decrease in the proportion of total siga that was anti-ha-specific siga during treatment was observed in 21.4% of patients in the osv group (those represented by the dotted lines and closed diamonds in figure 1 ), whereas an increase in the proportion was observed in most patients of the other groups (except for one patient of the untreated group). despite the low or unchanged induction of anti-ha-specific siga in the majority of osv-treated patients, the additional use of cam with osv boosted the mucosal immune response and restored local mucosal siga levels. we are currently engaged in detailed immunological studies of the effects of cam and osv on the levels of mediators controlling iga class switching in nws of influenza patients and airway secretion of mice infected with iav. further studies should clarify the boost mechanisms of cam and the suppression mechanisms of osv in iga class switching. our findings suggest the risk of re-infection in patients showing a low mucosal response following osv treatment and cam effectively boosts the siga production for protection of re-infection. to date there is an urgent need to develop new antivirals against influenza. most of the molecules reported target influenza proteins that acquire rapid mutations of resistance. the development of new molecules that have a broad antiviral activity and are not subjected to influenza mutation is of particular interest. our laboratory and others recently showed that proteases can participate to the innate immune response in the airways through the activation of a family of receptors called par. in particular, through the release of interferon, par2 agonists curbed viral replication significantly in infected cells. in this study, since erk activation is crucial for virus replication, we investigated whether par2 could inhibit virus replication through inhibition of the erk pathway. results showed that while influenza a infection alone or par2 stimulation alone induced erk activation, par2 stimulation does not inhibit erk activation in influenza infected cells. thus, par2 agonists may be a potential new drug against influenza viruses that could be used in combination with other anti flu therapy such as the inhibition of the erk pathway. respiratory tract-resident proteases are key players during influenza virus type a infection. 1, 2 in addition to their direct activating effect on surface viral proteins, lung mucosal proteases can regulate cellular processes by their ability to signal through protease-activated receptors (pars). 3 after cleavage of the receptor by proteases, the new aminoterminal sequence of par binds and activates the receptor internally. these receptors are highly expressed at epithelial surfaces, in particular in the lung, where human influenza virus replicate in vivo. pars are thus directly exposed to proteases present in the airways. among the four different pars, par2 acts as an antiviral through an interferondependent pathway. 4, 5 thus, agonists of par2 are potential new drugs against a broad range of influenza viruses, which is in accordance with the broad antiviral action of interferon. however, the signalling pathway induced by par2 agonists in influenza a infected cells has still to be investigated. in this manuscript, we showed that influenza infection or activation of par2 induced erk activation, a crucial step for efficient virus replication. 6, 7 however, par2 agonists do not impaired erk activation in influenza a virus infected cells. since the pathway of par2 protection is likely to be erk-independent, the use of anti erk molecules in combination with par2 agonists maybe of potential interest in future anti-influenza therapy. influenza viruses a ⁄ wsn ⁄ 33 (h1n1) (a kind gift from nadia naffakh) was used in the present study. mdck (madin-darby canine kidney) and the human alveolar type ii a549 cell were obtained from atcc and grown as previously described. 8 for western blot analysis, the following antibodies were used: monoclonal antibody for phospho-erk1 ⁄ 2 (t202 ⁄ y204) and for erk1 ⁄ 2 antibodies from cell signaling technology (beverly, ma), horseradish peroxydase (hrp)-coupled rabbit polyclonal antibodies against mouse or rabbit igg from paris (compiègne, france). a549 cells were infected with iav at an moi of 1 in emem medium, as previously described. 9,10 at various time points post infection, cells were collected and proteins were analysed as previously described. 6,11 par2 stimulation was performed at 37°c in emem medium as previously described. 4 after infection and ⁄ or stimulation, cells were lysed in ice-cold lysis buffer. lysates were centrifuged at 12 000 g for 20 min, and total proteins of the supernatants were analyzed by western blot analysis as previously described. 12, 13 results since activation of the erk pathway is essential for efficient influenza replication, 7 we first investigated the kinetics of erk activation after influenza infection in human a549 alveolar epithelial cells. for this purpose, a549 cells were infected with influenza viruses at a moi of 1 at different time point post-infection, and activation of erk1 ⁄ 2 pathway was assessed by western blot analysis using an anti-erk antibody. results showed that erk was phosphorylated after influenza infection in a time course depen-dent manner when compared to uninfected cells. in contrast, erk phosphorylation was not observed with heatinactivated viruses, suggesting that productive infection is needed for erk activation ( figure 1a ). antibodies against erk1 ⁄ 2 were used as controls. since erk is activated after influenza infection, we then tested whether activation of par2 in uninfected cells also leads to activation of this pathway. for this purpose, a549 cells were stimulated with the selective human (h) or mouse (m) par2 agonist or a control peptide for the indicated time ( figure 1b ). when exposed to the par2 agonists and compared to controltreated cells, erk phosphorylation increased over the time course of stimulation. thus, influenza infection or stimulation of par2 without infection in a549 cells induced activation of the erk pathway at different time point post-infection. since influenza infection and par2 stimulation induced erk activation, we then investigated whether par2 could inhibit erk activation in influenza infected a549 cells. results in figure 2 showed that in influenza infected cells, par2 activation for ten minutes does not inhibit erk activation after influenza infection. thus, erk activation is not inhibited by par2 activation in influenza stimulated cells. in this manuscript, we studied the activation of the erk pathway after par2 stimulation and or influenza infection. particularly interesting is the fact that either influenza infection or par2 stimulation alone induce erk phosphorylation in a549 epithelial cells, while erk activation is not inhibited in a549 infected cells compared to uninfected ones after par2 stimulation. proteases are key factor in the pathogenicity of influenza viruses. in addition to the cleavage of ha, necessary for iav replication, extracellular proteases also play a role in the modulation of the immune system against influenza viruses through the activation of pars. particularly par2, activated by extracellular trypsin-like proteases, could inhibit virus replication through the release of interferon, 4,5 thus, strengthening the immune system via agonist peptides and providing new therapeutic potential against a broad range of influenza strains. in addition, targeting the host instead of the virus could provide a way to escape from virus resistance. 14 thus, a better understanding of how virus escapes from immune surveillance may provide new therapeutic strategies to block iav. in addition, combinations of drugs that block virus replication via different pathways are of interest. the non classical molecules hla-g maybe an interesting new target as we recently showed that it is upregulated after influenza infection, 13 and it is a well known immunotolerant molecule. 15 indeed, it inhibits the innate immune response 16 as well as the adaptive immune response. 17, 18 also, as previously suggested, the erk signal transduction cascade is also of potential interest since it is crucial for virus replication and particularly influenza replication. 6, 7 as shown here, it is unlikely that par2 protection occurs through an erkdependent pathway. thus strengthening the immune response with par2 agonists and blocking nuclear retention of the viral ribonucleoprotein complexes with inhibitors of the mek ⁄ erk pathway may be alternative combinatory approaches for influenza therapy. in addition, since those potential drugs target the host instead of the virus, this could help in the design of new antivirals molecules more resilient to iav mutations and thus to virus resistance. the initial waves of the first influenza pandemic of the 21st century have passed. in june 2009, vaccine companies estimated they could produce in 6 months almost 2.5 billion doses of pandemic vaccine. 1 instead, they actually produced only 534 million doses, of which 64% were non adjuvanted preparations. had these doses been produced with adjuvants (i.e., 3.75 lg instead of 15 lg ha per dose), an additional 1 billion doses could have been made available. yet there was public opposition to adjuvants in many countries, especially by regulatory officials in the united states. misperceptions about the safety of both adjuvanted and nonadjuvanted vaccines were widespread. added to this, shortfalls in vaccine production, delays in vaccine delivery, and the ''mildness'' of the pandemic itself meant that only a few countries achieved reasonable levels of vaccine coverage. millions of doses went unused and had to be destroyed. supplies of antiviral agents were even more limited. thus, despite the best efforts of influenza scientists, health officials, and companies, more than 90% of the world's people did not have timely access to affordable supplies of vaccines and antiviral agents. instead, they had to rely on 19th century public health ''technologies.'' given current understanding of biology in the early 21st century, they should have had -and probably could have had -something better. this report reviews evidence for an alternative approach to serious and pandemic influenza that could be used in all countries with basic health care systems. instead of confronting the influenza virus with vaccines and antiviral agents, it suggests that we might be able to modify the host response to influenza virus infection by using anti-inflammatory and immunomodulatory agents. this idea was introduced several years ago 2 and has been reviewed in several publications. [3] [4] [5] [6] [7] [8] the central importance of the host response in the 1918 pandemic, young adults had high mortality rates. ever since, influenza virologists have sought to answer the question ''why did young adults die?'' by defining the molecular characteristics of the 1918 virus that were responsible for its virulence. 8 in doing so, they have overlooked a crucial piece of clinical evidence from the 1918 pandemic: compared with young adults, children were infected more frequently with the same virus, yet they seldom died. consequently, the more important question is ''why did children live?'' this can only be explained by recognizing that children must have had a different host response to the 1918 influenza virus than adults. physicians have long recognized that for several other medical conditions, both infectious (e.g., pneumococcal bacteremia) and non-infectious (e.g., multiple trauma), children have a more benign clinical course than adults. 6, 8 a corollary of this observation is that secondary bacterial pneumonia, although commonly found in young adults in 1918, could not have been the primary cause of death. children must have had the same or higher rates of nasopharyngeal colonization with the same bacteria that were associated with pneumonia deaths in adults, yet children seldom died of secondary bacterial pneumonia. 6 if young adults died with secondary bacterial pneumonia, underlying host factors must have made them more susceptible. few people who die of influenza do so during the first few days of illness when pro-inflammatory cytokine levels are high. instead, like patients with sepsis, they usually die in the second week, when anti-inflammatory cytokines and immunosuppression dominate. 6, 8, 9 influenza deaths occur more frequently in older persons with cardiopulmonary conditions, diabetes, and renal disease, but as seen in the 2009 h1n1 pandemic, they also occur in younger adults with obesity, asthma, and in women who are pregnant. regardless of age, people with all of these conditions share one characteristic in common: they have chronic low-grade inflammation. in effect, their ''innate immune rheostats'' have been set at different, and perhaps more precarious, levels that make them more vulnerable to influenza-related complications. 10 laboratory studies of influenza virus infection confirm the importance of the host response. in several studies in mice in which the host response has been modified (e.g., cytokine knockout), survival has been improved without increasing virus replication in the lung. 5 in fact, severe disease can be induced without any influenza virus replication. for example, fatal acute lung injury has been induced in mice by inactivated (not live) h5n1 virus. 11 in this model, antiviral agents would be useless; only the host response could be responsible for disease. these observations raise the following question: could the host response be modified so patients with severe seasonal and pandemic influenza might have a better chance of surviving? influenza is associated with acute coronary syndromes, and influenza vaccination and statins reduce their occurrence. these associations led to the suggestion in 2004 that statins might be used to treat pandemic influenza. 2 other agents that might also be effective include ppara and pparc agonists (fibrates and glitazones, respectively) and ampk agonists (e.g., metformin). 5, 8 these agents have been studied in laboratory models of inflammation, sepsis, acute lung injury, ischemia ⁄ reperfusion injury, energy metabolism, mitochondrial function, and programmed cell death. the results of these studies cannot be reviewed in detail here, but the major findings for cell signaling are summarized in the table 1 . unfortunately, the results of experimental studies are not always clear cut. for example, in one study of influenza virus infected mice, il-10 was necessary for containing infection, 12 but in another study il-10 appeared to be harmful. 13 nonetheless, overall understand-ing of cell signaling pathways in influenza virus infections and the actions of statins, glitazones, fibrates, and ampk agonists strongly suggest that these agents could benefit patients with severe influenza. laboratory studies in mice infected with pr8 (h1n1) h2n2 and pandemic h1n1 viruses show that resveratrol, fibrates, glitazones, and ampk agonists reduce mortality by 30-50%, often when treatment is started 2-4 days following infection. 14-17 (resveratrol is a polyphenol found in red wine. it shares with these other agents many of the same cell signaling effects.) in h5n1-infected mice, treatment with celecoxib and mesalazine, together with zanamivir, showed better protection than zanamivir alone. 18 remarkably, these immunomodulatory agents have not increased virus replication. even more remarkable, in another model of a highly inflammatory and frequently fatal conditionhepatic ischemia ⁄ reperfusion injury -glitazone treatment ''rolled back'' the host response of ''young adult'' mice (8-10 weeks old) to that of ''children'' (3-4 weeks old). 19 this unique study suggests that immunomodulatory treatment might roll back the damaging and sometimes fatal host response of young adults with influenza to the more benign and rarely fatal response of children. several, but not all, observational studies have shown that outpatient statins decrease hospital admissions and mortality due to community-acquired pneumonia. 8 for influenza itself, preliminary evidence presented in october 2009 suggests that immunomodulatory treatment of influtable 1 . cell signaling targets that might be affected by immunomodulatory treatment of severe seasonal and pandemic influenza* down regulate pro-inflammatory cytokines (e.g., nf-kappab, tnfa, il-1, il-6) up regulate anti-inflammatory cytokines (il-10, tgfb) up regulate pro-resolution factors (lipoxin a4, resolvin e1) up regulate ho-1 and decrease tlr signaling by pamps and damps up regulate enos, downregulate inos, restore inos ⁄ enos balance and stabilize cardiovascular function decrease formation of reactive oxygen species and decrease oxidative stress improve mitochondrial function and restore mitochondrial biogenesis decrease tissue factor and its associated pro-thrombotic state stabilize the actin cytoskeleton in endothelial cells and intracellular adherins junctions, and thereby increase pulmonary barrier integrity and decrease vascular leak differentially modify caspase activation and apoptosis in epithelial and endothelial cells, macrophages, neutrophils and lymphocytes in the lung and other organs increase the bcl-2 ⁄ bax ratio in influenza virus-infected cells and prevent the apoptosis necessary for virus replication. *see references 2,5,7,8 for details. nf-kappab, nuclear factor kappab; tnfa, tumor necrosis factor alpha; tgfb, transforming growth factor beta; ho-1, heme oxygenase -1; tlr, toll-like receptor; pamp, pathogen-associated molecular pattern; damp, damage associated molecular pattern; enos, endothelial nitric oxide synthase; inos, inducible nitric oxide synthase. enza patients with severe illness could be beneficial. in a study of almost 4000 patients hospitalized with laboratoryconfirmed seasonal influenza, inpatient statin treatment reduced hospital mortality by 66%. 20 in these patients, the cell signaling effects of statin treatment, summarized in the table 1 , probably acted to reduce pulmonary infiltrates, maintain oxygenation, stabilize myocardial contractility and the peripheral circulation, reverse immunosuppression, restore mitochondrial biogenesis, and prevent multi-organ failure. achieving these clinical effects led to a decrease in mortality. because of the molecular cross-talk between statins, fibrates, glitazones, and ampk agonists, 5,8 similar clinical benefits might be expected from other members of this ''family'' of immunomodulatory agents. simvastatin, pioglitazone, and metformin are produced as inexpensive generics in developing countries. they are used throughout the world in the daily treatment of millions of patients with cardiovascular diseases and diabetes. global supplies are huge. because most people with influenza recover without specific treatment (this was true in 1918), not all patients would require immunomodulatory agents. instead, only those at risk of ards, multi-organ failure, and death would need to be treated. importantly, the cost of treatment for an individual patient would be less than $1.00 (d.s. fedson, unpublished observations). moreover, unlike vaccines they could be used on the first pandemic day. thus far, influenza scientists and the institutions that support their work (e.g., nih and cdc, national health agencies in many countries, the bill and melinda gates foundation, the welcome trust, and the world health organization) have shown little interest in immunomodulatory treatment. nonetheless, when more than 90% of the world's people have no access to influenza vaccines and antiviral agents, their physicians must have access to an effective ''option,'' especially one that might be lifesaving. research on immunomodulatory agents for influenza must involve investigators in many fields outside influenza science -those with expertise in the molecular and cell biology of inflammation, immunity, sepsis, cardiopulmonary diseases, endocrinology and metabolism, ischemia ⁄ reperfusion injury, mitochondrial function, and cell death. laboratory studies needed to identify promising treatment agents would probably cost $5-15 million (d.s. the results of these studies would inform clinical trials that critical care physicians are already eager to undertake. 21, 22 this work will be especially important for people in developing countries where critical care capacity is extremely limited and not likely to improve. 23 like critical care physicians, influenza scientists too must recognize that they cannot afford not to undertake research to determine whether generic immunomodulatory agents might be useful in managing severe seasonal and pandemic influenza. the nf-kappab-inhibitor sc75741 efficiently blocks h5n1 influenza virus propagation in vitro and in vivo without the tendency to induce resistant virus variants introduction influenza is still one of the major plagues worldwide. the appearance of highly pathogenic avian influenza (hpai) h5n1 viruses in humans and the emergence of resistant h5n1 variants against neuraminidase inhibitors highlight the need for new and amply available antiviral drugs. we and others have demonstrated that influenza virus misuses the cellular ikk ⁄ nf-kappab signalling pathway for efficient replication, suggesting that this module may be a suitable target for antiviral intervention. 1 here, we show that the novel nf-kappab inhibitor sc75741 efficiently blocks replication of influenza a viruses, including avian and human a ⁄ h5n1 isolates in vitro in concentrations that do not affect cell viability or metabolism. in a mouse infection model with hpai a ⁄ h5n1 and a ⁄ h7n7 viruses, we were able to demonstrate reduced clinical symptoms and survival of sc75741 treated mice. moreover, influenza virus was reduced in the lung of drug-treated animals. besides this direct antiviral effect, the drug also suppresses h5n1-induced overproduction of cytokines and chemokines in the lung, suggesting that it might prevent hypercytokinemia we hypothesise to be associated with pathogenesis after infections with highly pathogenic influenza viruses, such as the a ⁄ h5n1 strains. thus, a sc75741-based drug may serve as a broadly active nontoxic anti-influenza agent. to assess the number of infectious particles (plaque titers) in organs a plaque assay using avicel ò was performed in 96-well plates as described by mastrosovich and colleagues. 2 virus-infected cells were immunostained by incubating for 1 hour with a monoclonal antibody specific for the influenza a virus nucleoprotein (serotec) followed by 30 minutes incubation with peroxidase-labeled anti-mouse antibody (dianova) and 10 minutes incubation with true blueô peroxidase substrate (kpl). stained plates were scanned on a flat bed scanner and the data were acquired using microsoft ò paint software. the virus titer is given as the logarithm to the basis 10 of the mean value. the detection limit for this test was <1ae7 log 10 pfu ⁄ ml. organs of infected and control mice were homogenized and incubated over night in 1 ml trizol ò reagent (invitrogen) at 4°c. total rna isolation was performed as specified by the manufacturer (invitrogen). rna was solubilised in 50 ll rnase free water and diluted to a working concentration of 50 ng rna ⁄ ll. reverse transcription real-time pcr was performed using quantifastô sybr ò green rt-pcr kit and quantitect primer assays (qiagen) . all samples were normalized to gapdh and fold expression analyzed relative to uninfected controls. 3 ct values were obtained with the smartcycler ò (cepheid). to answer the question whether the nf-kappab inhibitor sc75741 shows antiviral properties against influenza virus, h5n1 infected mdck cells were treated with different concentrations of the inhibitor (figure 1 ). already treatment with 1 nm of sc75741 led to a reduction of viral cpe of more than 70%. almost 100% protection of cells was achieved when cells were treated with 50 lm sc75741. the results indicated that sc75741 has antiviral properties at concentrations ranging from 1 to 5 nm. we next tested whether sc75741 would also be effective in the mouse model of influenza virus infection. when h7n7 mice were treated i.v. once daily for 5 days with 5 mg ⁄ kg sc75741, survival rate of the animals increased significantly (p < 0ae05). the same results were found when h7n7 influenza virus infected mice were treated i.p. with 15 mg ⁄ kg sc75741 (data not shown). moreover, sc75741 treatment was not only effective when the inhibitor was given prior to h5n1 influenza virus infection, but also in a therapeutic setup when sc75741 was applied to the animals 4 days after infection (data not shown). since influenza virus infected mice showed increased survival after lethal infection, we next questioned whether the amount of influenza virus was reduced in the lung. therefore, we performed quantitative real-time (qrt) pcr to detect viral mrna. mice were treated with either sc75741 or the solvent, and 48 hour later the lungs were prepared to perform qrt-pcr. as shown in figure 2a the amount of viral mrna was reduced by 90% in sc75741 treated mice compared to solvent treated controls, indicating that sc75741 leads to a reduced expression of h5n1 specific mrna in the lung of infected mice. since infection of mice with h5n1 leads to hypercytekinemia, 4 we also investigated the expression of cytokines in sc75741 treated mice. as shown in figure 2b the amount of il-6 specific mrna was drastically reduced in sc75741 treated mice compared to solvent treated controls. moreover, also the expression of ip-10 was altered in sc75741 treated h5n1 influenza virus infected mice. here, roughly 90% reduction of specific mrna was detectable ( figure 2c ). thus, sc75741 leads to a reduced transcription of il-6 and ip-10 in h5n1 infected mice. there is an urgent need for new concepts to develop antiviral drugs against influenza virus. targeting cellular factors is a promising but challenging approach, and the concerns about side effects are obvious. however, it should be considered that drugs targeting viral factors, such as amantadine or oseltamivir, also exhibit a wide range of side effects in patients. thus, drug safety has to be rigorously tested in clinical trials regardless whether a drug targets a cellular or a viral factor. moreover, resistance against human h1n1 influenza viruses and highly pathogenic avian h5n1 virus strains to oseltamivir and amantadine have been reported. 5 in that respect, the strategy to target cellular factors 6,7 might be one way to ensure that new drugs against influenza virus will be useful and effective for a long time without causing the development of resistant virus variants. we were able to demonstrate that the nfkappab inhibitor sc75741 is able to reduce influenza virus activity in cell culture. moreover, the compound was also effective against highly pathogenic avian influenza viruses of the h5n1 and h7n7 subtypes in the mouse model. next to the reduction of virus sc75741 was also able to reduce h5n1-induced overproduction of cytokines and chemokines in the lung in the lung of mice after infection with h5n1. most importantly, the drug did not show any tendency to induce resistant virus variants (data not shown). thus, a sc75741based drug may serve as a broadly active non-toxic antiinfluenza agent. [1] [2] [3] [4] [5] in hong kong, the first confirmed case was a tourist from mexico reported on may 1, 2009. the local government made its first attempt to contain the spread of h1n1 in the local community by closing the metropark hotel where that tourist was staying, and quarantining 350 guests and staff for 7 days. following identification of the first local case around 6 weeks later on june 11, 2009, the government closed all kindergartens and primary schools from june 12 until early july. fever clinics were also opened, the alarm levels in hospitals were raised to the highest, and a public education campaign was implemented. previous studies of the community responses to severe acute respiratory syndrome (sars) and human-to-human h5n1 avian flu identified the importance of understanding the background perceptions of risk and psychological impact on the community. [6] [7] [8] [9] [10] in this study we investigated the psychological and behavioral responses of the general local community throughout the first wave of ph1n1, and we also examined the factors associated with greater use of preventive measures. 11 a total of 13 surveys were conducted between april and november 2009, covering the entire first wave of the ph1n1 pandemic. computer generated random-household telephone numbers from all land-based local telephone numbers covering over 98% of hong kong households were used to recruit a total of 12 965 local adults. one cantonese-speaking adult (age ‡18) was invited for interview in each selected household on the basis of a kish grid. the survey instrument was based on previous experience in sars and avian influenza projects. information, including knowledge on modes of transmission, psychological responses to pandemic influenza, preventive behaviors, attitudes towards the new vaccines and socio-demographics, was collected. informed consent was obtained prior to the interview. ethics approval was obtained from the institutional review board of the university of hong kong. descriptive statistics were weighted by sex and age based on the reference population data provided by the hong kong government census and statistics department. 12 multivariable logistic regression analyses were used to examine the association between the use of preventive measures and knowledge, perceptions and behaviors, sociodemographic characteristics, and psychological responses to pandemic influenza. multiple imputation was used to cope with a small proportion of missing data and make the best use of all available data. 13 statistical analyses were conducted in r version 2.9.1 (r development core team, vienna, austria). twelve thousand and nine hundred and sixty-five local adults were recruited throughout the study period, with a total of 127 715 telephone calls being made; the response rate among eligible participants was 69.9%. 11 hong kong entered the containment phase after the world health organization (who) announced a global alert, and policies including border screening, tracing, and quarantine of doi:10.1111/j.1750-2659.2011.00219.x www.influenzajournal.com suspected cases were implemented. hong kong transitioned to the mitigation phase on june 10, 2009 when the first local case was reported. the chronology of these and other events plus the epidemic curve of laboratory-confirmed ph1n1 cases are shown in figure 1(a) . the anxiety scores and risk perception of the respondents are shown in figure 1(b,c) . anxiety, measured by the state trait anxiety inventory, remained steady throughout the study period. in response to the announcement made by who and the unknown nature of the new virus, a higher proportion of the respondents expressed worry (more, much more, or extremely more worried than normal) if developed ili and perceived ph1n1 severity (same, more, or much more serious than sars) initially in early may 2009. fewer respondents reported worry if they developed ili as the pandemic proceeded, with a slight perturbation around the first deaths in july 2009 and a steady decline to 40.0%, while perceived severity of ph1n1 declined more dramatically after an early high. perceived risks of infection of respondents (absolute susceptibility) and risk relative to others (relative susceptibility) were also investigated and found to remain relatively stable throughout the first wave, with no indication of an increase during the period of peak ph1n1 activity in september (figure 1c) . as the first wave of ph1n1 progressed, knowledge on modes of transmission did not improve. on the contrary, later in the epidemic increasing proportions of respondents reported oral-fecal and cold weather as modes of transmission of ph1n1. around 35-40% of the respondents did not recognize direct and indirect contact or touching infected persons and contaminated objects as transmission routes for ph1n1 throughout the first wave ( figure 1d ). higher proportions of respondents avoided crowded places and rescheduled travel plans in the second half of june 2009 when local kindergartens and primary schools were closed and the first ph1n1-associated deaths were announced. social distancing measures such as avoiding crowded places and rescheduling travel plans remained stable with slightly decreasing trends thereafter. the use of hygiene measures and other social distancing strategies was relatively stable with slightly decreasing trends during the study period ( figure 2 ). female sex and older age were generally associated with greater reported use of hand hygiene measures, home disinfection, avoidance of crowded places, and rescheduling of travel plans. 11 female sex was also positively correlated with use of face masks and cough etiquette. we found a negative correlation between anxiety and use of all hand hygiene measures and cough etiquette, but a positive correlation between anxiety and use of home disinfection and (c) proportion of the respondents reporting higher worry if developed flu-like symptoms (more, much more, or extremely worried), higher perceived seriousness of h1n1 compared to sars (much more or more severe), higher probability to contract h1n1 over the next 1 month (certain, much more, or more likely), higher probability to contract h1n1 over the next 1 month compared to others outside family (certain, much more, or more likely). (d) proportion of the respondents identifying 5 possible modes of transmission as the actual modes of transmission of h1n1. social distancing measures. 11 other significant factors contributing to greater use of preventive measures were worry and knowledge. 11 greater worry was associated with higher probability of home disinfection, social distancing measures, and use of face masks. knowledge that h1n1 could be spread by indirect contact was associated all the investigated preventive measures, and knowledge that h1n1 could be spread by droplets was associated with cough etiquette, but not face masks. there were no consistent trends between all the investigated preventive measures and absolute and relative susceptibility. 11 community transmission emerged in hong kong in mid-june 2009, and prior to emergence of community transmission, perceived risk and perceived severity were high. as ph1n1 spread in hong kong, risk perception declined, even at the same time as incidence was increasing. anxiety was low throughout, at around 1.8 on the 4-point scale, compared to a maximum of 2.5 during sars on the same scale. 9 anxiety has been showed to be positively correlated to personal hygiene measures and social distancing in previous studies; 9,14 however, we found a negative correlation between anxiety and use of all hand hygiene measures, cough etiquette, and face masks, and a positive correlation between anxiety and home disinfection. 11 the differences in findings may be due to the fact that our anxiety measure was not specific to h1n1, and the score could be affected by other factors including economics. unlike hygiene measures, higher anxiety level, greater worry, and higher risk of perception were all associated with more social distancing. 6, 7, 9, 14 social distancing is the most direct strategy in avoiding infection from other people, and it is commonly observed in an outbreak that the general public avoids crowded places, travelling to other countries, and social gatherings, 9, 14 but the economic impact could be substantial. 15 as community incidence of h1n1 peaked, we did not observe any increase in use of preventive measures (figure 2) . we found that face mask use peaked at the early stage of the pandemic, while hand hygiene remained fairly constant, and the knowledge on the modes of transmission of ph1n1 did not improve over time. the lack of substantial change in preventive measures or knowledge about the modes of ph1n1 transmission in the general population suggests that community mitigation measures played little role in mitigating the impact of ph1n1 in hong kong. on the other hand, knowledge that ph1n1 could be spread by indirect contact was associated with all of the preventive measures studied. consistent with reports during the sars period, 9, 16 this study also showed that females and those of older age were more likely than others to use hygiene measures, avoid crowded places, and reschedule travel plans. this study has some limitations. first, this was a crosssectional study that was carried out at different time points, rather than a longitudinal study following the same individuals over time, and so the inferences on changes in behavior may need to be interpreted more cautiously. second, we recruited samples from all land-based local telephone numbers that cover 98% of hong kong households, but the response rate was not high enough to guarantee a representative sample, and this could be a source of selection bias. third, the responses were self-reported, and this may lead to social desirability bias in estimating knowledge, attitudes, and preventive behaviors. fourth, since the hong kong population has previously gone through unique experiences from sars in 2003 and avian flu in 1997, our results may not be comparable to other countries or settings. in conclusion, this study revealed that the ph1n1 pandemic failed to generate an increase use of preventive measures in the local community. there was no association between anxiety level and the events of the pandemic. with a relatively low mortality and morbidity rates compared to sars, ph1n1 was not a matter of concern in the hong kong community. the lack of substantial change in the use of preventive measures and improvement in knowledge on the modes of transmission of ph1n1 suggested that public health campaigns during the pandemic may not have had substantial effects on the general public. london is a major tourist destination, the seat of government and finance in the uk, and in 2012 will host much of the olympic and paralympic games. along with the rest of the global community, in 2009 and early 2010 london faced the challenges of responding to the first pandemic of the 21st century. at the time, nhs in london was composed of 72 organisations, including the london ambulance service, acute hospitals, mental health and primary care trusts, and the strategic health authority. while london's nhs is well practiced at responding to large, big bang incidents, the influenza a ⁄ h1n1v pandemic was a rising tide event that lasted many months. significant preparatory work had been undertaken prior to april 2009, which meant that the nhs in london was ready to respond. nhs london (the strategic health authority for london) led the response in partnership with local managers in all nhs organisations. the first uk cases of influenza a ⁄ h1n1v were reported in scotland on 27 april, with the first in london on 30 april. cases continued to increase, and the first wave peaked in london in july. cases reduced over the school summer holidays, but increased again when children returned to school at the start of september, and a second, smaller wave occurred. it is essential that the nhs learns from the 2009 ⁄ 10 influenza a ⁄ h1n1v pandemic to ensure it is prepared for future challenges. nhs london provided a standardised debriefing pack to all nhs organisations in the region to identify, capture, and learn lessons. each debrief event involved health and inter-agency partners to ensure all viewpoints were considered and brought together in a single local report. all local reports were compiled in an over-arching document, which brings together common themes to inform ongoing preparedness in the region. 1 the debrief process identified a number of common themes, such as the need for clear and appropriate communication, the importance of working with partners, and the benefits of strong and early leadership. however, differences between and within organisations were also highlighted; for example, some wanted more freedom for local decision making, whereas others would have preferred more stringently applied central direction. the following paragraphs considers individual areas assessed in the debrief process. command and control was in the main effective, with clear direction delivered from the national centre through nhs london to local nhs organisations. effective leadership is essential; the identification of senior local individuals to lead the response with teams of people to support them was critical. appropriate use of technology to communicate messages and coordinate command and control processes greatly aided the response. this included the development of the nhs london noon brief, a daily digest and associated web portal, and regular teleconferencing. key points are: • operational management at all levels must be considered in pandemic planning. • appointing an executive lead in each organisation was invaluable in the response. • pandemic flu planning for london must continue to be regionally led. communication is an essential component of the response to any incident. it must be clear, timely, and accurate. in the main, communication was excellent and met these criteria. one of the most challenging aspects was when messages from partner organisations differed, which occasionally led to confusion, unnecessary work, or frustration. the use of technology greatly aided communication across the region and supported the response; this included secure web sites, bluetooth, and text messaging etc. key points are: • regular internal communications and staff briefings are critical in the response to emergencies. • regular teleconferencing should be incorporated into future plans. • organisations should consider proactive and innovative methods for communicating during emergencies. robust partnership working was an essential component of pandemic preparedness work; however in the event, the a ⁄ h1n1v pandemic had little impact on sectors in london other than health. resilient communication networks between organisations, a common understanding, and the ability to make decisions were essential to the response at local level. ipcs proved an excellent mechanism to maintain local working relationships and resolve problems. clarity on the seniority of those attending these meetings and whether multi-site organisations such as mental health trusts should attend every ipc should be considered on a local and regional basis. key points are: • pandemic planning must remain part of inter-agency working. • social care resilience and planning must be embedded and integrated in health planning. 'vulnerable groups' is a universal term that covers a large and fluid group of individuals with different needs. ensuring access to healthcare during the pandemic for those who became vulnerable due to the situation, or those identified as such prior to the event, was the role of the pct in partnership with the local authorities. work continues to ensure that communication with vulnerable people is appropriate and timely in all incidents, and that organisations work together to achieve this. key points are: • planning to support the breadth of vulnerable people must continue. • pandemic preparedness for the prison sector should be further developed. • red ⁄ amber ⁄ green ratings for assessing vulnerabilities of mental health service users in an emergency should be further developed across the region. correct and appropriate usage of ppe is an essential component of reducing influenza spread, particularly in healthcare settings. london's nhs had been working towards developing local stockpiles of ppe when the pandemic commenced; however, there was little in place. the unanticipated national stockpile, while providing ppe to all organisations, was accompanied with some challenges in that it was often unfamiliar stock. key points are: • work around local stockpiling of non-standard consumables should continue. • regular training and fit testing of respirators should be embedded in all organisations. antiviral treatment was a core component of the response to influenza a ⁄ h1n1v, and was provided free of charge from a national stockpile. npfs reduced pressure on frontline nhs services once it was activated; however, there were concerns that patients could 'cheat' the system and obtain the drugs prior their clinical need. information about storage requirements of countermeasures must be clearly explained when they are delivered to frontline services, and the potential for recall into national stockpiles should be planned for. key points are: • regular exercising of local mass countermeasures centres and antiviral collection points (acps) should continue. • the use of community pharmacies as acps should be further considered in the capital. pandemic influenza vaccine uptake by healthcare workers was better than usual seasonal influenza uptake in the majority of nhs organisations, but could have been even better. this was largely due to the second pandemic wave not being as significant as expected, lack of clarity around when the vaccine would be delivered, and limited amounts being available initially. • gp-led and mass vaccination models for pandemic vaccination should be considered in local plans. • local lessons from the pandemic vaccination campaign should be applied to seasonal flu vaccination. the ability to maintain or increase capacity in response to a surge in demand, no matter what the cause, must be planned for. any of a number of situations could result in reduced staff or more patients, such as industrial action, transport disruption, disease outbreak, major incident, or poor weather. the work undertaken during planning for and responding to the pandemic will stand organisations in good stead for future disruptions. the importance of robust business continuity planning locally cannot be overlooked, as this is a key component of maintaining and increasing capacity. key points are: • local gp 'buddy schemes' should be encouraged for response to extreme pressure events. • organisations should regularly run staff skills audits so as to be aware of their overall capability for managing emergencies. • less emphasis should be placed on the use of retired staff when planning service continuity. reporting is a necessary but onerous task, and is often one of the most time-demanding parts of any incident response. it is also the aspect least likely to be tested through exercising. nhs london worked with organisations to endeavour to reduce reporting pressures, but much of this was dictated by central government. it is essential that future reporting requirements are proportional, informative, and realistic. while recognising it is not possible to predict the detail of information that may be requested, some broad assumptions can be made. key points are: • organisations should consider how they would collect and collate data from disparate parts of their organisation, rather than focussing on the detail of what that might be. • national and regional planning should consider the need for information and how this is balanced with the demand this places on organisations. • the introduction of the concept of a daily dashboard to identify areas of pressure should be incorporated into pandemic flu planning. the winter and pandemic influenza resilience assurance process undertaken in autumn 2009 was a useful process to inform planning for the first winter when the pandemic virus would be circulating in the uk. this consisted of a regional inter-agency exercise and a comprehensive review of the winter and pandemic plans of all nhs organisations in london. • regular assurance of pandemic flu preparedness should be maintained. • future resilience assurance processes should be undertaken in a timely and measured manner. • local organisations should continue to undertake regular pandemic flu exercises. the recovery period is as important as the response, but often receives minimal attention and has the potential to suffer as staff return to their normal jobs. one of the aspects that was not anticipated during the pandemic was the amount of stock (ppe, antivirals, and vaccine consumables) that would be recalled into national stockpiles. this proved particularly challenging for pcts who had to coordinate the process across their local areas. key points are: • the recovery period of an emergency must be given the same status and importance as the response. • future pandemic flu planning must include the recovery of national stockpiles of equipment and medicines. it is essential the lessons from the 2009 ⁄ 10 influenza a ⁄ h1n1v pandemic are learnt and embedded into business-as-usual and emergency response processes in preparation for the next pandemic and other incidents. even though the a ⁄ h1n1v pandemic was generally milder than previous pandemics, it still presented challenges to the nhs in london. the biggest challenge that remains is to ensure that the public and nhs staff are aware that a more virulent virus could cause significantly more illness, death, and disruption, and that we must maintain our preparedness should this happen. the influenza a ⁄ h1n1v pandemic has been a major stimulus to business continuity planning and emergency preparedness across health in london, and many of the experiences during the pandemic proved invaluable in the unusually severe weather in early 2010. it is important that this impetus and focus is maintained. changes to the nhs landscape in london will be considered in ongoing pandemic and emergency preparedness to ensure we remain as well prepared as possible for future events, particularly as london approaches the 2012 olympic and paralympic games. one of the major lessons learnt from all global pandemic events is that better preparedness of national health systems to deal with influenza viruses could make a significant difference. the way national health systems operate during inter-pandemic and the pandemic alert periods and the methods they use to address potential threats posed by zoonotic viruses with pandemic potential, as well as sea-sonal influenza epidemics, can clearly indicate whether the countries have enough capacities to respond adequately to unexpected influenza outbreaks. these public health decisions to ensure the maximum of efficiency require a robust scientific knowledge base. the who public health research agenda for influenza developed by the global influenza programme (gip) in cooperation with international influenza experts identified specific research topics and their importance in meeting stream-specific breakout discussion groups during the global consultation meeting included representatives of researchers and public health professionals. funding organizations were invited to observe the process with no direct participation in the deliberations. the methods used to design the research roadmap for an influenza pandemic scenario are closely related to the process of development of the final document of who public health research agenda for influenza. during a pandemic scenario, the group prioritized topics and questions relating to rapid action and response. five to 10 key public health needs associated with a pandemic scenario have been identified for each of the research agenda streams: five priority public health topics were identified for a pandemic scenario as follows: • examination of host range and transmission dynamics of animal influenza viruses to guide surveillance, control strategies, and risk communication. • enhanced surveillance in animals and humans to monitor virus evolution: o early detection of novel reassortants or changes in genotype and ⁄ or phenotype related to virulence. o development of epidemiological and laboratory diagnostic tools and capacity building to optimize case finding. o develop a framework for surveillance in animals that address ethical, legal, and social barriers to intra-pandemic surveillance and reporting. • deconstruct the origins of the pandemic virus to identify factors that permitted efficient human transmission. • develop strategies to limit economic, social, and cultural disincentives of animal-based interventions to reduce intra-and inter-species transmission. • operational research to optimize risk communication in the early phases of the pandemic linked to animal husbandry and food safety. stream 2: limiting the spread of pandemic, zoonotic and seasonal epidemic influenza ten priority research topics were identified for both pandemic and inter-pandemic scenario as follows: transmissibility of influenza across the progression of infection and spectrum of disease: • relative contributions of the different modes of transmission for influenza. five priority public health topics were identified for a pandemic scenario as follows: • identification of groups at higher risk of infection and severe disease outcome through enhanced surveillance. • understanding disease severity and identification of predictors of severe outcomes. • investigation of vaccine effectiveness, especially in high risk groups in diverse geographic areas. • establishment ⁄ enhancement of pharmacovigilance, particularly for adverse events among at-risk groups. • optimization of strategies for rapid and targeted vaccine deployment. • rapid assessment to optimize acceptance of pandemic vaccine. six priority public health topics were identified for a pandemic scenario as follows: • collaboration and coordinated sharing of data, protocols, regulatory, and other implementation strategies and databases from different countries on all aspects of patient management and outcome to accelerate improvements in patient care. • development of best practices in patient management in different settings, including checklists and algorithms for clinical care and treatment, prognostic parameters, and tests to predict potential for the development of severe disease. • rapid, reliable, simple, low-cost point-of-care diagnostic tools for influenza. • best use of current antiviral drugs and optimal formulations in different target populations, such as parenteral and other routes of administration for severe infections. • use of combination therapies, including use of adjunctive therapies (e.g., use of convalescent serum and immunomodulators). • role of ongoing viral replication, host responses, and the effect of co-infections in the pathogenesis of severe disease. modern tools for early detection and monitoring of disease the group on surveillance tools concluded that the agreed topics of interest were equally applicable during a pandemic or inter-pandemic period: • studies to appraise and adapt modern technologies for early detection of influenza outbreaks in surveillance at the human-animal interface. • develop, integrate, and evaluate innovative approaches for influenza surveillance and monitoring with other existing disease monitoring systems. • study efficient mechanisms on sharing data, clinical specimens, and viruses with consideration for local, ethical, legal, and research perspectives. • examine the timeliness and quality of data required for early detection from local to national and global levels for the respective stakeholders. five priority public health topics were identified for a pandemic scenario as follows: • identify environmental determinants of seasonal variation in influenza transmissibility in tropical and temperate regions. • estimate the transmission risk associated with types of contacts by comparing measured contact patterns with outbreak data. • incorporation of validated models of behavioral responses to risk and control measures in virus transmission. • development and implementation of novel technology for real-time sero-surveillance during a pandemic. • develop experimental and theoretical framework to assess host adaptation to study host receptor, antigenicity, and virulence. modern tools for strategic communication three priority public health topics were identified for a pandemic scenario as follows: • evaluate tools to more rapidly and accurately assess and monitor knowledge, attitudes, beliefs, and practices in different population groups to guide future communication efforts; develop tools and methods to more rapidly and accurately assess and monitor knowledge, attitudes, beliefs, and practices in different population groups, and thereby, guide future communication efforts. for communicating in different cultural settings, which engage and empower individuals and communities to practice and promote appropriate risk reduction measures. implementation of the identified research priorities is expected to underpin public health decision making at all levels with proven knowledge that will help to save large numbers of lives, reduce health costs and economic loss, and mitigate potential social disruption. complemented by an analogous research roadmap for a pandemic influenza scenario, the research recommendations for an interpandemic period represent a framework to provide evidence to guide public health policies on influenza control. one of the major lessons learnt from all global pandemic events is that better preparedness of national health systems to deal with influenza viruses could make a significant difference. these public health decisions to ensure the maximum of efficiency require a robust scientific knowledge base. the who public health research agenda for influenza 1 developed by the global influenza programme (gip) in cooperation with international influenza experts identified specific research topics and their importance in meeting public health needs for inter-pandemic periods according to its five key research streams: • stream 1. reducing the risk of emergence of pandemic influenza. • stream 2. limiting the spread of pandemic, zoonotic, and seasonal epidemic influenza. • stream 3. minimizing the impact of pandemic, zoonotic, and seasonal epidemic influenza. • stream 4. optimizing the treatment of patients. • stream 5. promoting the development and application of modern public health tools. stream-specific breakout discussion groups during the global consultation meeting included representatives of researchers and public health professionals. funding organizations were invited to observe the process with no direct participation in the deliberations. the methods used to design the research roadmap for an influenza inter-pandemic scenario are closely related to the process of development of the final document of who public health research agenda for influenza. during an inter-pandemic phase, a more comprehensive approach was applied to establish research topics and prioritizing a range of questions that will build a solid foundation to guide research activities to support public health decision making. five to ten key public health needs associated with an inter-pandemic scenario have been identified for each of the research agenda streams: stream 2: limiting the spread of pandemic, zoonotic, and seasonal epidemic influenza ten priority research topics were identified for both pandemic and inter-pandemic scenario as follows: 1. transmissibility of influenza across the progression of infection and spectrum of disease 2. relative contributions of the different modes of transmission for influenza 3. biological, behavioral, and social host factors that influence the risk of transmission and infection 4. patterns, drivers, and mechanisms affecting the seasonality of transmission 5. viral and population factors that influence transmission and spread of different influenza types, subtypes, and strains 6. strategies to reduce the transmission of influenza in community, household, and health care settings, especially in less-resourced areas 7. impact and cost effectiveness of social measures, such as school closures, and the role of surveillance in assessing timing of these interventions 8. impact, effectiveness, and cost effectiveness of individual measures, such as isolation and quarantine 9. role of vaccination in limiting the spread of influenza and strategies for its use 10. impact of antiviral treatment and prophylaxis in reducing transmission of influenza stream 3: minimizing the impact of pandemic, zoonotic, and seasonal epidemic influenza 1. identify higher risk groups and severe disease through surveillance; disease severity and identification of predictors of severe outcomes 2. evaluate vaccination preventable disease burden and the potential impact of immunization programs through vaccine demonstration projects 3. enhancement of the properties of existing vaccines, including duration and breadth of protection, safety, immunogenicity, and dosesparing 4. development of new vaccines and vaccine platforms, especially suitable for under-resourced country settings 5. study the effectiveness of vaccine strategies to reduce disease burden in children and other high risk groups in a wide range of settings 6. improved uptake and acceptability of vaccines for both seasonal and pandemic influenza seven priority public health topics were identified for an inter-pandemic seasonal influenza scenario as follows: inter-pandemic seasonal influenza scenario 1. research on the burden of severe disease with a focus on regionalspecific factors, such as the burden of tb and hiv and optimization of pandemic and management 2. development of new antiviral strategies and validation of surrogate endpoints which may aid in advancing understanding of disease progression 3. further clinical evaluation of current antiviral drugs, particularly in populations at risk 4. integration of seasonal influenza with pandemic preparedness; strengthen surveillance, health care systems, capacity, and preparedness planning 5. improving diagnostics (e.g., multiplex assays for viruses and bacteria), including antiviral resistance testing at point-of-care 6. dissemination of best practices, situation analysis, preparation for next epidemic (e.g., establish protocols for rotating stockpiles of antiviral drugs) 7. increased attention to basic science research such as studying immunomodulatory drugs five priority public health topics were identified for an inter-pandemic zoonotic influenza scenario as follows: inter-pandemic zoonotic influenza 1. antiviral susceptibility of circulating zoonotic viruses (e.g., h5, h9, h7 influenza viruses) 2. reassortment between zoonotic and human influenza viruses and the potential for inter sub-type spread of antiviral resistance and virulence modern tools for early detection and monitoring of disease the group focusing on surveillance tools concluded that the agreed topics of interest were equally applicable during both pandemic and inter-pandemic period: 1. identify modern technologies for early detection of influenza outbreaks as well as their application in surveillance at the human-animal interface 2. develop and evaluate innovative approaches for influenza surveillance and monitoring with other existing disease monitoring systems 3. studies to address challenges on data, clinical specimens, and viruses sharing with consideration for local, ethical, legal, and research perspectives 4. examine the timeliness and quality of data required for early detection from local to regional, national, and global levels role of modeling in public health decision making five priority public health topics were identified for an inter-pandemic seasonal influenza scenario as follows: 1. integration of genetic and epidemiological data to understand spatiotemporal spread to forecasts evolution for vaccine strain selection and to anticipate likely burden of disease 2. quantifying the relative contributions of different modes of transmission of human influenza and developing mechanistic modeling of transmission processes 3. research using data-capture technologies to characterize human contact and mobility patterns at local, regional, and global scales, and their correlation with transmission risk 4. integration of genetic, antigenic, and epidemiological analyses to optimize surveillance for newly emerging pathogens at the animal ⁄ human interface 5. identifying and quantifying human and environmental ecological, behavioral, and demographic determinants of the risk of cross-species transmission and pandemic emergence modern tools for strategic communication four priority public health topics were identified for an inter-pandemic seasonal influenza scenario as follows: 1. review of evidence and experience related to health crisis communication from fields to organize knowledge and support evidencebased practice in strategic communication 2. identify and develop tools to rapidly and accurately monitor knowledge, attitudes, and practices in different population groups and guide future communication efforts 3. identify and develop communication tools and approaches for cultural settings and communities to practice and promote appropriate risk reduction measures 4. understand the potential ethical, social, economic, and political communication in crisis and develop strategies to work within constraints while maximizing opportunities complemented by an analogous research roadmap for a pandemic influenza scenario, the research topic recommendations for an inter-pandemic period represent an important outcome of joint international efforts by who, academicians, and public health experts. implementation of the identified research priorities is expected to underpin public health decision-making at all levels with proven knowledge that will help to save large numbers of lives, reduce health costs, and economic loss and mitigate potential social disruption over a medium-tolong term period. the impacts of school resumption on the incidence of pandemic (h1n1) 2009 in school students introduction school closure is one non-pharmaceutical intervention that is often suggested in pandemic preparedness plans, and it was widely implemented in pandemic (h1n1) 2009 to reduce transmission amongst school students. however, from past epidemiological studies, the effect of school closure in reducing respiratory disease transmission was inconclusive. 1 given this public health intervention causes major disruption to the education system and potentially raises childcare issues to working parents, evaluating its effect in the recent pandemic is necessary to improve future pandemic planning. in hong kong, since school closure was implemented early in the pandemic and closure was effectively continued with the commencement of summer holiday, the lack of incidence data in the absence of school closure makes it difficult to analyse its effect directly. this has prompted us to analyse the situation indirectly from the angle of school resumption after summer holiday. in hong kong, public health surveillance on pandemic (h1n1) 2009 was effective from 25th april-30th september 2009: healthcare professionals were advised to report suspected cases of infection to centre for health protection, department of health, hksar, for further laboratorial confirmation. demographics of reported cases were subsequently recorded into a computerised system (the ''e-flu'' database). following institutional approval, a dataset of all confirmed cases diagnosed from may to september 2009 was obtained, which included the age, gender, confirmation date, and notification date of each report. all cases were classified into four defined socio-economic classes by age: pre-schoolers (0-5), school students (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) , adults (20-60), and retirees ( ‡61). assuming cases had contracted infection on the earlier date between confirmation and notification, daily incidence in each age class was counted for epidemic curve construction. upon observing an unusual rise in the epidemic curve of school students when school season resumed in september, interrupted time series analysis (also known as intervention analysis) 2 was applied to obtain the statistical significance of this observation. the analysis was applied to the incidence in school students from 12th july to 26th september 2009, which covered the period from the start of summer holiday to the end of the 4th week of new school season. incidence in school students before summer holiday was deliberately dropped since not all schools were closed when the school closure policy was effective: all primary schools were closed proactively, whereas secondary schools were individually closed on a reactive basis if students were identified to have contracted the infection. school activity was formulated as a step function, which takes value from 1st september 2009 onwards (st = 0: t < 1st september, st = 1 otherwise). a range of times series models were fitted by the maximum likelihood method and aic (akakine information criterion) was used to select the one with best fit. all computations were performed in sas version 9.2. a total of 3905 (14ae8%) pre-schoolers, 13758 (52ae1%) school students, 8383 (31ae8%) adults, and 338 (1ae3%) retirees were diagnosed with the infection in the surveillance period. the epidemic curves of preschoolers, school students, and adults showed a steady rise from 11th june onwards when local transmission of pandemic influenza was identified. an upsurge in the epidemic curve of school students can be observed in early september, coinciding with the commencement of the new school year (figure 1) . interrupted time series analysis on the epidemic curve of school students returned an arima(0,1,0) model with equations: where st, yt, yt denote school activity, predicted and actual incidence in school students on day t, respectively. standard error and significance for model constants were: 60ae27 (se = 33ae5, p = 0ae08), 2ae73 (se = 3ae84, p = 0ae48). in short, the model can be interpreted as: the number of infected school students rose by 2ae73 per day on average during the entire study period, with a sharp increase by 60ae27 coming into effect when the new school year began. time series analysis showed, at the marginally significance level, that daily incidence in school students had a major increase when school season resumed. on the assumption that the increase was not caused by any change in health seeking behaviour, this result suggests that school resumption had facilitated transmission amongst school students. on the basis that school activity significantly increases incidence of pandemic influenza in school students, this study suggests closures of schools in the early phase of pandemic (h1n1) 2009 and subsequently in the summer holiday probably had a major effect in mitigating transmission amongst school students. youngsters were postulated to be major vector for transmission in pandemic (h1n1) 2009. if this were true, it would be reasonable to expect the epidemic curves of the other age classes to show a similar upsurge when one is observed in school students. the absence of such observation in the epidemic curve of hong kong suggests school students were mostly disseminating the virus amongst themselves, but not to the other age groups. in november 2009, gip convened the first global consultation on a public health research agenda for influenza to identify key research topics in each of the five main streams of public health research. during this meeting, the scientific working group (swg) of the sub-stream in ''modern tools for risk communication'' identified the requirements in research during influenza pandemics and inter-pandemic periods to provide clear, credible, and appropriate messages which meet the needs of diverse communities. the swg suggested that who hold a follow-up workshop to assess the use of modern tools related to strategic and risk communication and to further promote research in these areas. communication'' in may 2010. one of the main objectives of the meeting was to generate a roadmap of public health research priorities related to strategic and risk communication. the research roadmap was developed by the group of invited experts on the basis of an analysis of available evidence and experience on public health and health crisis communication from relevant disciplines across global regions, as well as critical assessment of existing communication methods related to influenza control in different cultural, social, and ethnic settings. the workshop consisted of a series of presentations by experts in relation to experiences and lessons learned about communication during the sars, h5n1 epidemic, and h1n1 pandemic. there were also a series of group discussions on identifying research needs for pandemic and interpandemic periods in order to strengthen the research agenda. the expert group identified important public health needs in relation to communication during pandemics as well as in the inter-pandemic times. the main topics of discussion centered on communicating issues of influenza virus transmission, the use of influenza vaccines safety and efficacy, and use of antivirals as well as definition of the severity of the pandemic and the phase changes. in this context a number of research areas were identified, which can be broadly classified into four areas: understanding of communication principles and mechanisms is associated with an array of research topics covering different subject areas. one of the key questions here relates to the link between communication and ''behaviour change'' models and their application and appropriateness for different settings. the expert group defined the term ''behaviour change'' in this context as the modification of behaviour towards better health practices that are supported by clinical and scientific evidence for personal protection against infectious diseases and other adverse health risks. research topics related to these models require understanding and differentiating information and ''behaviour change'' needs of different audience segments, such as stakeholder mapping, target audience analysis, research into behaviour motivation, social norms, and the cultural, religious, social, legal, and political barriers and enablers of particular behaviors that are beneficial in influenza control. this research area also includes the analysis of media consumption among different audiences, role models, including ways to analyse how rumours and misinformation are spread, and ways to provide evidence-based information correctly. other important areas of investigation embrace methods to communicate uncertainty, learning how to build trust while communicating about a pandemic, and understanding what needs to be done before, during, and after a pandemic in order to create the best environment for influenza pandemic communication. critical key audiences identified for more intensive analysis were health workers, religious, public health, and societal (political and community) leaders. • investigation of the role of different communication channels and communication formats for different target audiences in a pandemic, particularly for groups that are ''hard-to-reach.'' • determining effects of perceptions related to pandemic influenza (severity, susceptibility, response efficacy, self efficacy, perceived social norms) on protective behaviours in different groups. • understanding audience in terms of their knowledge, preventive activities, and reasons why engaged ⁄ not engaged. • developing mechanisms to synergies between risk communication and behavior oriented approaches in the pandemic and inter-pandemic phases. • determining social, economic, cultural, and religious factors which support behaviours to limit spread and minimize impact in different settings. • identification of the key predictors ⁄ factors that influence people's behavior among different groups and populations vis-à -vis pandemic flu behaviors. • identification of elements that contribute to trust among populations and in different settings (country, public, professional, community), particularly where trust was previously compromised. • understanding psychology of different groups regarding their response to uncertainty, and finding the best way to communicate uncertainty. the research questions in this section relate to the planning, development, and evaluation of tools that can be quickly accessed and used in a pandemic situation. these may include communication materials and channels; the setting up of key stakeholder and champion communication networks; research protocols that are ready for rapid assessment during a pandemic or new communication tools. the use and understanding of terminology and language by both lay and professional groups and communities in planning for and ⁄ or reacting to a pandemic are important areas of research. acute examples, such as the naming of the viruses or the use of the word ''pandemic,'' illustrate this need well. the research focus of this area is to look at lessons learned from the a(h1n1) 2009 pandemic and to document and evaluate case studies, both looking at best practices, challenges, and barriers that were experienced. different communication strategies need to be evaluated and models to be built not only in terms of reach, but also in terms of impact on thinking, emotional response, and behavioural modification. a key question was how to prepare communication for a pandemic and how can the pandemic communication contribute to longer term ''behavioural change.'' mathematical modelling on gauging outcomes of such ''behaviour change'' would provide strategic approaches in risk communication. this section aims to answer the question whether the modeling, mapping, and scenario planning are actually useful in the pandemic situation. the expert group agreed that the research on the above issues should use a variety of methods and engage a number of disciplines. this would include literature reviews, case studies, trials, ethnographic studies, modelling, surveys, network analysis, as well as any other useful methodology. in an inter-pandemic situation for actual behaviour under pandemic conditions. • study the synergies and develop priority research topics on strategic ⁄ risk communication for influenza under inter-pandemic situations that includes zoonotic and seasonal infections. the who public health research agenda for influenza initiated and facilitated a multi-disciplinary discussion for communication during pandemic and inter-pandemic situations. it focused on both theoretical and practical issues to improve practice and ensure the health of the public for influenza. critical areas for research were identified to build evidence in this field. it was recognized that there are extensive bodies of knowledge in a number of disciplines, 2,3 such as health promotion, behavioural psychology, social sciences, social and behaviour change communication, social marketing, and communication for development relating to these questions, and that these should be explored. outcomes of these research activities are expected to widen the evidence base which will support developing communication strategies for influenza by countries, institutions, and individuals and will, consequently, help to improve public health world-wide. abstract background: cytokine dysregulation contributes to the unusual severity of h5n1 (reviewed in 1 ). previously, we demonstrated that interferon regulatory factor 3 (irf3) and p38 map kinase (p38) signaling pathways separately contribute to the induction of pro-inflammatory cytokines and chemokines in h5n1-infected cells. 2 here we investigate the role of innate sensing receptors in the induction of these cytokines and chemokines in response to h5n1 and seasonal h1n1 infection. materials and methods: human macrophages derived from peripheral blood monocytes were infected with h5n1 (483 ⁄ 97) or seasonal h1n1 (54 ⁄ 98) viruses. the role of innate sensing receptors in cytokine and chemokine induction by h5n1 virus was investigated using transient knock-down of these receptors with sirnas. the expression of innate sensing receptors in infected cells, and as a result of paracrine activation (by virus free supernatants of infected cells) of adjacent uninfected cells were also monitored by real-time pcr and ⁄ or western blotting. the involvement of janus kinase (jak) signaling pathways in these autocrine ⁄ paracrine cascades was investigated using a jak inhibitor. results: we previously showed that tnf-alpha, ifn-beta, and ifn-lambda 1 are the key mediators directly induced by the h5n1 virus in primary human macrophages with other cytokines and chemokines being induced as part of a secondary autocrine and paracrine cascade. here we demonstrated that retinoicacid-inducible gene i (rig-i) rather than toll-like receptor 3 (tlr3) plays the predominant role in h5n1-induced cytokines and chemokines in human macrophages via the regulation of irf3 and nf-kb nuclear translocation. in addition to the effects on virus infected cells, paracrine interactions between macrophages and alveolar epithelial cells contributed to cytokine cascades via modulation of jak signaling and by the upregulation of sensing receptors. conclusions: h5n1 directly induced tnf-alpha and ifnbeta mainly via rig-i signaling, and the subsequent activa-tion and nuclear translocation of irf3 and nf-kb in human macrophages. in addition to the effects on cytokine signaling, the innate immune sensing regulators themselves were also up-regulated by h5n1 infection, much more so than by seasonal influenza infection, via jak signaling. the up-regulation of innate sensing receptors was not limited to the infected cells, but was also found in adjacent uninfected cells through paracrine feedback mechanisms. this may lead to broadened and amplified cytokine signals within the microenvironment of the infected lung. a more precise understanding of the signaling pathways triggered by h5n1 virus leading to cytokine induction may provide novel options for the design of therapeutic strategies for severe human h5n1 influenza and also for treating other causes of acute respiratory disease syndrome. human h5n1 infection is associated with a mortality rate of more than 60%. the basis for the unusual severity of h5n1 disease has not been fully explained. cytokine dysregulation has been suggested to contribute to the disease severity of h5n1 (reviewed in 1 ). however, signaling pathways involved in the cytokine induction by h5n1 virus are not fully understood. previously, we demonstrated that irf3 and p38 map kinase (p38) are separate signaling pathways which contribute to the induction of pro-inflammatory cytokines and chemokines in h5n1-infected cells. 2 rig-i and melanoma differentiation-associated gene 5 (mda5) are important cytosolic sensors of nucleic acid of pathogens, while tlr3 and tlr8 also recognize nucleic acid species of pathogens, but they are localized at the endosomal membrane. 3 rig-i was found to be responsible for the recognition of influenza a virus infection, 4 and the transfection of vrnps induces ifn-beta expression. 5 while many studies have shown the role of rig-i in the induction of ifn-beta by influenza virus infection, the majority of these studies used either immortalized cell lines or mouse embryonic fibroblasts. there is a lack of data on the role of these innate sensing receptors in highly pathogenic avian influenza h5n1 infection in primary human cells in vitro, which are more physiologically relevant. furthermore, there is little data on the autocrine and paracrine up-regulation of these innate immune sensors following virus infection. human macrophages were obtained from peripheral blood monocytes by adhesion and differentiation in vitro for 14 days in rpmi medium supplemented with 10% autologous plasma. the cells were infected with h5n1 (483 ⁄ 97) or seasonal h1n1 (54 ⁄ 98) viruses at a moi of 2ae0. a549 cells were obtained from atcc and cultured in mem medium supplemented with 10% fcs and 1% penicillin and streptomycin. the role of innate sensing receptors in cytokine induction by h5n1 and h1n1 viruses was investigated using transient knock down of these receptors with sirnas in human macrophages as previously described 2 using specific sirnas purchased from qiagen. immunofluorescence staining assay of irf3 and nf-jb was employed to detect the nuclear translocation of these transcription factors after h5n1 infection. rabbit polyclonal antibodies against human irf3 and and nf-kb were obtained from santa cruz biotechnology. goat anti-rabbit igg antibody conjugated with alexa fluor 488 was a product of molecular probes. for investigation of paracrine effects on rig-i and tlr3 expression, culture supernatants collected from mock, 54 ⁄ 98 or 483 ⁄ 97 infected human macrophages were used to treat uninfected cells. the supernatants were first passed through a filter with 100-kda cut-off. virus particles as well as molecules with a molecular weight higher than 100 kda were retained and removed, while the filtrate was collected for treatment of uninfected cells. the expression of innate sensing receptors in infected cells and in adjacent uninfected cells following paracrine activation by virus free supernatants of infected cells was monitored by real-time pcr. the involvement of jak signaling pathways in these paracrine cascades was investigated using a jak inhibitor (calbiochem). we previously showed that tnf-alpha, ifn-beta, and ifnlambda 1 are the key mediators directly induced by the h5n1 virus in primary human macrophages with others being induced as part of a secondary autocrine and paracrine cascade. 2 in this study, we demonstrate that knockdown of rig-i or tlr3 led to the reduction of ifn-beta and tnf-alpha in human macrophages by both 54 ⁄ 98 (h1n1) and 483 ⁄ 97 (h5n1) infection. as shown in figure 1a , 483 ⁄ 97 virus induced higher level of ifn-beta mrna expression than 54 ⁄ 98 infection. cells transfected with rig-i or tlr3 sirna significantly reduced the expression of ifn-beta after 483 ⁄ 97 infection, by 63% and 29%, respectively. rig-i silencing also significantly reduced the ifn-beta expression in 54 ⁄ 98 infected cells by 52%. in contrast, silencing of mda5 or tlr8 did not suppress the induction of ifn-beta by either 54 ⁄ 98 or 483 ⁄ 97 infection; in fact, there was a slight (15%) increase of ifn-beta in cells transfected with mda5 sirna. based on these results we conclude that while both rig-i and tlr3 contribute to h5n1-induced interferon-beta induction in human macrophages, rig-i plays the dominant role. in order to investigate the relationship between these innate sensing receptors and the activation of transcription factors irf3 and nf-jb, we next measured the nuclear translocation of irf3 and nf-jb in cells with rig-i or tlr3 silencing after h5n1 infection. immunofluorescence staining assay on irf3 and nf-jb was performed and the number of cells with nuclear translocation was quantitated. the percentages of cells with nuclear translocation were plotted in figure 1b . we demonstrated that rig-i knockdown led to a significant reduction of irf3 nuclear translocation after 483 ⁄ 97 infection, whereas the nuclear translocation of nf-jb after 483 ⁄ 97 infection was significantly suppressed by rig-i or tlr3 silencing. these results suggest that the involvement of rig-i and tlr3 in the cytokine induction by 483 ⁄ 97 was via the regulation of irf3 and nf-jb nuclear translocation. since rig-i and tlr3 are important in influenza a virus-induced cytokine expression, we next explored the expression of these innate receptors in neighboring uninfected human macrophages by treating the uninfected macrophages with the filtered culture supernatants collected from mock, 54 ⁄ 98, or 483 ⁄ 97 infected macrophages. as shown in figure 2a , 483 ⁄ 97 supernatant differentially induced the mrna expression of rig-i, mda5, and tlr3 compared to 54 ⁄ 98 supernatant treated human macrophages. the induction of rig-i was higher than the induction of mda5 and tlr3. in the presence of 1 lm of jak inhibitor, the up-regulation of all three innate sensing receptors was significantly reduced showing their induction was dependent on jak activity. human lung epithelial a549 cells were also treated with the supernatants collected from macrophages infected with mock, 54 ⁄ 98, or 483 ⁄ 97 virus. differential induction of rig-i, mda5 and tlr3 by 483 ⁄ 97 supernatant compared to 54 ⁄ 98 supernatant treated cells was observed (figure 2b) . 483 ⁄ 97 supernatant dramatically induced all three innate sensing receptors, while 54 ⁄ 98 supernatant only marginally induced rig-i and mda5, but not tlr3. as in human macrophages, treatment with 1 lm of jak inhibitor caused a significant suppression of 483 ⁄ 97 supernatantinduced rig-i, mda5, and tlr3 expression in a549 cells. these results, taken together with the direct effects on virus infected cells, suggest that paracrine interactions between macrophages and alveolar epithelial cells contributed to cytokine cascades via modulation of jak signaling and by the up-regulation of innate sensing receptors. h5n1 directly induced ifn-beta ( figure 1 ) and tnf-alpha (data not shown) mainly via rig-i signaling and the consequent activation and nuclear translocation of irf3 and nf-kb in human macrophages. these results were consistent with a previous study using beas-2b cells showing the essential role of rig-i in ifn-beta reporter activity by h3n2 influenza virus infection. 4 while tlr3 also played a role in induction of ifn-beta and the activation of irf3 and nf-kb, it plays a less important role compared to rig-i. the reduction of irf3 and nf-kb activation was also confirmed with the study by le goffic 4 showing differential regulation of irf3 and nf-kb by rig-i and nf-kb can also be regulated by tlr3. in addition to the direct role of rig-i and tlr3 in sensing and signaling the presence of influenza virus, the innate immune sensing regulators were themselves also highly upregulated in both infected (data not shown) and adjacent uninfected cells by influenza virus infection. compared with seasonal h1n1 virus, the h5n1 viruses had a much more dramatic effect on inducing innate sensing receptors via jak signaling pathways activated by autocrine and paracrine mediators. the up-regulation of rig-i, mda5, and tlr3 was markedly induced by virus free culture supernatants from h5n1-infected macrophages, while supernatant from 54 ⁄ 98-infected cells induced the expression of these receptors only to a lesser degree. the soluble mediators in the virus infected cell supernatant caused paracrine upregulation of rig-i, mda5, and tlr3 in uninfected macrophages as well as human lung epithelial cells. these effects may lead to broadened and amplified cytokine signals within the microenvironment of the infected lung. taken together these results provide, at least, part of the explanation on the hyper-induction of cytokines in h5n1 infection. a more precise identification of the signaling pathways triggered by h5n1 virus leading to cytokine induction may provide novel options for the design of therapeutic strategies for severe human h5n1 influenza and also for treating other causes of acute respiratory disease syndrome. we generated mutants of y55 (h9n2) and a ⁄ duck ⁄ hokkaido ⁄ vac generation and characterization of mutant viruses rgy55sub (h9n2), rgvac1sub (h5n1), and rgvac1ins (h5n1), which have a serial basic amino acid residues at their ha cleavage sites were generated by site-directedmutagenesis and reverse genetics. rgy55sub (h9n2) and rgvac1ins (h5n1) required trypsin to replicate in mdck cells, and showed similar levels of growth to their parental viruses (table 1) . chickens intravenously inoculated with rgy55sub (h9n2) or rgvac1ins (h5n1) did not show any signs of disease. rgvac1sub (h5n1) replicated in mdck cells without exogenous trypsin, and one of the eight chickens inoculated with the virus showed slight depression at 1 day post-infection. the h9 and h5 mutant viruses were serially passaged in the air sacs of chicks to assess their ability to acquire pathogenicity. plaque formation in mdck cells and pathogenicity in 3-day-old chicks and 4-week-old chickens are shown in table 1 . rgy55sub (h9n2) replicated in mdck cells in the absence of trypsin and killed all of the chicks after six consecutive passages. two of the eight-four-weekold chickens inoculated intravenously with rgy55sub-p8 (h9n2) died within 5 days. eventually, over 75% of the chickens intravenously infected with rgy55sub-p10 (h9n2) died by 2 days post inoculation, and its pathogenicity was comparable to that of hpaivs. 10 rgvac1sub-p1 (h5n1) was pathogenic to both chicks and 4-week-old chickens, and mortality increased after one more passage. rgvac1ins-p1 (h5n1) replicated in mdck cells in the absence of trypsin, killed all of the chicks, and caused 75% mortality among 4-week-old chickens. the lethal effect of rgvac1ins-p1 (h5n1) on chickens increased with one additional passage in the air sacs of chicks, as in the case of rgvac1sub (h5n1). to examine whether the pathogenicity of each virus via the natural route of infection correlated with that by intravenous infection or not, three 4-week-old chickens were challenged intranasally with the viruses at an eid 50 of 10 6ae5 and observed for clinical signs until day 14 post-infection (data not shown). all chickens inoculated with rgy55sub-p10 (h9n2) or its parental viruses survived without showing any clinical signs, and serum antibody responses were detected in the hi test. on the other hand, rgvac1sub-p2 (h5n1) and rgvac1ins-p2 (h5n1) were pathogenic as in the intravenous experiment, killing two of three chickens by day 11 post-inoculation. one of three chickens were not infected with rgvac1sub-p2 (h5n1) or rgvac1ins-p2 (h5n1) via intranasal route (data not shown), indicating these p2 viruses had not been completely adapted to the host. to investigate the possibility of these p2 viruses to acquire further pathogenicity for chicken, rgvac1sub-p3 (h5n1) and rgvac1ins-p3 (h5n1) were obtained from the brain homogenates of the chickens that died on 11 days post intranasal inoculation with the p2 viruses. although mortality rate of chickens inoculated with the p3 viruses was equal to that with p2 viruses, enhancement of pathogenicity was observed in intranasal inoculation study; all of the chickens inoculated with rgvac1sub-p3 (h5n1) were infected, and time to death was shortened to 4-6 days post inoculation in chickens with rgvac1ins-p3 (h5n1) (data not shown). to investigate whether tissue tropism of the viruses was involved in their pathogenicity, we determined viral titers in the tissue and blood samples from 4-week-old chickens intranasally inoculated with each virus on 3 days post infection ( table 2) . rgy55 (h9n2) and rgvac1 (h5n1) were scarcely recovered from the samples, and the mutant strains before passage showed broader tissue tropism than the parental viruses. none of the chickens inoculated with rgy55sub-p10 (h9n2) showed any signs of disease, and viruses were recovered from each of the samples except the brain and the blood. one chicken inoculated with rgvac1sub-p2 (h5n1) showed clinical signs such as depression, and viruses were recovered from virtually all of its organs and blood samples. two of three chickens inoculated with rgvac1ins-p2 (h5n1) showed disease signs, and one died 2 days post inoculation. the viruses were recovered from almost all samples of the two chickens showing signs of disease. p3 viruses were efficiently replicated in systemic organs of the chickens as compared with p2 viruses. throughout the study, the viruses were recovered from the brains of all of the chickens showing clinical signs. here, we demonstrated that the h9 influenza virus acquired intravenous pathogenicity after a pair of di-basic amino acid residues was introduced into the cleavage site of the ha and passaged in chicks. rgy55sub-p10 (h9n2) killed 75% of chickens infected intravenously, and its pathogenicity was comparable to that of hpaivs (table 1) . however, chickens intranasally inoculated with rgy55sub-p10 (h9n2) did not show any clinical signs of disease (data not shown). these results are consistent with a previous study in chickens that found some h10 influenza viruses did not show intranasal pathogenicity although their intravenous pathogenicity index was over 1ae2, classified as hpaiv according to the definition by european union. 11 ohuchi et al. 12 reported that the insertion of additional basic amino acids into the h3 ha cleavage site resulted in intracellular proteolytic cleavage. other groups reported that h3 and h6 has tolerated amino acid mutations into their cleavage sites, and the viruses with the mutated has replicated in mdck and ⁄ or qt6 cells in the absence of trypsin. 13, 14 the results in the present study is in agreement with these, namely, cleavage-based activation by a ubiquitous protease is not restricted to the h5 and h7 has. the intranasal pathogenicity of the h9 and h5 mutants were different (data not shown), although these viruses similarly replicated in mdck cells in the absence of trypsin and killed chickens by intravenous inoculation ( table 1) . the viruses were recovered from the brain and the blood of some chickens infected with rgvac1 mutants (h5n1), and morbidity was closely associated with viral titers in the brain (table 2) . on the other hand, no viruses were recovered from the brain of chickens infected with rgy55 mutants (h9n2), explaining why rgy55sub-p10 (h9n2) did not show intranasal pathogenicity. all the viruses passaged in the air sacs of chicks killed chicken embryos by 48 hours post allantoic inoculation (data not shown). rgvac1sub-p3 (h5n1) and rgvac1ins-p3 (h5n1) were more pathogenic to chicken embryos than rgy55sub-p10 (h9n2); the allantoic fluid obtained from the embryonated eggs inoculated with the h5 viruses passaged in air sacs was turbid. it has been reported that infection of a highly pathogenic h7 virus were strictly confined to endotherial cells in chicken embryos or chickens. 15, 16 therefore, it is suggested that endotheliotropism differed between the h9 and h5 viruses passaged in air sacs and affected their intranasal pathogenicity. taken together, it is assumed that rgvac1sub-p3 (h5n1) and rgvac1ins-p3 (h5n1) showed marked intranasal pathogenicity with high levels of viremia caused by replication in vascular endothelial cells, leading to invasion of the brain. in the intravenous experiment, rgy55sub-p10 (h9n2) easily reached systemic organs, including the brain hematogenously, replicated through the cleavage of ha by a ubiquitous protease, and then exerted its pathogenicity. further study including a pathological analysis is currently underway to test this hypothesis. for all hpai viruses of subtypes h5 and h7 known to date, the cleavage of ha occurs at the c-terminal r residue in the consensus multibasic motifs, such as r-x-k ⁄ r-r with r at position p4 and k-k ⁄ r-k ⁄ t-r with k at p4, and leads to a systemic infection. early studies demonstrated that the ubiquitously expressed furin and pcs are activating proteases of hpai viruses. 1 furin and pcs cleave the consensus multi-basic motif r-x-k ⁄ r ⁄ x-r with r at position p4. 2 however, replacement of p4 r by k and a nonbasic amino acid significantly suppresses the processing activities of furin and pcs. 2 most of the type ii transmembrane serine protease identified so far recognize a single r at position p1, but the newly isolated mspl and its transcript variant tmprss13 preferentially recognize paired basic residue, particularly r and k at position p4, at the cleavage site. [3] [4] [5] thus, mspl and tmprss13 can activate various bioactive polypeptides with multibasic residue motifs, including fusogenic viral envelope glycoproteins. the present study was designed to characterize the proteolytic processing of the hpai virus ha by mspl and tmprss13 in comparison with furin. hpai virus a ⁄ crow ⁄ kyoto ⁄ 53 ⁄ 2004 (h5n1) 6 was isolated from embryonated eggs inoculated with tracheal homogenates from dead crows. then, the mutant ha sequence was constructed by changing r residue to k residue (n'-rkkr-c' to n'-kkkr-c') at the ha cleavage site by sitedirected mutagenic pcr as described. 7 we used human cell line ecv304, which expresses mspl and tmprss13 at levels below detection, and established the cells stably expressing mspl and tmprss13, such as ecv304-mspl and ecv304-tmprss13. 7 to determine the cleavage specificities of mspl ⁄ tmprss13 and furin, peptides (20 lg each) were incubated with 0ae25 mu mspl ⁄ tmprss13 for 1 hour and furin for 8 hours at 37°c, respectively. after incubation, the samples were separated by reverse-phasehigh-performance liquid chromatography (rp-hplc) with the use of a c 18 column. the elution samples were then identified by amino acid sequence analysis and by maldi-tof-ms. we analyzed the cleavability of 14-residue synthetic peptides derived from ha cleavage sites of hpai strains, such as a ⁄ chick ⁄ penn ⁄ 1370 ⁄ 83 (h5n2) 8 and a ⁄ fpv ⁄ rostock ⁄ 34 (h7n1), 9 and low pathogenic strain a ⁄ aich ⁄ 2 ⁄ 58 (h3n2). after incubation with human mspl or human furin, the digested samples were separated by rp-hplc, and peptide fragments were characterized by mass-spectrometry and protein sequencing. in contrast to the low cleavage efficiencies of the h3 ha peptide with a single r at the cleavage site ( figure 1a) , both the h5 ha peptide with the k-k-k-r motif ( figure 1b ) and the h7 ha peptide with the r-k-k-r motif ( figure 1c) were fully processed at the correct positions by mspl within 1 hour. in the case of h7 ha peptide with multiple basic residues, mspl cleaved two carboxyl-terminal sides of r in the cleavage site sequence of n'-k-k-rfl-k-k-rfl-g-c', while furin cleaved only at a single site of r with r at position p4, n'-k-k-r-k-k-rfl-g-c' in the presence of 1 mm cacl 2 . these cleavage site specificities of furin were consistent with that reported for the h5 ha peptide of hpai virus a ⁄ hong kong ⁄ 156 ⁄ 97 (h5n1) with r-k-k-r motif. 1 however, the h5 ha peptide with k at position p4 ( figure 1b) was hardly cleaved by furin under the same experimental conditions. tmprss13 showed similar results (data not shown). these findings suggest that mspl and tmprss13 cover diverse cleavage specificities, including non-susceptible specificity to furin. full length recombinant ha of hpai virus with kkkr cleavage motif was converted to mature ha subunits with membrane-fused giant cell formation in mspl or tmprss13 transfectant cells. 7 in addition, this conversion was suppressed by bowman-birk trypsin inhibitor, a membrane non-permeable highmolecular mass inhibitor against mspl ⁄ tmprss13. to test for the generation of infective virus, the conditioned media of 1-day culture of ecv304-wt and ecv304-mspl cells infected with wt and mutant hpai h5n1 viruses were inoculated into newly prepared cells and cultured for 24 hours. although spreading of wt virus infection with ha cleavage motif of r-k-k-r was detected from the conditioned medium of both ecv304-wt and ecv304-mspl cells, that of mutant virus with ha cleavage motif of k-k-k-r was only detected from the condition medium of ecv304-mspl cells. these results strongly suggest that the expression of mspl, but not furin, potentiates multicycles of hpai virus with k-k-k-r ha cleavage motif. seasonal human influenza a virus has have consensus monobasic cleavage site sequence, n'-q ⁄ e-x-rfl-g-c', and all hpai virus has have two types of cleavage site sequences with multiple basic amino acids, n'-r-k ⁄ r-k ⁄ r ⁄ x-rfl-g-c' with r at position p4 in a large number of hpai viruses and n'-k-k ⁄ r-k ⁄ t-rfl-g-c' with k at position p4 in a small number of hpai viruses. figure 1 shows furin efficiently cleaved synthetic hpai a ⁄ hong kong ⁄ 156 ⁄ 97 (h5n1) ha cleavage site peptide with the r-k-k-r motif, but hardly cleaved the hpai virus a ⁄ chick ⁄ penn ⁄ 1370 ⁄ 83 ha cleavage site peptide with the k-k-k-r motif. furthermore, cleavage of the full-length ha of hpai virus with r-k-k-r motif was detected, but cleavage of hpai virus ha with k-k-k-r motif was hardly detected in ecv304-wt cells containing furin ( figure 2 ). these substrate specificities of furin suggest that proteases other than furin and pc5 ⁄ 6 play a role in the processing of has of hpai virus with k-k ⁄ r-k ⁄ t-r cleavage motif. mspl and tmprss13 show unique cleavage site specificities of the double basic residues at the cleavage site, and r or k at position p4 greatly enhanced the efficiency, which none of the other ttsps have shown similar substrate specificities so far. furthermore, infectious and multicycle viral replication along with ha processing was also noted in genetically modified mutant recombinant live hpai virus a ⁄ crow ⁄ kyoto ⁄ 53 ⁄ 2004 (h5n1) with k-k-k-r cleavage motif in ecv304-mspl cells (figure 2) . these results were supported by the data of two cleaved peptides by mspl in figure 1c . these findings suggest that mspl has diverse cleavage specificities and may cleave ha at least two sites, although multiplicity of the mutant hpai virus was observed under the conditions. these results also suggest that mspl and tmprss13 in the membrane might potently activate the ha membrane fusion activity of hpai viruses and promote their spread. highly pathogenic avian influenza viruses replicate in various organs in birds, and the ha processing proteases might be widely distributed in these organs. indeed, tmprss13 and mspl are ubiquitously expressed in almost all human organs tested and are highly expressed in lungs, leukocytes, pancreas, spleen, and placenta. 4, 5 in addition, mspl and tmprss13 are strictly localized in the plasma membranes, suggesting that proteolytic activation of hpai virus ha occurs not only through the trans-golgi network by furin and pc5 ⁄ 6, but also on the cell surface by mspl and tmprss13. the pb1-f2 protein, which is translated from the +1 reading frame of the pb1 gene segment, has been linked to the pathogenesis of both primary viral and secondary bacterial infections in a mouse model. 1-3 a mitochondrial targeting sequence is located in the c-terminal portion of the pb1-f2 open reading frame, and expression of full length pb1-f2 has been associated with mitochondrial targeting and apoptosis in a monocyte dependent manner. 1, 4 it has been theorized that enhanced virulence could result from mitochondrial disruption with subsequent cell death mediated by pb1-f2. 4, 5 a suggested second function of the pb1-f2 protein is that it enhances immunopathology by triggering the inflammatory response. 3, 6 in earlier studies from our group, the pro-inflammatory phenotype was markedly upregulated when the pb1-f2 from the 1918 pandemic strain was expressed, arguing that this protein may be an important virulence factor for highly pathogenic pandemic viruses. 3, 6 in this report we analyze the pb1-f2 protein's contribution to pathogenesis in a mouse model, examining both inflammation and cell death. pb1-f2 proteins from a variety of epidemiologically important iav strains including all pandemic strains from the 20th century, a highly pathogenic avian influenza virus of the h5n1 subtype, and representative seasonal strains were utilized to determine the relevance to pandemic disease. we demonstrate that macrophage mediated immunopathology, but not apoptosis, are relevant functions of pb1-f2 proteins from past or potential pandemic influenza viruses. using the predicted amino acid sequences of the pb1-f2 proteins from pr8, a ⁄ brevig mission ⁄ 1 ⁄ 1918, 1 ⁄ singapore ⁄ 1 ⁄ 1957, a ⁄ hong kong ⁄ 1 ⁄ 1968, a ⁄ wuhan ⁄ 359 ⁄ 1995, and a ⁄ vietnam ⁄ 1203 ⁄ 2004, peptides from the c-terminal end were synthesized as described. 3 an additional n-terminal peptide was synthesized from the pr8 sequence as a positive control (mgqeqdtpwilstghistqk) as described. 3 a panel of viruses were reverse engineered as described 7, 8 and included laboratory strain pr8, a virus unable to express pb1-f2 (dpb1-f2 ⁄ pr8), or expressing the pb1-f2 of the 1918 pandemic strain (1918 pb1-f2 ⁄ pr8) or the truncated 1956 h1n1 strain (beij pb1-f2 ⁄ pr8). 3, 7 in addition, 7:1 reassortants encoding pb1 gene segments from a *current address: department of immunology and microbiology, university of melbourne, melbourne, vic., australia. 2004 highly pathogenic avian influenza of the h5n1 subtype (h5n1 pb1 ⁄ pr8), or from a 1995 human h3n2 strain (h3n2 pb1 ⁄ pr8) were utilized along with their isogenic deletion mutants for pb1-f2 (h5n1 dpb1-f2 ⁄ pr8 and h3n2 dpb1-f2 ⁄ pr8). cell lines and cell death assays raw264.7 cells were grown under conditions as described. 9 cells were infected with one multiplicity of infection (moi) of virus for 2-12 hours, or exposed to 50 lm (final concentration) of peptides derived from the c-terminal portion of pb1-f2 for 1 hour. cells from the supernatant and monolayers were harvested, washed, and stained with annexin (apc) and propidium iodide (pi) (becton dickinson, san jose, ca, usa), then analysed for cell death as described. 3 six-to eight week old female balb ⁄ cj mice (jackson laboratory, bar harbor, me, usa) were maintained in a biosafety level 2 facility in the animal resource center and procedures approved by the animal care and use committee at sjcrh. infectious agents and peptides were diluted in sterile pbs and administered intranasally to anesthetized mice (n = 6-10) in a volume of 100 ll (50 ll per nare) and monitored for overt signs of illness and weight loss daily. following euthanasia by co 2 inhalation, the trachea was exposed and cannulated with a 21 gauge plastic catheter (bd insyte; becton dickinson, sandy, ut, usa). bronchoalveolar lavage fluid (balf) was collected, red blood cell depleted, and cellular content analyzed via flow cytometry as described. 3 one way analysis of variance (anova) was used for multiple comparisons of cell death and cellularity of balf. a p-value of <0ae05 was considered significant for these comparisons. graphpad prism version 5.00 for windows (graphpad software, san diego, ca, usa) was utilized for all statistical analyses. to assess the contribution of pb1-f2 to inflammation, we utilized a panel of previously described reverse engineered viruses in the mouse infection model. 3, 7 the effect of pb1-f2 expression was observed clearly in the inflammatory infiltrate in response to infection in the lungs. deleting pb1-f2 from pr8 or expression of the c-terminally truncated beij pb1-f2 had a significantly reduced influx of macrophages ( figure 1a) . expression of the 1918 pb1-f2 caused similar inflammatory effects as the pr8 virus. disruption of pb1-f2 expression the virus containing the h5n1 pb1 gene segment in a pr8 background also significantly decreased the inflammatory response compared to the virus maintaining the ability to express full length pb1-f2 ( figure 1a) . however, no differences were seen that could be attributed to the 1995 h3n2 derived pb1-f2. the lungs of mice infected with the panel of pb1-f2 variant viruses were examined at 72 hours. pathologic changes typical of pr8 viral infection were observed in all lungs. these typical findings included perivascular inflammation, airway necrosis, hemorrhage, and deposition of cellular debris (figure 2 ). in the lungs of mice infected with pr8 or 1918 pb1-f2 ⁄ pr8, however, significantly more perivascular cuffing was noted, with a prominent increase in numbers of macrophages (figure 2a, c) . the overall number of inflammatory cells throughout the lungs, including both airways and alveoli, was quantitatively greater in these mice than in mice infected with dpb1-f2 ⁄ pr8 or beij pb1-f2 ⁄ pr8 ( figure 2b, d) . as the function and influence of pb1-f2 protein on normal viral function is not currently understood, and given the abrogation of enhanced inflammation induced by the truncated pb1-f2 beij ⁄ pr8 virus, we sought to elucidate whether the c-terminal domain of pb1-f2 could alone induce this inflammatory response. mice were exposed to a panel of peptides and were euthanized 24 hours later for collection of balf. significant influxes of macrophages into the balf were seen following exposure to c-terminal pb1-f2 peptides derived from pr8, the pandemic strains from 1918 (h1n1), 1957 (h2n2), and 1968 (h3n2), and the 2004 h5n1 virus compared to controls ( figure 1b) . similar effects were not seen with the peptide derived from a more recent h3n2 strain, a ⁄ wuhan ⁄ 359 ⁄ 1995. when peptide exposed mice were followed for morbidity for 7 days, peptides proven to induce a heightened inflammatory response correlated strongly with overt clinical signs of illness (data not shown). thus, the ability to cause lung inflammation appears to be a property of pb1-f2 proteins of viruses containing pb1 gene segments reassorted directly from the avian reservoir. the pb1-f2 protein may contribute to virulence by rendering the host cellular immune response ineffective through inducing apoptosis. 5 we sought to determine whether this was an epidemiologically important function for combating the host immune response to infection by testing the ability of pb1-f2 proteins from several different iav strains to cause cell death. we therefore infected raw264.7 cells with the panel of recombinant viruses at an moi of 1 for 2-12 hours. as has been demonstrated previously, 1,4,5 pr8 virus induces significant cell death compared to uninfected controls ( figure 1c ). when raw264.7 cells were infected with pr8 virus, necrotic death peaked 8 hours after infection. viruses lacking the c-terminal portion of pb1-f2, including the dpb1-f2 ⁄ pr8 and the beij pb1-f2 ⁄ pr8 were unable to cause cell death ( figure 1c ). in addition, expression of the 1918 pb1-f2 also did not cause significant increases in cell death over controls. expression of pb1-f2 or deletion of pb1-f2 in either an h3n2 or h5n1 pb1 gene segment background similarly did not alter the cell death phenotype. to examine additional strains for which we did not have isogenic virus pairs, we next exposed the balbcj mouse derived macrophage cell line raw264.7 to the panel of pb1-f2 peptides derived from pr8, the pandemic strains from 1918 (h1n1), 1957 (h2n2) and 1968 (h3n2), and the 2004 h5n1 for 1 hours. cell death in raw264.7 cells was caused only by the peptides derived from the laboratory strain pr8 and the peptide derived from the 1918 pandemic strain ( figure 1d ). viability was not affected by exposure of raw264.7 to peptides derived from other virus strains. we conclude from these data that the mechanism by which pb1-f2 contributes to the pathogenicity of pandemic influenza is unlikely to be through its reported ability to cause cell death. these data presented here demonstrate that the lung inflammatory response is enhanced by the influenza a virus pb1-f2 protein in a mouse model. this inflammatory response was characterized by increased cellular infiltration of macrophages into the interstitial and alveolar spaces of the lungs, as well as enhanced perivascular inflammation, airway necrosis, hemorrhage, and deposition of cellular debris. this augmentation was shown to be induced by pb1-f2 proteins only from those strains contributing to the formation of all pandemic strains of the 20th century and from the currently circulating, highly virulent h5n1 strains that constitute an imminent pandemic threat. the iav h1n1 strains circulating in humans since around 1950 code for a truncated pb1-f2. these viruses may lack the cterminal residues responsible for the inflammatory effects demonstrated in this publication. additionally, recently circulating h3n2 strains, in contrast to their pandemic forbear from 1968, have lost the capacity to cause pb1-f2 mediated inflammation through mutation of the c-terminus of this protein. in 2009 a novel h1n1 iav emerged from an animal reservoir and caused a human pandemic. disease burden from this strain has been considered mild 10 in contrast to the three pandemics of the 20th century. 11 the reasons for this disparity in pathogenesis are unclear. an examination of the origins of the three 20th century pandemics shows that only 2 the hemagglutinin (ha) and pb1 gene segments were reassorted directly from the avian reservoir in every case, suggesting gene products of one or both of these may be important. 12 the ha surface glycoprotein provided the antigenic novelty required for the each virus to achieve pandemic status. however, the significance of inclusion of a novel pb1 gene segment in each of the 20th century pandemics is not yet understood. we show here that the pb1-f2 of these pandemic strains contributes to virulence through induction of inflammatory responses. thus pb1-f2 may serve as a marker of the pathogenicity of pandemic strains. since the 2009 h1n1 strain codes a truncated pb1-f2 of only 11 predicted amino acids, the lack of pb1-f2 mediated inflammation may account in part for its relatively lower virulence. 13, 14 of the panel of pb1-f2 proteins studied, only that from the laboratory strain pr8 was capable of rendering responding host-immune cells ineffective by induction of cell death. we therefore hypothesize that molecular signatures specific to induction of apoptosis may have been lost through genetic mutation of the pb1-f2 gene throughout the evolution of the iavs. our findings suggest that this apoptotic function is unlikely to be important for the virulence of any of the known pandemics. rather, the inflammatory phenotype appears to be the dominant contribution of pb1-f2 to pandemic disease. influenza virus-cytokine-protease cycles are principal mechanisms of multi-organ failure in severe influenza and therapeutic approaches introduction influenza a virus is the most common infectious pathogen in humans, causing significant morbidity and mortality, particularly in infants and the elderly. mof with severe edema is observed in the advanced stage of influenza pneumonia. 1 however, the relationships amongst factors that induce vascular hyper-permeability in severe influenza remain unclear. it is reported that significant increases in levels of pro-inflammatory cytokine levels, such as tnf-a, il-6, and il-1b, affect host survival both positively and negatively. 2 the inflammatory response affects cell adhesion, permeability, apoptosis, and mitochondrial reactive oxygen species, potentially resulting in vascular dysfunction and mof. 3 in addition, iav infection up-regulates several cellular proteases including ectopic trypsin 4 and mmp-9. 5 up-regulated ectopic trypsin mediates the post-translational proteolytic cleavage of viral envelope hemagglutinin (ha), 6 which is crucial for viral entry and replication 7 and the subsequent tissue damage in various organs. 8 the aim of the this study was to define the pathogenic impact of cytokine storm in iav infection and the molecular mechanisms by which pro-inflammatory cytokines and proteases cause vascular dysfunction in animal model. weanling female mice aged 3 weeks (c57bl ⁄ 6crslc) were infected with iav ⁄ wsn ⁄ 33 (250 pfu) with and without treatment of pdtc (2.5 mg ⁄ kg), nac (10 mg ⁄ kg), and ndga (10 mg ⁄ kg). these inhibitors were administrated once daily for 4 days after infection. the levels of cytokines in tissue homogenates were measured by elisa kits. the effect of inhibitors on viral replications was determined by real-time pcr. gelatin zymography and western blotting were conducted as reported previously. 9 host cellular responses in the airway after iav infection figure 1 shows schematic view of typical biological responses in the airway of mice after iav infection. an initial response before viral proliferation is significant increases in pro-inflammatory cytokine levels. immediately after cytokine inductions, there is a marked up-regulation of ectopic trypsin along with an increase in virus titer in the airway, lung, and brain. 4 ectopic trypsin mediates the post-translational proteolytic cleavage of iav ha, which is crucial for viral entry and replication and the subsequent tissue damage in various organs. we also found that iav infection markedly induces mmp-9 and matrix degradation. 5 just after the peak of viral proliferation, the innate and adaptive immune responses of protective immunity are induced for defense and recovery, or oppositely on rare occasions, mof with vascular hyper-permeability is started into the advanced stage of influenza. the levels of tnf-a and il-6 in the lungs were increased persistently for 6 days after iav wsn infection, and that of il-1b peaked at days 4-6 post-infection (figure 2a ). since these cytokine responses are associated with activation of nf-jb and ap-1, we treated mice once daily for 4 days with anti-oxidant inhibitors: pdtc and nac against nf-jb activation, and ndga against ap-1 activation. pdtc and ndga significantly suppressed the up-regulation of tnf-a and il-1b (p < 0.001), and nac suppressed tnf-a (p < 0.001), and il-6 (p < 0.01) at day 4 post-infection. gelatin zymography showed up-regulation of ectopic trypsin and mmp-9 in mice lung, brain, and heart during infection for 4 days ( figure 2b ). trypsin and mmp-9 induction was inhibited by treatment with pdtc, nac, and ndga, probably via blockade of nf-jb and ap-1 binding in the promoter region of the genes. viral rna replication in various organs at day 4 post-infection was suppressed by more than one order of magnitude by pdtc, nac, and ndga ( figure 2c ). suppression of viral multiplication and induction of cellular factors by pdtc, nac, and ndga, significantly improved the survival of mice at day 14 post-infection, the late stage of infection ( figure 2d ). to elucidate the mechanisms underlying brain vascular dysfunction of influenza-associated encephalopathy, changes in the levels of tight-junction proteins, intracellular zonula occludens-1 (zo-1) and transmembrane occludin, and the matrix protein laminin, were analyzed by western blotting. marked reductions in the expression levels of tight-junction constituents were detected at day 4 post-infection, which were partly rescued by pdtc, nac, or ndga (figure 2e ). 9 no other tight-junction protein, claudin-5 or matrix fibronectin and type iv collagen, were affected. the present study reports several new observations: (i) proinflammatory cytokines, tnf-a, il-1b, and il-6, when up-regulated by iav infection, induce trypsin and mmp-9 expression in various organs in mice; (ii) inhibitors of nf-jb and ap-1 effectively suppress the up-regulation of proinflammatory cytokines, trypsin, and mmp-9 and improve survival rates of infected mice. based on these results, we propose the 'influenza virus-cytokine-protease cycle' hypothesis as one of the mechanisms of vascular dysfunction in mof with cytokine storm in severe influenza and influenza-associated encephalopathy. 9 the significance of pro-inflammatory hyper-cytokinemia, or 'cytokine storm,' in the pathogenesis of iav infection remains unclear. on the positive effects, cytokines promote lymphocyte activation and infiltration at the sites of infection and exert direct antiviral effects. however, on the negative effects of excess cytokines, the hyper inflammatory process evoked by viral infection may become harmful through intracellular activation of nf-jb, ap-1, and the janus kinase-signal transducers and activators of transcription signaling pathways. 3, [10] [11] [12] the in vivo experiments presented here showed that nf-jb and ap-1 inhibitors markedly suppress the expression of cytokines, trypsin, mmp-9, and viral replication, resulting in a significant increase in the survival of infected mice. furthermore, cytokines interact with mitochondria to increase the production of reactive oxygen species, resulting in the production ⁄ activation of vasodilatory mediators such as nitric oxide and bradykinin, and subsequent endothelial dysfunction and edema in various organs. 3 the molecular mechanisms underlying tight-junction disruption in endothelial cells and vascular hyper-permeability following the 'cytokine storm' remain unclear. tnfa up-regulation alters the cellular redox state, reduces the expression of four complex i subunits by increasing mitochondrial o 2 ) production and depleting atp synthesis, decreases oxygen consumption thereby resulting in mitochondrial damage, 3, 13 and increases [ca 2+ ] i 14 atp depletion dissociates zo-1 from the actin cytoskeleton and thereby increases junctional permeability. 15 endothelial dysfunction induced by 'influenza virus-cytokine-protease cycle' in the early stage of severe influenza may further affect various circulating factors, coagulation factors and complement systems, and vascular interacting cells, such as neutrophils, macrophages and lymphocytes. mof is the final outcome of metabolic and mitochondrial fuel disorder, immunosuppression, endocrine disorder, and tissue injury followed by endothelial dysfunction in many organs. another key pathway of acute lung injury in the highly pathogenic avian influenza virus h5n1and acute respiratory syndrome-corona virus infection reported recently involves oxidative stress and formation of oxidized phospholipids, which induce lung injury via toll-like receptor 4 signaling pathway. 16 in addition to these data, up-regulated trypsin and pro-inflammatory cytokines may also affect tissue destruction and immunosuppression in the late stage of iav infection. further studies are required on the role of the 'influenza virus-cytokine-protease cycle' in the pathogenesis of mof, particularly in the late stage of viral infection. though influenza a virus replication kinetics and host responses have been previously studied in umbilical vein endothelial cell or transformed endothelial cell lines, the tropism of influenza a virus including h5n1 and pandemic h1n1pdm for primary human lung microvascular endothelial cell has not been well defined. 1 in this study we employed primary human lung microvascular endothelial cells, which are more physiologically relevant for understanding pathogenesis of influenza in the lung as to obtain a better understanding of the links of endothelial cell infection to systematic virus dissemination and multiple organ involvement in severe human influenza. supernatants of cells infected at moi of two were collected for cytokine protein assays, and total rna was extracted for gene expression analysis using qpcr. we found that seasonal influenza h1n1 and h3n2 viruses initiated viral gene transcription and viral protein expression, but did not produce infectious progeny, while the highly pathogenic avian influenza h5n1 and the pandemic influenza h1n1pdm virus could replicate even with the absence of exogenous protease (figure 1) . furthermore, when compared to seasonal h1n1 and h3n2, the h5n1 virus was a more potent inducer of cytokine and chemokine including ifn-b, mcp-1, rantes, ip-10 (figure 2) , and il-6, in virus infected endothelial cells, whereas h1n1pdm induced intermediate levels of cytokine and chemokine. avian influenza h5n1 and pandemic h1n1pdm virus (but not the seasonal h1n1 and h3n2 virus) can productively replicate in human lung microvascular endothelial cells. this is likely to be of relevant to pathogenesis and provides a possible explanation for the extra-pulmonary infection seen in animal infection models. this extra-pulmonary spread may support the previous speculation and anecdotal evidence that h5n1 and h1n1pdm virus can infect the gastrointestinal tract through the virus dissemination from the infected respiratory tract as the first target cells for influenza infection. [2] [3] [4] in addition, the release of proinflammatory cytokine and chemokine induced by influenza h5n1 and h1n1pdm virus infection in lung microvascular endothelial cells may be important contributors to the pathogenesis of severe human influenza disease leading to endothelial cell dysfunction that contributes to severe pulmonary disease symptoms. during its replication, influenza virus utilizes the host cellular machinery for many aspects of its life cycle. characterization of such virus-host protein-protein interactions is a must to identify determinants of pathogenesis. the m2 ion channel protein plays a crucial role during the entry and late stages of the viral life cycle where its c-terminal domain, well conserved among influenza a viruses, is accessible to cellular machinery after fusion with endosomal membrane and during its trafficking along the secretory pathway prior to assembly and budding. 1 the aim of the study is to identify cellular interactants of m2 that play important regulatory roles during influenza infection. to identify cellular partners of m2 we performed a genome-wide yeast-two-hybrid (y2h) screening approach 2 using the cytosolic domain of m2 as bait and a human placenta random primed cdna library as prey and tested more than 60 million interactions. from the y2h screening, an interesting interaction with the human annexin a6 (anxa6) protein, 3 a member of annexin family proteins that binds to phospholipds in a ca 2+ -dependent manner, was identified. co-immunopre-cipitation of myc-tagged anxa6 and viral m2 proteins coexpressed in hek293t cells after transfection and infection confirmed the direct interaction between anxa6 and m2. we further investigated whether this interaction had any functional significance with regards to influenza life cycle. using a rna interference strategy to silence the anxa6 gene in human lung epithelial a549 cells, we observed increased progeny virus titers either in a single or multiple viral growth kinetics study, suggesting a negative regulatory role for anax6 during viral infection (figure 1 ). a novel interaction between m2 and anxa6 was identified. more functional studies are in progress to define precisely the potential negative regulatory role of this interaction during viral infection. a systematic dissection of the viral life cycle will be performed to identify the step(s) affected by the anxa6 cellular factor using specific assays such as real-time quantitative rt-pcr in a single or multiple viral growth kinetics study, cell transduction with ha-and m2-pseudotyped lentiviral particles, virion attachment and internalization assay, immunofluorescence staining of np protein as a marker of viral ribonucleoproteins localization, viral polymerase activity measurement, and viral budding observation by electron microscopy. rna extraction was achieved by qiagen biorobot ez1 prior to respiratory multiplex pcr analysis. what remained of the extracted material of each specimen was stored by refrigeration at 4°c. electronic patient records were searched for parameters, such as c-reactive protein (crp), white cell count (wcc), length of admission in days, and patient co-morbidities. patients were divided into three groups according to clinical severity: mild, moderate, and severe. the 'mild' group comprised of those admitted for three days or fewer, or not admitted at all. the 'moderate' group comprised those who required admission to hospital for more than 3 days as a result of swine flu, but who did not require admission to an intensive care unit (itu). the 'severe' group comprised those who had required itu admission. invitrogen '2· reaction mix': 0ae4 mm of each dntp + 6 mm magnesium sulphate. primer ⁄ probe mix recipe applied biosystems 7500 fast real-time pcr system, 'respiratory multiplex' program. well content 25 ll; thermocycler initial stage 50ae0°c for 15 minutes, then 95°c for 2 minutes. subsequent cycles of 95ae0°c for 15 seconds followed by 60°c for 33 seconds for 45 cycles. sequence detection software version 1.4 (applied biosystems). of 126 clinical isolates analyzed, all samples produced amplification of pdh material; 100 produced amplification of both swine flu and pdh material. human male dna (lot no. 36048611039 at 10 ng ⁄ l, applied biosystems) at concentration calculated at 16 666ae6 cells ⁄ ll was diluted from 10 )1 to 10 )4 , yielding mean average ct values of respectively 30ae10, 33ae60, 37ae78, and 37ae22. plotting log of cell number versus ct gave a y = mx + c line from which ct could be interpolated into cell numbers. for swine flu quantification, a sample of swine flu ct 19ae28 was diluted through 10 )1 to 10 )10 . it must be noted that due to variability in resultant swine flu ct values, repetitions at these dilutions were done using an rna carrier (1350 lg ⁄ l, qiagen; cat no. 1017794) in place of rnasefree water. the 10 )4 concentration was positive in nine out of 15 assays; this fraction was used in the calculation described by simmonds 2 to obtain a copy number of targets per reaction by the equation copy value = )ln(f), where f is decimal fraction of failure rate. here, f = 6 ⁄ 15 = 0ae4; )ln0ae4 = 0ae916 copies. a control curve was generated with ct values of 26ae73, 30ae02, 33ae67, and 37ae23 giving copy values of 916, 91ae6, 9ae16, and 0ae916, respectively. using excel (microsoft office, 2010), these control series were adapted into formulae to convert swine flu and pdh ct values into copy numbers of these elements per reaction. simple division derived a value for swine flu copy per pdh copy, but this was chosen to be expressed as swine flu copy number per 100 human cells. this will be referred to as the 'c' value. forty-two patients had known clinical details; average age was 29ae74, female to male ratio 63:37, and average admission length of 7 days. of the mild group (n = 26), nine cases were not admitted to hospital. of the remainder, the mean average admission length was 2ae5 days. mean average c value for all samples was 1ae49 · 10 6 , with a standard deviation of 1ae49 · 10 7 ; geometric mean was 4ae28, and median average was 6ae45. log(mean average c value) is shown for each severity group and for identified risk factors in the 'mild' severity group (figures 2a, b respectively) . in each case variation was too great to yield statistical significance. figure 1 shows the range of c values observed in the 'moderate' severity group. > -4 · 0 ; < -4 · 5 > -3 · 5 ; < -4 · 0 > -3 · 0 ; < -3 · 5 > -2 · 5 ; < -3 · 0 > -2 · 0 ; < -2 · 5 > -1 · 5 ; < -2 · 0 > -1 · 0 ; < -1 · 5 > -0 · 5 ; < -1 · 0 > -0 · 0 ; < -0 · 5 > 0 · 0 ; < 0 · 5 > 0 · 5 ; < 1 · 0 > 1 · 0 ; < 1 · 5 > 1 · 5 ; < 2 · 0 > 2 · 0 ; < 2 · 5 > 2 · 5 ; < 3 · 0 in a study by duchamp et al., 3 no significant correlation was observed between viral ct value and presence or absence of cardiaorespiratory disease, myalgia, digestive symptoms, or upper or lower respiratory tract infection (although a trend was observed towards patients presenting with signs of upper respiratory tract infection). to our knowledge, no other study has used a dual pcr for analysis of respiratory virus concentrations, and no study has attempted to correlate biochemical markers with respiratory virus concentration. the data exhibited a spectrum of c values, from values <5 · 10 )5 to over 1 · 10 8 . the three severity group standard deviations all overlapped with each other, preventing statistical significance. analysis of co-morbidities showed a high mean average c value when asthma was present (6ae05 · 10 7 ), but again this was associated with an excessive standard deviation. whereas the median average c value in the presence of asthma was higher than the overall average c value (9ae09 versus 5ae85), it was significantly lower than the median c value when no co-morbidity was documented (15ae55). there are multiple caveats that may be the cause of such variety of c values obtained. the duration between initial rna extraction and study pcr had a range of 21 to 285 days, with mean average delay of 211 days. the degradation of viral rna is an important contributor to assay variance and failure; rna degradation in clinical samples has been studied. [4] [5] [6] degradation of human dna in clinical samples may have occurred. several studies have chartered degradation of stored human dna. 2, 7 with regards to sampling, the clinical collection of throat swabs is naturally variable according to the method of the collector. a small number of bronchoalveolar lavage samples were analyzed, yet did not amplify, presumably due to rna degradation. the upper respiratory tract may be only a physical stepping stone for the virus, and take no further role in pathogenesis of severe disease (although undoubtedly is crucial for transmission). interestingly, a ferret study of pathogenesis observed that swine flu yields from the upper respiratory tract were greater than those given by ordinary seasonal h1n1, with consequently increased shedding. 8 the review by mansfield 9 cites significant findings regarding influenza pathogenesis, including the predilection of h5n1 strains for type ii pneumocyte cells and alveolar macrophages. it also highlights the limitation of knowledge through dearth of human autopsy studies; an exception is the recognition of haematophagocytic syndrome in severe cases. it is known that specific immunoglobulin is effective against establishment of infection in the upper respiratory tract, whereas specific cytotoxic t lymphocytes (ctls) are necessary for clearance of the virus from the lower respiratory tract. 10 it is also suggestive that a gap of two whole days transpires between initial infection and instigation of a specific immune response. 11 it is plausible that in the healthy individual, virus progression is confounded by efficient natural mucosal immunity, in part through good secretory immunoglobulin levels. airway inflammation associated with asthma exacerbation is known to increase both risk of respiratory viral infection and poorer outcome. it is unproven but likely that the local inflammatory processes give rise to increased virion burdens in the upper airways; however, the same effect is conceivable for epithelial cell turnover. there will likely be variance within each clinical category due to patient circumstances and clinicians' judgment of required admission. unfortunately, the duration of symptoms prior to swab collection was often omitted in the clinical notes. finally, stratification of patient group by receipt of antiviral treatment was not studied. no correlations were observed with c values and crp, wcc or admission length. trends were observed towards higher c values in 'mild' cases, but without statistical significance. the relative small study size, coupled with the intrinsic variability of the parameters studied, warrants larger, better controlled, prospective studies to elucidate clinical use of the c value for influenza illness prediction and management. in mid-april 2009 a novel variant of a(h1n1) influenza virus began to spread rapidly throughout the world, causing the first pandemic of the 21st century. the majority of the cases associated with this new virus show to be mild, but severe and fatal cases have been reported. molecular markers associated with severity have already been identified, as is the case of the mutation d222g. 1 resistant viruses to antiviral drugs have also been identified, highlighting the importance of rapid determination of the antiviral drug profile. global a(h1n1) 2009 genetic characterization, molecular evolution dynamics, antiviral susceptibility profiles, and inference of public health implications require nation and region wide systematic analysis of circulating virus. the objective of this ongoing research study was, primarily, to thoroughly characterize the genetic profile and evolution of the emergent influenza a(h1n1) 2009 virus circulating in portugal and its phenotypic expression on antiviral drugs susceptibility. the cases considered in this study were obtained from the community and from two collaborating hospitals in lisbon -a reference hospital for adults (hospital de curry cabral) and a reference hospital for children (hospital dona estefânia). the cdc real-time pcr protocol, recommended by world health organization (who), was the method used to confirmed all influenza a(h1n1) 2009 cases. from a total of 577 a(h1n1) 2009 positive cases diagnosed and confirmed, 163 were selected for this study, taking in consideration that they should cover the period of epidemic activity in portugal and include cases from persons belonging to risk groups and cases associated with more severe clinical features. ninety-six a(h1n1) 2009 strains were isolated in mdck-siat1 cells, from combined naso-oropharyngeal swabs. for the evaluation of the genetic profile of a(h1n1) 2009 virus circulating in portugal, 37 of the 96 isolates were characterized by genetic analysis of the ha, na, and mp genes. the remaining five gene segments (pb1, pb2, pa, ns, and np) were also sequenced for six of this 37 isolates. briefly, sequencing was performed according to the protocol developed by cdc and recommended by who, 2 using bigdye terminator v.1.1 technology. nucleotide sequences were determined in a dna automatic sequencer abi prism 3130xl genetic analyzer. for each genomic segment, genetic analysis was performed with lasergene v.4.05 software (dnastar inc, usa) using an average of 4-6 overlapping readings, including sense and antisense, for precise nucleotide and amino acid sequence determination. genetic mutation and phylogenetic analysis were performed by neighbor-joining method, using mega4.0 software, against published sequences from the vaccine strain (a ⁄ california ⁄ 7 ⁄ 2009) and from selected a(h1n1) 2009 strains available on gisaid epiflu database. all mutations were identified with reference to the vaccine strain genome sequence. antiviral drug susceptibility profile of a(h1n1) 2009 influenza virus circulating in portugal was evaluated both phenotypically and genotypically for nais and genotypically for amantadine. phenotypic evaluation to nais, oseltamivir and zanamivir, was performed for all 96 isolates by ic 50 determination through munana fluorescence assays. 3 genotypic evaluation was performed by searching for mutations associated with resistance to nais in all 37 na gene sequences. amantadine susceptibility profile was performed for all 96 isolates by searching on m2 sequence for the 5 molecular markers associated with resistance to this antiviral drug (l26f ⁄ i; v27a ⁄ d; a30t; s31n; g34e). genetic characterisation of the ha1 subunit of ha reveals point mutations in different strains. all 37 analysed strains present p83s and i321v mutations, which distinguish them from the vaccine strain ( figure 1a ). thirty-three of the 37 sequenced strains group in the s203t branch. this mutation is referred in the literature as being associated with the putative antigenic site ca. 4 most of these strains (19) further subgroup in the d222e branch, this mutation being associated with one loop of the receptor-binding site. 1 from the early to the late epidemic period, an increased circulation of virus carrying the mutation s203t was observed. this is in agreement with the association between this mutation and an enhanced viral fitness that is described in the literature. 5 additional mutations were also observed in a small number of virus, of which we highlight: regarding the genetic characterisation of na, the majority of strains analysed (34 of 37) presents the mutations n248d and v106i ( figure 1b) . as mutation s203t in ha gene, these two na mutations are described in the literature as associated with enhanced viral fitness. 5 the few strains not carrying these mutations have circulated in the beginning of the epidemic period. fifteen of the 37 analysed strains further subgroup in y155h branch. additionally, mutation i223v was identified in two strains. for the remaining gene segments available for the six analysed strains, the observations include: (i) no previously described virulence markers in pb2, pb1-f2, and ns1 were detected; (ii) pb1-f2 protein is present in the truncated form of 11 amino acids; (iii) the presence of mutations i123v and l122q in ns1 and v100i in np; (iv) the described association of mutation i123v in ns1 and v100i in np genes with viral fitness. phenotypic evaluation of nais susceptibility revealed the existence of three minor and two major outliers to oseltamivir ( figure 2 ). the two minor outliers exhibited a reduction of approximately twofold in the susceptibility to this antiviral drug, comparing to the baseline level, while the reduction exhibited by the two major outliers was of approximately three-and fourfold. regarding zanamivir, two minor outliers were identified with a reduction of approximately twofold in the susceptibility, compared to the baseline level. these two minor outliers (a ⁄ portugal ⁄ 17 ⁄ 2009 and a ⁄ portugal ⁄ 82 ⁄ 2009) correspond to the two major outliers identified for oseltamivir. genetic analysis revealed the presence of the mutation i223v in the na sequence of these two strains. the contribution of this mutation for the profile of reduced susceptibility identified for both nais is not known, but a mutation in the same na position (i223r) has been referred to as being associated with a reduction in nais susceptibility. 6 full genome sequence analysis of these strains shows that both strains also present the v480i mutation in pb2 gene. however, no association of this mutation with antiviral drug susceptibility is referred in the literature. concerning genetic evaluation of susceptibility to amantadine, all 96 analysed strains present a serine in position 31, which is a molecular marker of resistance to m2 inhibitors. these preliminary results allow us to discuss several points. however, the additional data that is being obtained through this ongoing study will be essential for a more complete analysis. for example, more information is needed to determine if the mutations found alter the biology and the fitness of the virus or if there are associated with an increased prevalence of the virus. the majority of the mutations identified in ha1 subunit have been detected in a(h1n1)2009 strains distributed throughout the epidemic curve, not evidencing a specific evolutionary trend. this is in agreement with the genetic and antigenic homogeneity that has being described for a(h1n1)2009 virus. 7 the occurrence of mutations in the position 222 of the ha1 subunit of a(h1n1)2009 virus have been described. however, more studies are needed to clarify the outcome of these mutations, as for example in patients with severe complications. it could also be relevant to investigate the presence of single and mixed variants in viruses and in clinical specimens and the possibility of these mutations affecting the binding specificity. regarding the susceptibility of a(h1n1)2009 pandemic viruses to antiviral drugs, all analysed strains were found to be resistant to amantadine. this resistant profile was not unexpected since the mp gene from this new variant had originated in the eurasian swine lineage, which is characterised by being resistant to this antiviral drug. 8 the majority of the a(h1n1)2009 strains analysed revealed to be susceptible to both nais, with only five strains exhibiting a profile of reduced susceptibility, three to oseltamivir and two to both nais. for these last two, the presence of the i223v mutation in the na sequence could explain the reduction observed, but a more complete analysis is needed to confirm this. the french national pandemic plan includes an early containment phase followed by a limitation phase. the efficacy of such a plan depends on pre-existing surveillance and laboratory networks. the grog community surveillance network and the hospital lab networks organized by the two french nics carried out the virological monitorthe efficacy of such plan depends on pre-existing influenza surveillance and laboratory networks. in france, the community surveillance is carried through the grog surveillance network. in addition, surveillance is also carried out in hospitals by the renal network. this renal network is divided in two sub-networks: the so-called h5-labs network, activated during the containment phase and the extended renal lab network activated in the limitation phase. the h5-labs have bsl-3 facilities that can be used for diagnosis purposes. as part of the national influenza surveillance system led by the french institute for public health surveillance (invs), the grog community surveillance network and the lab networks linked to the two french nics carried out the virological monitoring of the a(h1n1)2009 pandemic from the early containment phase up until the end of the pandemic phase. during the containment phase, all suspected cases were hospitalized and declared to invs. each patient was tested on the same day by specific virological diagnosis. hospital admission was not mandatory during the limitation phase, (i) the clustered cases were monitored to study transmission chains, and (ii) the circulation of the virus in the community was monitored through grog swabs collected by practitioners. the nics organized the influenza surveillance to fulfill several objectives according to the epidemiological situation. first, rt-pcr tools (influenza a m gene rt-pcr and a(h1n1)2009 specific h1 and n1 genes rt-pcrs) were developped and distributed to the lab networks on the 14th of may 2009. 1 from the early phase, the nics and the h5-lab network analyzed all the samples collected from hospitalized and community patients. during the early phase of the limitation phase, an increasing number of labs were performing the specific assays. when the pandemic wave started, all hospital labs could do the testing. results were centralised by nic and reported on a weekly basis. in addition, nics carried out the monitoring of antiviral resistance emergence (na pyrosequencing, specific h275y rt-pcr, and phenotypic assays), and real-time surveillance of genetic changes involved in virus adaptation (pb2) virulence factors or antigenic variations (ha). this sequencing was carried out by the pf8 sequencing platform of the institut pasteur. the first imported a(h1n1)2009 influenza cases were observed from the 28th of april 2009. a limited number of cases have been reported in may. local transmission could be detected end of may. clusters were observed in schools in june and in summer camps during summer. as opposed to the epidemiology of the a(h1n1)2009 virus in other european countries, no summer wave was observed in france. only a limited number of sporadic cases were reported up until october. early september, a significant number of cases presenting with influenza-like illness was reported (figure 1 ). the virological investigation of these cases showed high prevalence of rhinovirus infection. this circulation of rhinovirus was a counfounding factor of the pandemic. the pandemic wave lasted 11 weeks between mid-october and the end of december (week 43 to week 53, figure 1 ). the pandemic wave started week 42-43 in the ile-de-france area, and only week 44-45 in the rest of france. the peak was recorded week 48 ( figure 1 ). the impact of the pandemic was mainly observed in the 5-15 years group of age. overall, 1334 severe cases have been admitted to the hospital, and 308 deaths have been recorded by the end of the pandemic wave. the major impact was observed in the 15-65 years group of age (66% of deaths recorded). amongst the severe cases and the deceased cases, 20% and 16% of cases had no risk factor, respectively. these specimens, 24 279 were positives for h1n1, representing 99ae7% of total influenza virus detections. only nine brisbane-like h1n1, 62 brisbane-like h3n2, and eight b viruses have been detected in the same period of time. the weekly positive rate ranged from 0% to 48%. phylogenetic and antigenic analyses of the viruses collected during the pandemic wave did not show any emerging genetic or antigenic variants (figure 2a,b) . eight patients, all among cases presenting with severe illness, were infected by a virus harbouring the d222g mutation in the ha. 2 amongst the virus tested for antiviral susceptibility or screened for the h275y mutation by or specific rt-pcr, only 11 oseltamivir-resistant viruses related to the na h275y mutation have been detected. one of these cases also had an i223r mutation associated to a reduced sensitivity to zanamivir. all but one resistant virus were detected in treated immunocompromised patients. overall, eight patients presented a virus with the d222g mutation in the ha. all these patients had a severe infection; one of these had also a h275y mutation in the na asociated to oseltamivir resistance. the pandemic started by the end of april 2009. although the first cases recorded were as early as the 28th of april, the epidemic wave associated with a widespread spread of the virus was only recorded in october. the french population did not have to face a summer wave, as observed in north america and in numerous european countries. 3, 4 it is difficult to speculate the reasons for the lack of summer wave; the specimens collected were negative for influenza. moreover, during september, it was anticipated that school openings would be the trigger for the beginning of the pandemic wave. as a matter of fact, a significant increase of influenza-like syndromes were observed at that time, but the virological investigation carried out by the laboratories showed thta is was related to a very large epidemic of rhinovirus. 5 the epidemic circulation of other respiratory viruses can be counfounding factors for the surveillance of the influenza epidemic clinical when the survellance is only based on collection of clinical information. the starting of the pandemic wave was heterogeneous in france. the ilede-france region (paris and its suburbean area), where the population is dense, experienced an early start as compared to the rest of france. however, once the pandemic started in the rest of the county, the epidemic curves were quite similar. the peak was reached at identical times, although it may have been delayed in some remote places in france. overall, we estimate that 10% of the french population consulted for an ili presentation. the impact was mainly observed in the 5-15 years groupe of. however, this age groupe represented only a limited number of severe cases and deaths. on the other hand, the 15-65 years groupe of age, where the prevalence was not high, was the age group where the majority of severe cases and deaths was recorded (74% and 66%, respectively). 6 this data is consistent with the observational data reported by numerous other countries. 7 according to the profile of hospitalized cases, a(h1n1)2009 was more aggressive than seasonal viruses. the number of admission to the hospital was ten-fold that observed during a normal influenza epidemic. even if the mortality was limited (312 cases), the age distribution of the deceased patients was different as compared to seasonal influenza (75% mortality in <65 years of age). the lack of recordeable excess mortality has been interpreted to be the consequence of a very mild pandemic, milder than some seasonal epidemics. however, the median age of the fatal cases was much younger than those observed during the seasonal flu, leading to a mis-interpretation of the real impact of the pandemic. when the impact is measurered in loss of years of life, the impact of this pandemic is larger than seen with seasonal influenza, and is quite comparable to these of the two last pandemics. 8 the pandemic preparadness of numerous countries, the develoment of new intensive care techniques and equipment, and the large use of antivirals have reduced the overall impact of this pandemic. these are new factors that should be taken into account when evaluating the real impact of the 2009 h1n1 virus. 8 the virological monitoring of the pandemic was achieved by the community-based and hospital-based seasonal influenza networks, reminding the importance of maintining such networks. the diagnosis of influenza in most of the patients was carried out by molecular techniques. it has been clearly stated from the beginning of the pandemic that near-patient tests were lacking of susceptibility and could not be used for patient management. the distribution of a set of validated and comprehensive techniques by the two nic was very helpfull for the monitoring of the pandemic and the patients. however, this diagnostic procedure change should not preclude maintaining virus isolation that is necessary for whole genome analysis, monitoring of antigenic changes, and phenotypic testing for antiviral testing. some of the mutants that have been recorded, including viruses with antiviral resistance phenotype or genotype, could be analysed from grown virus strains. it is striking that despite a large antiviral usage, only a limited number of isolates had mutations associated to resistance. however, the frequent isolation of such resistant virus was observed in immunocompromised patients that presented severe infections and long virus shedding. 9 the impact of the pandemic is still under evaluation. sero-epidemiological analysis will be performed to asses for the real attack rate of the 2009 pandemic virus. as in other countries, it has been recorded that asymptomatic infections could be observed frequently. 10 it is quite unlikely that the impact of the pandemic was reduced by the vaccination campaign, although this vaccination started on the 12th of november, just when the pandemic started in france. it is estimated that 6 millions received the vaccination. pandemic strains of the influenza virus sporadically emerge, deviating from the regular endemic strains of seasonal influenza. in april 2009, a novel pandemic influenza virus a ⁄ h1n1 emerged, swiftly spreading across the world. immediately, domestic and international public health agencies were forced to develop containment and mitiga-tion strategies in response to the pandemic. however, the dynamics and transmission patterns of this novel virus are yet to be fully understood. simultaneously, seasonal strains of influenza (a ⁄ h1n1, a ⁄ h3n2, and b) continued to circulate in many nations. both pandemic and seasonal variants of influenza are responsible for significant morbidity and mortality. 1 to characterize the dynamics of this disease and the variation within strains, a more detailed understanding of the patterns in viral shedding during natural infection is required. the majority of data on the patterns of viral shedding during influenza infection are a result of volunteer challenge studies. 2 in these studies, volunteers are commonly screened for pre-existing immunity against the challenge strain and are of a certain demographic and age. information on the patterns of viral shedding in natural influenza infections, pandemic or seasonal, is limited but should provide greater generalizability. we describe the trends of viral shedding and clinical illness in community acquired cases of pandemic and seasonal strains of influenza. in 2008, a community-based study was conducted to analyse the effectiveness of non-pharmaceutical interventions to prevent the spread of influenza in households. 3 in 2009, a similar community-based study was initiated to collect comparative data from individuals infected with seasonal and pandemic influenza. 4 both studies were conducted with very similar protocols, involving 617 households in total. the specimens and symptom data required for this study all arise from secondary infections ascertained in these two community-based studies. the recruitment process in both studies was essentially identical. index cases were first recruited from their healthcare provider if they presented with influenza-like illness (ili). this individual would be included in the follow-up if he ⁄ she tested positive for influenza virus infection by rapid antigen test (quickvue) and was the first person in his ⁄ her household that showed signs of ili in the previous 2 weeks. follow-up consisted of three home visits that spanned approximately 7-10 days. at each home visit, nasal and throat swab (nts) specimens were collected from all household members, regardless of the presence or absence of symptoms. symptoms were recorded in daily symptom diaries provided for every household member, and digital thermometers were provided to record daily tympanic temperature. the symptoms recorded were fever ‡37ae8°c, headache, myalgia, cough, sore throat, runny nose, and phlegm. influenza virus infection and subtype was identified by reverse transcription polymerase chain reaction (rt-pcr) on the nts specimens. viral shedding was quantified from the same specimens by rt-pcr to determine viral loads, as well as by quantitative viral dilutions to determine median tissue culture infectious dose (tcid 50 ). the details concerning laboratory methods have been described in a previous study. 5 all analyses in this study focus exclusively on secondary cases; these are household contacts of recruited index cases who acquire influenza virus infection following the initial home visit. index cases generally presented with a certain threshold of illness severity requiring medical attention, whereas infections among household contacts can vary from asymptomatic to severe representing naturally acquired influenza infections. these secondary cases must be negative for influenza for their first nts specimen, and subsequently tested positive. we analysed mean viral loads measured by rt-pcr and quantitative culture by plotting by day since acute respiratory illness (ari) onset according to strain of influenza (pandemic a ⁄ h1n1, seasonal a ⁄ h1n1, seasonal a ⁄ h3n2, and seasonal b). ari is the reference time point, because the day of infection is unknown and is defined as the presence of ‡2 of the symptoms mentioned above. average symptom scores were also plotted according to ari onset and grouped into upper respiratory symptoms (sore throat and runny nose), lower respiratory symptoms (cough and phlegm), and systemic signs and symptoms (fever ‡ 37ae8°c, headache, and myalgia). mean daily tympanic temperatures were also plotted since date of ari onset and according to strain of influenza virus. all analyses were conducted using r software (version 2.10.1; r development core team). 6 a total of 617 households and 2499 individuals were followed-up in the two studies. of 1887 household con-tacts tested by rt-pcr, 153 were found to be influenza positive. among these influenza infections, 13 (8ae5%) were asymptomatic (rt-pcr positive plus 0 symptoms recorded), 20 were subclinical (rt-pcr positive plus 1 symptom recorded), and 88 presented with an onset of ari during the follow-up period. from the cases with ari onset, seven pandemic a ⁄ h1n1, 40 seasonal a ⁄ h1n1, 22 seasonal a ⁄ h3n2, and 19 seasonal b influenza virus infections were identified. the age distribution among secondary cases was observed to be largely comparable across the four strains of interest (table 1 ). there were a lower proportion of males who acquired pandemic a ⁄ h1n1 compared to the seasonal strains of the virus. cough was the most commonly reported symptoms during follow-up in cases of pandemic a ⁄ h1n1 and seasonal b, whereas runny nose was most common in seasonal a ⁄ h1n1 and a ⁄ h3n2 cases. cumulatively, fever ( ‡37ae8°c) was reported in approximately half (51%) of the secondary cases. patterns of viral shedding were analysed in a subset of 88 influenza positive individuals who recorded an onset of ari in their symptoms diaries (figure 1 ). household contacts that were asymptomatic, subclinical, or did not have an ari onset were excluded from the analysis. viral shedding in all three influenza a strains were recorded to occur on the day of ari onset or 1 day post-ari onset. following the peak, measured levels of viral shedding declined steadily to undetectable levels over 5-6 days. the trend of viral shedding in influenza b infected individuals rose 2 days before ari onset, fluctuated for around 6 days before eventually resolving. the patterns of viral shedding over time measured by quantitative viral culture were generally similar to the patterns measured by rt-pcr. the patterns of symptoms and signs were comparable in the four strains of influenza included in this study, peaking on the day or 1 day post-ari onset, and gradually declining over a period of 5-7 days. in all strains, systemic symptoms and signs were observed to resolved faster than upper and lower respiratory symptoms. the trend of tympanic temperature in each influenza strain was comparable to the respective symptom pattern. patterns of viral shedding observed in influenza a strain infections (pandemic a ⁄ h1n1, seasonal a ⁄ h1n1, and seasonal a ⁄ h3n2) were broadly similar. the pattern differed from the observed pattern of viral shedding in seasonal influenza b infections. the majority of viral shedding in influenza a strains occurred at and near ari onset, whereas there were variable amounts of viral shedding preand post-ari onset for those with influenza b. the biological reason for this difference is yet to be clarified. these differences are consistently observed regardless of laboratory method used to quantify the viral loads. it was observed that viral shedding measured by tcid 50 resolved more quickly than when measured by rt-pcr, suggesting that rt-pcr is more sensitive, but it could be detecting inactivated fragments of rna instead of active virus. the trends observed for the seasonal strains of influenza in this study were similar to those reported in literature. 2 the patterns of symptoms and signs as well as tympanic temperature in the four different strains of interest in this study were found to be comparable. these patterns closely resemble the patterns of viral shedding observed in the influenza a virus strains, but not in the influenza b virus strain. the trends of viral shedding, symptom scores, and tympanic temperature for pandemic a ⁄ h1n1 were similar to trends observed for seasonal a ⁄ h1n1 and seasonal a ⁄ h3n2 infections, suggesting that the dynamics of these viruses are largely the same. the clinical course of infection with pandemic a ⁄ h1n1 influenza virus appeared to be similar to the seasonal b influenza virus, but the patterns of viral shedding over time diverges. in general, our results suggest that the dynamics of the pandemic a ⁄ h1n1 virus were similar to the seasonal a ⁄ h1n1 and a ⁄ h3n2 viruses, and clinically similar to the seasonal b virus. this study faced sample size limitations; very few cases of pandemic a ⁄ h1n1 were detected and the secondary attack rate in general was low, though a total of 617 households were followed up. this lack of power led to the inability to analyse the differences between adult and children and other characteristics that could be correlated with amount of viral shedding. there are also biases that must be factored in during recruitment. the eligibility criteria of only healthy households could select for households with higher innate immunity. on the other hand, recruitment at health care providers can be biased towards index cases that had more severe illness that required medical attention. the strength of the study is the broad generalizability of the results due to the strict classification of secondary cases. the infections reported in this study were all community-based and should represent true natural infections. pandemic potency of the influenza virus is largely determined by its transmissibility. the first objective of this study was to model the transmission of influenza h1n1 and h5n1 viruses. at present, vaccination with laiv has been used as a widespread, effective public health measure for influenza prophylaxis. some unsubstantiated concerns have been raised about a potential possibility of reassortment of circulating influenza viruses with laiv viruses following vaccination with laiv. thus, another objective of this study was to assess the probability of pig-to-pig transmission of cold-adapted viruses and their potential reassortment with wt influenza strains. female albino guinea pigs weighing 300-350 g were inoculated intranasally with 10 5 eid 50 of virus without anaesthesia. transmission studies were then performed 24 hours after inoculation. inoculated animals were housed at 25% relative humidity and 22°c in the same cage with noninfected guinea pigs or in cages placed 3 m away from non-infected pigs. virus replication was determined by virus isolation in hen eggs and by pcr. sera were collected at 0 and 28 days post inoculation. seroconversions were assessed by routine hai test. genome composition of reassortants was monitored by rflp analysis. capacity of the viruses to grow at optimum, low, and elevated temperatures (ca ⁄ ts phenotype) was evaluated, and virus growth properties were observed following virus titration in hen eggs. when infected pigs were co-caged with non-infected (naïve) individuals, vn1203, indo ⁄ 5, a ⁄ california ⁄ 7 ⁄ 2009, and nibrg-23 were isolated in 0%, 25% 83ae3%, and 100% of contact animals, respectively. serological confirmation of virus transmission was higher than virological data (25%, 100%, 100%, and 100%, respectively). in addition, it was shown that when pigs inoculated with a ⁄ california ⁄ 7 ⁄ 2009 were co-caged with animals inoculated with nibrg-23, they got infected with both viruses ( table 1) . the ability of direct transmission of cold-adapted viruses was also investigated. data show that the a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 laiv candidate was detected in the upper respiratory tract of 87ae5% vaccinated pigs. the mdv was identified in 100% of infected animals. however, neither group of contact pigs, co-housed with the vaccinate pigs, had evidence of infection with cold-adapted viruses. in addition, none of the contact pigs had any evidence of seroconversion to the coldadapted viruses as determined by hai assay. it was also most interesting to note that pig-to-pig transmission of the highly transmittable nibrg-23 reassortant virus was not seen when pigs, vaccinated with mdv, were co-caged with animals infected with nibrg-23 virus (table 1) . this strongly implies a form of interference or protection from transmissibility that was provided by the cold-adapted virus. the results show that nibrg-23 and indo ⁄ 5 viruses were able to spread between cages over the 3 m distance (100% and 50% naïve animals were successfully infected, respectively). a ⁄ california ⁄ 7 ⁄ 2009 influenza and vn1203 viruses did not transmit between infected and non-infected guinea pigs housed in separated cages (table 1) . pigs with confirmed a ⁄ california ⁄ 7 ⁄ 2009 virus replication were also infected with nibrg-23 virus if h1n1-and h5n1-infected animals were separated by a space. thus, influenza virus transmission from h5n1-to h1n1-infected pigs has been shown, but the reverse pattern did not occur. transmission of nibrg-23 or a ⁄ california ⁄ 7 ⁄ 2009 viruses was not observed when contact pigs were first vaccinated with the mdv and housed at a 3 m distance ( table 2) . it was also shown that efficiency of transmission of nibrg-23 was much higher than of other studied h5n1 viruses; it can be transmitted between naïve guinea pigs separated from infected animals at a distance of 4-5 m (data not shown). five reassortants were isolated from animals which were infected with a ⁄ california ⁄ 7 ⁄ 2009 virus and co-caged with pigs inoculated with nibrg-23. two reassortants possessed different combinations of pr8, nibrg-23, and a ⁄ california ⁄ 7 ⁄ 2009 genes and demonstrated the non-ca ⁄ non-ts phenotype typical of wt viruses. unexpectedly, two other reassortants inherited ha gene from nibrg-23, na gene from a ⁄ california ⁄ 7 ⁄ 2009, and other genes from pr8 became ca and ts. 7:1 non-ts reassortant inherited pa gene from pr8 and seven other genes from a ⁄ california ⁄ 7 ⁄ 2009, gained ca properties. in spite of aforesaid experimental data, we cannot exclude the theoretical possibility of simultaneous infection of human host with cold-adapted and wt influenza viruses. to better understand possible consequences of such a reassortment event, we co-infected guinea pigs with a mixture of mdv and nibrg-23 viruses. nasal washes were collected and cloned by limited dilutions in hen eggs in the presence or absence of immune serum to the mdv. cloning of nasal washes without antiserum led to isolation of over 100 clones, which were all identical to the mdv (data not shown). when nasal washes were cloned in the presence of antiserum, only nine clones were isolated. genome composition analysis showed that all isolates were triple reassortants, which had inherited pb2 and na genes from mdv, pa gene from pr8, and ha gene from nibrg-23. the origin of the other gene segments (pb1, np, m, ns) in the genome of guinea pig-derived reassortants varied. reassuringly, all reassortants generated in vivo had the phenotype typical of the mdv. the severity of influenza outbreaks is partly determined by efficient spreading of the causative virus strain between human hosts. however, little is known about mechanisms underlying influenza virus transmission in humans. guinea pigs have been shown to be a suitable model for influenza transmission studies. 1 our in vivo study showed that influenza a viruses vary in their transmissibility. nib-rg-23 and indo ⁄ 5 viruses were able to transmit to naïve animals caged distantly from infected animals. in contrast, cold-adapted viruses, the same as those used for licensed laivs, showed no signs of transmission from one guinea pig to another. our study also provided evidence of a lower level of transmissibility of the novel pandemic h1n1 virus compared to the nibrg-23 and indo ⁄ 5 h5n1 strains evaluated. benefits of vaccination with laiv to aid in the control of influenza outbreaks are acknowledged by the who. 2 in our study, the mdv inoculated into guinea pigs appeared to interfere with and even offer protection from transmission of the highly transmissible nibrg-23 virus. the ability to immunize with the laiv and subsequently block the spread of a homologous h3n2 subtype and a heterologous h3n2 subtype influenza virus between guinea pigs has been shown. 3 interference between cold-adapted and wildtype influenza virus infection was the most likely explanation for the data observed in our study. the mdv inoculated into guinea pigs might in some way interfere with transmission of highly transmissible influenza viruses. it is believed by some that widespread use of laiv could increase the potential risk of reassortment of the vaccine strain with circulating influenza viruses immediately following vaccination. however, it was shown that any such potential reassortments would most likely lead to yet attenuated viruses. 4 our in vivo studies have shown that introduction of mdv genes into the genome of nib-rg-23 virus led to the generation of triple reassortants inherited pb2 and na genes of mdv and ha gene of h5n1 virus. all isolates possessed phenotypical markers associated with attenuation of mdv. our data suggest that even if a reassortment event of such rare occurrence between a laiv strain and a circulating virus were to occur, it would most likely lead to a reassortant that would retain highly attenuated phenotypic properties of the vaccine strain. our data strongly support the safety of laivs, especially those developed against highly transmissible h5n1 and h1n1 pandemic influenza viruses. this information builds upon databases that have clearly shown the low likelihood of transmitting an laiv, as well as the high likelihood of any field reassortment of laiv with a circulating influenza virus to retain important properties of the cold-adapted, temperature-sensitive vaccine master composition. very interestingly, we also present data that show the potential of a laiv to prevent the transmission of highly infectious influenza viruses, perhaps identifying a broader role for laiv in the overall scheme of influenza virus prophylactic use. background: schlieren imaging is a non-invasive, real-time airflow visualization technique that relies on differences in air temperatures (and the resulting changes in the refractive index) to allow exhaled human airflows to be seen clearly against the background of more-stationary, ambient air. recently, this technique, well-known to engineers, has been applied to better understand and characterize airflow behaviors associated with everyday, as well as healthcarerelated, human respiratory activities. materials and methods: as a surrogate marker for the behavior of airborne infectious agents, schlieren imaging was used to visualize the airflow patterns produced by adult human volunteers of different ages while coughing with and without the wearing of standard surgical and n95 masks. results: the cough plumes were generally similar in shape and range for all the adult volunteers used in this study. although both the surgical and n95 masks decelerated and blocked some of the forward momentum of the coughed airflows, much of the cough plume was redirected and escaped around the top, bottom, and side edges of the masks to merge with the volunteer's natural, verticallymoving thermal plume. conclusions: schlieren imaging is a safe technique for visualizing exhaled airflows from human volunteers without the need for potentially-irritant or toxic particle tracers. findings from these schlieren imaging experiments will assist the development of more effective aerosol infection control guidelines in healthcare premises where patients infected with potentially airborne infectious agents (e.g., influenza and tuberculosis) are present. these infectious agents may be transmitted to healthcare workers, other patients, and their visitors by way of exhaled airflows. with the recent influenza pandemic 1,2 and the ongoing concerns about human cases of avian influenza h5n1 infections, 3 there is now a very real concern about the potential for the aerosol transmission of respiratory pathogens. such concerns amongst staff and patients in healthcare environments have led to a greater emphasis on the understanding and control of infectious airflows. 4,5 previous visualization techniques have used potentially-toxic or irritant gas or particulate tracers with hazardous laser light sources that have precluded the use of human volunteers as subjects. instead, various forms of lung models that simulate human respiratory patterns with such particulate tracers have been used. 6, 7 schlieren imaging is a technique familiar to engineers and offers a non-invasive (i.e., no tracer required) airflow visualization method that depends only on differences in the refractive index of the warmer, human-exhaled air and the cooler ambient air. 8 the use of a simple incandescent or light-emitting diode (i.e., non-laser) light source is safe and allows human volunteers to be used as experimental subjects, where their exhaled airflows are then observed using a large, precise spherical or parabolic telescopic mirror and a camera, and are recorded for later analysis and presentation. [9] [10] [11] the analysis of these patterns of 'real-life' human airflows will be useful in optimizing aerosol infection control guidelines, which aim to reduce the transmission of airborne infectious agents to other healthcare personnel, patients, or their visitors. the images and analysis presented here have all been obtained from the large 1 m diameter parabolic mirror (figure 1 ) situated at the gas dynamics laboratory of penn state (directed by gary s. settles). this large schlie-ren imaging system has been in use for over 30 years to obtain high quality schlieren images for various engineering applications. it has only recently been applied to clinically-relevant imaging. the objective of this paper is to augment and expand upon the details of the methods and results presented in an earlier study using this same schlieren imaging system. 10 the aim of this series of studies is to visualize and capture a series of airflow images produced by coughing from adult human volunteers of different ages (24-80 years old). these included males (three of 26 years, one of 80 years of age) and females (one of 24 years, one of 30-40 years, and one of 40-50 years of age). each volunteer was tested with and without wearing either a standard surgical mask or n95 mask. more specifically, the aim was to visualize the extent and direction of leakage around the mask whilst each subject was coughing. penn state institutional approval for experiments involving human subjects was also obtained. each volunteer was asked to stand approximately 1 m in front of the schlieren mirror, facing across the surface of the mirror on one side, and to cough several times as the real-time, color image and video footage was recorded by the operator (using a nikon d90 camera; nikon inc. melville, ny, usa). this process was repeated whilst each volunteer was wearing a standard surgical mask then an n95 mask (supplied by 3mô, st paul, mn, usa). some of the schlieren images obtained from some of these volunteers have been published previously: for a 26-year old male, 9 the 24 year-old female and a 26-year old male, 10 and the 40-50 year-old female. 11 this article completes this series of schlieren images obtained from these experiments by including the images recorded for the older, 80 year-old man. generally, it was found that the shape of the cough plumes (shown in the figure as darker shadows emanating from the subject's mouth) produced by adult humans of different ages was relatively similar. cough plumes are roughly conical in shape and very turbulent, usually passing beyond the extent of the 1 m mirror (figure 2a) . a previous detailed study of one of these images measured a maximum airflow velocity of 8 m ⁄ second for an adult cough. 9 similarly, the effects of wearing surgical and n95 masks can be generalized across different ages. wearing a surgical mask allows leakage of the coughed air from the sides, top, and bottom of the mask ( figure 2b ). there is also some leakage through the mask, as indicated by the darker patches of air directly in front of the mask ( figure 2b, c) . the useful effect of the mask appears to be a deceleration and redirection of this coughed (and potentially infectious) air into the natural, upward-rising human thermal plume, which captures it and carries it upwards where it is diluted and less likely to transmit infection to others. the effects of the n95 mask are similar (i.e., deceleration and redirection), yet due to its tighter (mask-fitted) face seal, more of the coughed air appears to penetrate the front of the mask ( figure 2c ). this penetrating air is, however, also decelerated sufficiently to allow the wearer's natural thermal plume to carry it upwards. 10, 11 discussion from these series of schlieren images presented in this and other related studies, [9] [10] [11] it is clear that schlieren imaging offers a safe, non-invasive, real-time technique to visualize human exhaled airflows for all age groups. it is apparent that, at least where airflow patterns are an acceptable surrogate marker for airborne transmission risks, there are beneficial effects of wearing either type of mask, even when the mask fit is relatively poor. this is often the case when n95-style masks are purchased and used by the general public -in contrast to the situation with healthcare workers, who are often accurately fit-tested for this type of mask. the immediate significance of this can be seen when masks are bought by parents for their children. often, these will not be of pediatric size and the mask-fit will be loose. children are well-known to be major sources of infection in the community because of their relatively poor immunity to many types of infectious agents due to their young age and, therefore, limited past-exposure history. 12 these images allow infection control teams to literally see how far and how fast potentially-infectious human exhaled airflows can travel from an individual. this may have significant implications for guidance on the wearing of masks for infected staff and patients, on ward bed-spacing, as well as for the types of masks to be used in different situations. the important practical potential lies in the non-intrusive visualization of airflows associated with human volunteers, to assist in heightening the awareness amongst healthcare workers of the risks and potential for the airborne transmission of infectious agents, as well as the development of more effective aerosol infection control policies. schlieren images can be analysed more quantitatively, e.g., with the 'schlieren-piv' technique, 9, 13 though this additional quantitative data is probably more of research interest than being of immediate practical use to everyday hospital infection control teams. these are the subtypes that we have studied. clearly, the question arises as to whether the changes in antigenicity are coupled with changes in germicide susceptibility. we have employed a modified log-reduction method 3 in a cell culture system employing mdck cells 4 in serum-free ex-cellô 5 medium supplemented with trypsin. microscopic examination of cpe was the marker for infectivity together with plaque assay. we confirmed antiviral potency by using specific subtype influenza identification subtype technology, quidel quickvue ò influenza a + b test. the log inactivation and percent inactivation by bac after a 60 second contact time for the h1, h2, and h3 pandemic strains are as follows: a ⁄ swine ⁄ iowa ⁄ 12 ⁄ 30 h1n1, 3ae5 log ⁄ 99ae97%; a ⁄ swine ⁄ cal ⁄ 2009 h1n1, 4ae8 logs ⁄ 99ae998%; a ⁄ j305 ⁄ 57 ⁄ h2n2, 5 logs ⁄ 99ae999%; and a ⁄ hong kong 8 ⁄ 68 h3n2, 5ae0 logs ⁄ 99ae999% (table 1 ). comparable results of antiviral efficacy are obtained with the tcid 50 and plaque assays against all subtypes studied. when performing the plaque assay the sensitivity of virus recovery was better in the vessel with a larger surface area and overall recovery was in agreement with the potency determined by tcid 50 assay. in our plaque assay, we inoculated a ⁄ hong kong ⁄ 8 ⁄ 68 virus dilutions into two different vessels with 2 hours adsorption time: 6-well plate and t-25 flask, 9 ml inoculum per replicate. virus titers obtained were: 1ae4 · 10 6 pfu ⁄ ml from 6-well plate and 2ae2 · 10 6 pfu ⁄ ml from t-25 flask ( table 2 ). the discrepancy on virus potency can possibly be explained as: the binding of virus to host cell occurs only when virus gets a chance to interact with the cell on the monolayer during adsorption time. the percentage of virus population in the inoculum that has the opportunity to bind to the cell mainly depends on the surface area where this interaction takes place. therefore, in our experiment the plaque assay in the t-25 flask gave higher virus recovery 2ae2 10 6 versus 1ae4 · 10 6 pfu ⁄ ml. the increased virus recovery can translate into better sensitivity of the test system for disinfectant and antiviral agents. the potency of the virus used in this study was determined by tcid 50 was 5 · 10 6 tcid 50 ⁄ ml. rapid diagnostic testing for influenza (quickvue ò influenza a + b test, quidel) for aj305 versus bac was studied. the presence of influenza viral nucleoprotein a determined by quickvue kit correlated 100% with the viral infection based on by cpe in viral culture. interestingly, the inactivation of viral nucleoprotein was able to be revealed with diagnostic kit in the dilutions of virus ⁄ bac reaction mixture, which possessed prominent cytotoxic effect for the host cells in viral culture system. this type of molecular testing method is useful for interpreting antiviral efficacy against a background of cytotoxicity. these experiments are intended for the sponsor to substantiate to us fda that their antiviral substances are safe and effective. the data shows that the three hemagglutinin subtypes were highly susceptible to the quaternary ammonium compound in the short term in vitro experiment. the appearance of novel subtypes in the future can be met with the assurance that disinfectant and ⁄ or antiseptic resistance will be unlikely. certainly, from the above data, although genetic reassortment of human and swine viruses may modulate influenza pathogenesis and limit existing vaccine benefit, it is not likely be a factor in control of viruses on environmental surfaces by benzalkonium-type disinfectant ⁄ cleaning agents in community or health care environments. table 2 . comparison of viral titer obtained in different vessels using quantal tcid 50 and plaque assay methods plaque assay tcid 50 assay t-25 (25 cm 2 ) 6-well plate (9 cm 2 ) tcid 50 ⁄ ml tcid ⁄ ml 2ae2 · 10 6 pfu ⁄ ml 1ae4 · 10 6 pfu ⁄ ml 5 · 10 6 2ae5 · 10 6 options for the control of influenza vii outbreak influenza in aged care facilities (acfs) is associated with an increased risk of poor health outcomes among residents, including death. in this paper we share our experience of managing an outbreak of viral respiratory infection in an acf very early in the 2009 influenza pandemic and also describe some of the emerging issues relating to crossreacting antibodies to the pandemic (h1n1) 2009 influenza virus in the very elderly. the outbreak investigation was conducted as part of an urgent public health intervention initiated by the new south wales (nsw) department of health during the early stages of the first southern hemisphere wave of the 2009 pandemic. nose and throat swabs for nucleic acid testing (nat) plus acute and convalescent serum samples (6 weeks apart) were collected from all the residents of an acf where an influenza-like illness (ili) outbreak occurred. the investigation revealed dual outbreaks of pandemic (h1n1) 2009 influenza and rhinovirus infection. out of 28 residents, three had laboratory confirmed influenza [two with pandemic (h1n1) 2009], and 10 had rhinovirus infection on nat. testing of acute sera collected from every subject found elevated ( ‡1:40) pandemic (h1n1) 2009 hai antibody in 60% (9 ⁄ 15) subjects aged 85 years or more (born before 1925 and median age 88 years; geometric mean titre-gmt 48ae1) compared with none of the 13 residents aged under 85 years (born after 1924 and median age 79 years; gmt 10ae1, p = 0ae01). the acf was closed to visi-tors for 7 days. the symptomatic residents received treatment-dose oseltamivir, and all other residents were given oseltamivir prophylaxis. more than one virus may be circulating in an acf with an ili outbreak at any one time in winter. a significant proportion of elderly residents had pre-existing cross reacting antibody to the pandemic (h1n1) 2009, which may explain the minimal clinical impact of pandemic (h1n1) 2009 in this elderly population. influenza is one of the leading causes of infectious death in elderly people, principally due to co-morbidities and declining immune competence with age. it is the most important agent in outbreaks of respiratory illness. 1 influenza in aged care facilities (acfs) is associated with an increased risk of poor health outcomes among residents, including death. 2 the clinical presentation of influenza in residents of acfs can be subtle, with a blunted febrile response and a non-specific decline in mental and functional status. 3 residents commonly have underlying diseases that can be exacerbated by influenza infection, and in addition, they are at higher risk of serious influenza-related complications than community dwelling elderly people. 4 people aged over 65 years are also at higher risk of influenza-related death, and more than 90% of annual influenza-related mortality is usually confined to this high risk group. 5 in australia, influenza and pneumonia have sub-stantial health impacts; recorded as being the underlying causes of death for 2623 persons in 2007. 6 since the world health organization declared an influenza pandemic in june 2009, australia has suffered one of the highest rates of confirmed infection during the first southern hemisphere wave. by late october 2009 there were 187 reported deaths due to pandemic influenza in australia, 7 and to date there have been about 18 449 deaths reported worldwide. 8 although disproportionately far fewer elderly people developed clinical influenza during the current pandemic than occurs with seasonal influenza, their case-fatality rate remained substantial. 9 early in the pandemic (june 2009), we investigated a suspected pandemic influenza outbreak in a rural acf in the state of nsw, australia. the epidemiology (including virulence and clinical outcome in the elderly) of the pandemic (h1n1) 2009 virus was mostly unknown at the time of investigation, and as time passed, this investigation provided clarity on some important issues of the influenza epidemiology in the elderly population. in this paper we share our experience of managing a dual outbreak of viral respiratory infections early in the pandemic, and also describe some of the emerging issues relating to the cross-reacting antibodies to pandemic influenza in the very elderly. the outbreak investigation was conducted as part of urgent public health intervention initiated by the nsw department of heath in conjunction with the local public health unit, the national centre for immunisation research and surveillance (ncirs), and the institute of clinical pathology and medical research (a who national influenza centre). to determine the extent and cause of the outbreak, a public health research doctor (gk) was dispatched from sydney over a weekend to assist with outbreak investigation and control. on june 12th 2009, the greater southern public health unit surveillance officer (bd) received a report of a possible pandemic (h1n1) 2009 outbreak in a local acf. on investigation, it was discovered that 3 days earlier a 77 year old female resident had become generally unwell, but without specific symptoms of influenza like illness (ili). soon after, nine of the 27 co-residents (but no staff) had developed symptoms suggestive of influenza. one other resident had returned from a melbourne (victoria) hospital (where pandemic (h1n1) 2009 was known to be circulating) the previous week after surgery, but did not have ili symptoms. on june 10th, the 10 symptomatic residents had nasal swabs taken by the local doctor for influenza [including pandemic (h1n1) 2009] nucleic acid testing (nat). there was rising concern due to reports of widespread pandemic (h1n1) 2009 influenza in a local army camp just over the border in nearby victoria, where pandemic (h1n1) 2009 influenza was known to be circulating widely. on june 12th, the 77 year old lady proved nat positive for pandemic (h1n1) 2009, but none of the other samples were pandemic (h1n1) 2009 nat positive. concern arose that there might be an outbreak of pandemic (h1n1) 2009 in the facility, and that some of the swabs from other residents might be false negatives. between 14 and 15 june, after consent was obtained, directly or through next of kin in 13 demented residents, all 28 submitted to venipuncture for serology, 27 successfully, and the other 18 as yet un-swabbed residents were swabbed. basic demographic data were collected from every resident with clinical information on co-morbidities and current medication use. convalescent blood samples were collected after 4 weeks on 16th july 2009 from 23 of the 28 residents. swabs were sent to icpmr where nat for influenza a [including pandemic (h1n1) 2009] and b was performed. the acute and convalescent serum samples were tested later (in december 2009), using haemagglutination inhibition assay (hai) to detect pandemic (h1n1) 2009 antibody. 10, 11 interventions the acf was closed to visitors from 12th until 18th june. treatment of the positive case and the nine symptomatic residents, with twice daily oseltamivir, was begun on saturday june 13th, and all other residents were started on once daily oseltamivir prophylaxis. the facility manager and local general practitioner (gp) monitored patient health on a daily basis, and none had to stop oseltamivir due to adverse events. one resident with ili who was known to have moderately impaired renal function was given once daily rather than twice daily oseltamivir treatment. the age range of the residents was 58-97 years with a median of 85 years. all residents had underlying medical conditions, e.g., chronic cardiac and respiratory diseases ( table 1) testing of acute sera collected from every subject found elevated ( ‡1:40) cross-reacting hai antibody to the pandemic (h1n1) 2009 in 60% (9 ⁄ 15) of subjects aged 85 years or more (born before 1925 and median age 88 years; geometric mean titre-gmt 48ae1). however, the hai titre was consistently <1:40 and significantly lower (gmt 10ae1, p = 0ae01) in the 13 residents aged under 85 years (range 58-83 years, median 79 years) (figure 1 ). the index case (nat positive) did not show a significant raise in hai level in convalescence (going from 20 to 40). the pandemic (h1n1) 2009 case that was determined by serology was pandemic (h1n1) 2009 nat negative. to our surprise, seven of the other 18 asymptomatic residents had rhinovirus detected on extended nat (reported on june 25th), despite being asymptomatic at time of swabbing and remaining so. the original nine influenza nat negative samples were then tested and three of these were also nat positive for rhinovirus; in total, ten proved nat positive for rhinovirus (35ae7%). the serologically confirmed pandemic (h1n1) 2009 case was also positive for rhinovirus infection. of interest was that only one resident had a documented fever. this investigation illustrates some of the difficulties in managing and investigating possible influenza outbreaks in real time in the context of an influenza pandemic. 12 finding a nat positive case of pandemic (h1n1) 2009 influenza among many other symptomatic cases raised the possibility (although not the probability) that pandemic (h1n1) 2009 was the cause of the outbreak. rhinovirus infection, however, was confirmed by nat in ten residents. this outbreak illustrates that more than one virus (in this case 2 and perhaps 3) may be circulating in an acf at any one time in winter. in ili outbreaks in acfs, broad laboratory testing is recommended; nat is the most sensitive method of detecting influenza or other viruses in respiratory tract samples. 13 studies have found that the pandemic (h1n1) 2009 haemagglutinin (ha) gene is more closely related phylogenetically to the 1918 h1n1 virus and classical swine influenza a ⁄ h1n1 viruses than more recent seasonal human influenza a ⁄ h1n1 viruses. 14 it is antigenically similar to the 1918 h1n1 pandemic virus in terms of the immunodominant antibody response to haemagglutinin. [15] [16] [17] it is likely that individuals alive during the emergence and initial persistence of the 1918 pandemic virus would have higher levels of cross-reacting hai antibodies to the pandemic (h1n1) 2009, which would contribute towards better clinical protection. 18 in our investigation, 60% of the residents born before 1925 (aged 85 years or above in 2009) had pre-existing cross-reacting hai antibody to the pandemic (h1n1) 2009. in elderly populations, severe illness may be associated with organisms typically considered to be mild, such as rhinovirus. however, studies have shown that nursing home residents may be susceptible to outbreaks of rhinovirus that may cause mild to severe respiratory illness, particularly in those with a history of lung disease. one rhinovirus outbreak in a nursing home in the usa caused 12 fatalities. 19 another outbreak showed residents with underlying lung disease are more likely to have longer infection, require antibiotics, develop bronchospasm, and have difficulty breathing; two residents with underlying lung disease required emergency treatment and one died. 20 a previous influenza outbreak in a nsw aged care facility in 2006 caused significant mortality and morbidity. that outbreak resulted in 14 hospital admissions and six deaths. 21 in our investigation we have found that 17% of the residents had chronic lung disease and 64% had chronic cardiac conditions both considered as high risk for severe complications of both rhinovirus and influenza infection. however, there were no hospitalisations or deaths in our outbreak investigation. indeed only one resident developed fever, indicating that non-specific signs of illness (such as in our index case) may be the only, or early, indication of an ili. our own experience with managing other ili outbreaks has also taught us that staff of acfs may not be vigilant enough to detect fevers. in this outbreak, the nursing home staff, local gp, public health unit and the outbreak investigation team and supporting laboratory staff acted quickly and in a coordinated way. pre-existing cross-reacting antibody in the very elderly (aged ‡85 years) probably helped to limit the spread of the pandemic virus (compared to the circulation of rhinovirus) within the acf. exposure to the 1918 pandemic (or a close variant occurring before 1925) appears to be responsible for a high hai titre in the very elderly, which contributed towards better clinical protection. however, wider testing early on would have alerted us more quickly to the main cause of the outbreak. treatment and prophylactic use of oseltamivir may also have contributed to halting the spread of pandemic (h1n1) 2009 and also to symptom relief. pandemic (h1n1) 2009 influenza virus (ah1pdm) has spread worldwide since march 2009. in a paper of ah1pdm, 25% of infected individuals have experienced gastrointestinal symptoms such as diarrhea and vomiting, which is higher than that of seasonal influenza. however, little is known whether viable virus shed from stool and replication of viruses are ongoing in the gastrointestinal tract. 1,2 viral load and isolation of ah1pdm in cell culture in stool samples has been reported. 3 stool specimens were collected from 35 patients suspected to have pandemic (h1n1) 2009 infection from november 2009 through may 2010. virus isolation was conducted in cell culture by using madin-darby canine kidney (mdck) cells and taqman based rt-pcr from 10% (w ⁄ v) stool suspension in phosphate-buffered saline. taqman based rt-pcr was conducted by using primers, probes, and positive controls provided by niid (national institute of infectious diseases of japan). to confirm presence of ah1pdm viral rna, lamp (loop-mediated isothermal amplification) was used as supplemental testing. of patients, one child (case 1) submitted one nasal swab and four stool samples, another one nasal swab and two stool samples, and the other one stool sample. informed consent was obtained. strand specific rt-nested pcr was performed for only case 1 by using only one primer at the rt reaction and also assayed neu5aca2-3gal and neu5aca2-6gal binding specificity about isolated strain derived from nasal swab and stool. receptor binding specificity was performed using a solid-phase binding assay with the sialylglycopolymers (poly a-l-glutamic acid backbones containing neu5aca2-3galb1-4glcnacb-pap or neu5aca2-6galb1-4glcnacb-pap bond as described. 4 ) nucleotide sequences of the ha gene of ah1pdm viruses isolated from stool sample and nasal swab were analysed. in order to exclude the possibility of contamination, the stool samples and nasal swabs were subjected to virus isolation separately. after getting the results on the nucleotide sequence, we also confirmed no strain harboring identical sequence was isolated in our laboratory before and after the day of sample collection. ah1pdm viral rna was detected in nine (25%) of the subjects from stool samples. among nine subjects, one case (case no. 1) was positive for viral isolation. case1, a healthy 9-year-old girl, experienced fever and abdominal pain, and the others had gastrointestinal symptoms without upper respiratory symptoms. in case 1, influenza a virus was diagnosed by rapid antigen test on the day of symptom onset. viable ah1pdm virus was isolated from the stool sample and nasal swab on the second day from onset using mdck cells (table 1 ). viral load decreased gradually after symptom onset. however, viral shedding was still present 8 days after symptom onset. positive stranded rna was detected 6 days after symptom onset from the stool specimen ( figure 1 ). above two ah1pdm strains (isolated from nasal swab and stool specimen) bound exclusively to human type receptor, neu5aca2-6gal. sequence analysis demonstrated that isolated virus from stool samples was identical with that from nasal swabs in comparison of ha gene (990 bp). ah1pdm influenza virus was isolated from the stool and nasal swab samples in the same patient simultaneously by using mdck cells. our results suggests the detection of viral rna and viable ah1pdm influenza virus from stool samples may serve as a potential mode of transmission and has important implications in understanding the context of ah1pdm influenza virus. strategies to prevent transmission of influenza include use of respirators. ffp2 and n95 respirators are certified to fil-ter at least 95% of particles (0ae3 lm in diameter), and many guidelines have recommended that healthcare workers wear respirators in certain healthcare settings to protect against infection from patients with pandemic influenza. [1] [2] [3] we have developed a proprietary acid-polymer formulation to coat a standard ffp2 respirator with an antiviral layer. we aimed to test this coated respirator for antiviral efficacy against a range of influenza viruses. a series of tests compared the antiviral efficacy of coated and uncoated respirators in conditions designed to simulate real-life exposure to influenza by varying the route of inoculation, contact time, temperature, humidity, moisture, and contaminating substances. we also investigated whether infectious viruses could be transferred from contaminated respirator surfaces to gloves. we tested human, swine, and avian influenza viruses, including influenza a and b viruses. influenza a subtypes were the a ⁄ h1n1 2009 pandemic strain, seasonal h1n1, h5n1, h3n2, h5n9, and h2n2. in each test, suspensions of influenza viruses were prepared to 4-8 log 10 tcid 50 ⁄ ml in mem. in some tests, organic contaminants (yeast, bsa, and mucin) were added. one set of respirators was maintained at 40°c and 75% relative humidity for 24 hours before the viral challenge, and repeatedly sprayed with he-pes buffer to simulate respiratory secretions. for each test, three coated (glaxosmithkline actiprotect) and three uncoated (sperian willson easy fit) ffp2 respirator samples were inoculated with 0ae2 ml of a viral suspension, which was applied with a pipette, sprayed, or aerosolised to create airborne droplets. after 1 minute at room temperature (on a shaker), the respirator samples were assayed for the presence of infectious viruses using standard methods. 4 in one test, after a 1 minute contact time of the respirator with the virus, nitrile gloves were applied with light pressure to the outer surface of inoculated respirator samples and then assayed after 1 minute. samples were put into test medium (mem, supplemented with antibiotics [penicillin, gentamycin, or streptomycin] and amphotericin b or l-glutamine). the supernatants were vortexed, extracted, and used to prepare serial 10-fold dilutions in mem. each dilution was used to inoculate four wells of rmk cells in a multi-well plate, and these cultures were incubated and scored over 7 days for cytopathic effects, cytotoxicity, and viability. (some tests substituted mdck cells; others used inoculated embryonated chick eggs.) all tests included negative cell controls, cytotoxicity controls, and neutralisation controls. the spearman-karber formula was used to calculate viral loads as tcid 50 or eid 50 . 4 antiviral efficacy was calculated from the difference between the geometric mean loads of influenza virus on the coated and uncoated respirators after 1 minute of exposure. the viral loads applied to respirators in these experiments ranged from 5ae5 to 8ae1 log 10 tcid 50 , and were therefore high in comparison with respiratory secretions from infected patients at the peak of influenza symptoms (range 3-7 log 10 tcid 50 ). 1 tables 1-2 show that the average viral loads detected on uncoated ffp2 respirator samples remained high in all conditions tested, ranging from 3ae2 to 6ae9 log 10 tcid 50 (or 4ae5-5ae0 log 10 eid 50 ). in contrast, the average viral load on coated respirators after 1 minute of exposure ranged from below the limits of detection to £1ae5 log 10 tcid 50 (1ae0 log 10 eid 50 ). therefore, the relative antiviral efficacy of the coating ranged from ‡2ae7 to 6ae4 log 10 . table 1 shows that the relative antiviral efficacy of the coated mask remained high in simulated-use conditions such as organic contaminants and repeated saturation at high temperature and humidity. in the experiment to test transfer of viruses from respirators, the gloves applied to regular uncoated inoculated respirators had a viral load of 3ae5 log 10 eid 50 (table 1) . by contrast, no viruses were detected on either the coated respirators or the gloves applied to them. the relative reduction in contamination was therefore ‡2ae5 log 10 . ‡4ae8 log 10 viral load with organic contaminants* 6ae1 0 ae7 5 ae3 log 10 viral load after heat, moisture, and simulated secretions** 5ae1 0 ae8 4 ae3 log 10 viral load transferred to glove** 3ae5 £1ae0 ‡2ae5 log 10 eid 50 *influenza subtype was a ⁄ h5n1, and strain was vnh5n1-pr8 ⁄ cdc-rg. **influenza subtype was a ⁄ h3n2, and the strain was hong kong ⁄ 8 ⁄ 68. results are mean log 10 tcid 50 , unless specified otherwise. results are mean log 10 tcid 50 , unless specified otherwise, based on an infectivity assay in triplicate. limits of detection varied. *2009 pandemic strains. **results are mean log 10 eid 50 , based on a haemagglutinin assay in duplicate. options for the control of influenza vii ª 2011 blackwell publishing ltd, influenza and other respiratory viruses, 5 (suppl. 1), 301-327 strategies to prevent transmission of influenza include use of respirators, and many guidelines have recommended that healthcare workers wear respirators in certain healthcare settings for protection against pandemic influenza. 1-3 ffp3 respirators are certified in europe to filter at least 99% of nacl particles (0ae3 lm in diameter), and ffp2 and ffp1 respirators must filter at least 95% and 80% of these particles, respectively. influenza a viruses are typically 0ae1 lm, and can be carried in aerosolised droplets smaller than 1 lm in diameter, which can disperse widely, remain airborne for 8 hours, and be inhaled deeply into the respiratory tract. 4 we have developed an acid-polymer formulation to coat the outer layer of a standard ffp2 respirator, in order to provide antiviral activity on the outer surface. we compared this coated respirator against standard ffp1, ffp2, and ffp3 respirators for filtration of aerosolised influenza viruses. the aim was to simulate protection against infectious viruses in droplets released when infected people cough and sneeze, and during aerosol-generating procedures in healthcare settings. the first assay compared three samples of coated ffp2 respirators (glaxosmithkline actiprotect) with three ffp2 controls (sperian willson easy fit). for each test, suspensions of influenza a (h3n2) at 7ae1 log 10 tcid 50 ⁄ ml in 0ae1· minimum essential medium (mem) were aerosolised with a nebulizer. the airborne droplets were introduced into a sterile chamber upstream of a respirator sample for 2 minutes, at a flow rate of 28ae3 l ⁄ minute. constant airflow was maintained for another 2 minutes after exposure to the virus. then the collection dish in the downstream sieve sampler (anderson) was assayed for infectious viruses using standard techniques. 5 briefly, serial dilutions of the collection medium (mem with 1% fbs, 5% gelatine, and 2% hepes, supplemented with antibiotics and amphotericin b) in mem + trypsin were used to inoculate madin-darby canine kidney epithelial (mdck) cells in quadruplicate in a multi-well plate. these cultures were then incubated and scored over 4-6 days for cytopathic effects, cytotoxicity, and viability. negative cell controls and cytotoxicity and neutralisation controls were also performed. the spearman-karber formula was used to calculate tcid 50 . the second assay compared five samples of coated respirators with five ffp1 controls (3m 9310) and five ffp3 controls (3m 1863). a suspension of influenza a (h1n1), at 8ae3 tcid 50 ⁄ ml, was nebulized for 1 minute and 40 seconds into the aerosol chamber, at a flow rate of 28ae3 l ⁄ minute, followed by constant airflow for 5 minutes after exposure to the virus. then the collection medium in the downstream chamber (as before, with 1% nahco 3 ) was assayed as described above. initial viral loads in the first and second assays were 8ae1 and 7ae9 log 10 tcid 50 , respectively, and were therefore high in comparison with respiratory secretions from infected patients at the peak of their influenza symptoms (range 3-7 log 10 t-cid 50 ). table 1 shows that the average viral load that passed through the uncoated ffp2 respirators in the first assay was 3ae6 log 10 tcid 50 . the average viral load that passed through the coated respirators was 1ae4 log 10 tcid 50 . therefore, for active filtration of viruses, the relative efficacy of the respirator with antiviral coating was 2ae2 log 10 greater than the uncoated respirator. for surface inactivation, the relative antiviral efficacy of the coated respirator was 3ae9 log 10. in the second study, table 2 shows that the average viral load that passed through the uncoated ffp1 respirators was 5ae2 log 10 tcid 50 . in contrast, 3ae1 log 10 tcid 50 passed through the coated ffp2 respirators. by comparison with the viral load when no respirator was present (8ae9 log 10 tcid 50 ), the ffp1 respirators reduced the viral load by 3ae7 log 10 , and the coated ffp2 by 5ae8 log 10 . therefore, for active filtration of viruses, the respirators with antiviral coating reduced the viral load by 2ae1 log 10 more than the ffp1 respirators. in this second study, the average viral load that passed through the uncoated ffp3 respirators was also 5ae2 log 10 tcid 50 . by comparison with the viral load when no respirator was present (8ae9 log 10 tcid 50 ), the ffp3 respirators reduced the viral load by 3ae7 log 10 . therefore, for active filtration of viruses, the respirators with antiviral coating reduced the viral load passing through the mask by 2ae1 log 10 more than the ffp3 respirators. table 2 also shows that the coated respirators reduced the infectious viruses remaining on the mask surfaces by 3ae8 log 10 more than the ffp1 respirators, and 3ae6 log 10 more than the ffp3 respirators. even with a very high viral challenge, the coated respirators prevented passage of at least an additional 2ae1 log 10 infectious viruses, compared with uncoated respirators. large numbers of infectious virions passed through all uncoated respirators tested. ffp3 respirators were no more effective than ffp1 respirators at blocking airborne influenza viruses. based on these in-vitro results, respirators with the antiviral coating could be expected to provide more protection than standard respirators from the risk of inhaling influenza viruses. strategies to prevent transmission of influenza include use of respiratory protection. ffp2 and n95 respirators are certified to filter at least 95% of nacl particles (0ae3 lm in diameter), and many guidelines have recommended that healthcare workers wear these respirators in certain healthcare settings to protect against infection from patients with pandemic influenza. 1,2 we have developed a proprietary acid-polymer formulation, designed to coat a standard respirator and inactivate influenza viruses on contact. we tested this coated respirator for cytotoxicity, skin irritation, and sensitisation potential. the antiviral coating was also tested for stability and leaching under extreme environmental conditions, such as physical abrasion and simulated breathing at different temperatures, levels of humidity and co 2 , and saturation with contaminants. eight coated respirators were tested at standard relative humidity (60% rh) for 8 hours, and one at elevated humidity (80% rh) for 2 hours. four coated masks were treated with synthetic blood or oral secretions, and then tested at 30% rh for 1 hour. the sample respirators were sealed onto a mannequin head inside an airtight chamber, and air at 30°c and 5000 ppm co 2 was pumped through the masks by a cyclic breathing machine at 24 l ⁄ minute. a 37 mm glass-fibre filter was placed behind the respirator, over the mannequin's mouth opening. at the end of all tests, these filters were eluted and analysed using high-performance liquid chromatography (hplc). standard in vitro methods were used to assess the cytotoxicity of the coated polyester and uncoated polypropylene layers of the respirator (glaxosmithkline actiprotect). 3 samples were extracted in minimum essential medium (mem), supplemented with serum, penicillin, streptomycin, amphotericin b, and l-glutamine, at 37°c for 24 hours. triplicate monolayers of mouse fibroblast cells (l-929) were dosed with each extract (including a reagent control and negative and positive controls), and incubated at 37°c in 5% co 2 for 48 hours. after 48 hours of incubation with samples or controls, the monolayers of mouse fibroblast cells were examined microscopically for abnormal cell morphology or cellular degeneration. samples of the coated respirator (comprising four polypropylene layers bonded to the coated polyester outer layer) were applied under occlusive patch conditions to the skin of 51 adults. controls, including individual layers, were applied in the same way. in a separate patch test, samples of the coated polyester outer layer and controls were applied under the same conditions to 219 adults. after 24 hours, test patches and controls were removed. sites were then scored for itching, erythema, oedema, epidermal damage, and papular response after 48 and 72 hours. the patches were applied three times a week for 3 weeks. to evaluate sensitisation, test patches were applied 10-15 days later for 24 hours at different sites to the original samples. after this challenge, skin was assessed and graded for sensitisation potential after 48 and 72 hours. table 1 shows that no residues of the antiviral coating or degradation products were detected in the air that had passed through any of the eight respirators. cytotoxicity tests showed that the coated respirator material caused 10% cell lysis or toxicity, classified as slight reactivity (grade 1), and that uncoated material caused no cell lysis or toxicity (grade 0) ( table 2) . results for positive and negative controls were severe reactivity and no reaction, respectively. from the results of the two human repeat-insult patch tests, neither the coated or uncoated layers nor the fullthickness respirator fabric caused irritation (including itching, erythema, edema, vesiculation, epidermal damage, papules, or reactions beyond the patch site) or sensitisation in any of the adult volunteers at any of the time points. based on these results, in conjunction with published data on acute and repeat-dose toxicity, mutagenicity, local irritation, dermal sensitisation, and inhalation safety for all components of the antiviral coating, the potential topical or inhalation exposure to the coated antiviral respirator does not pose a safety risk. the antiviral coating is durable and stable, and stays on the outer surface of the respirator, even in extreme environmental conditions. the coated respirator is non-irritating and non-sensitising. therefore, this respirator is considered to be well-tolerated and safe for its intended use. ies were funded by gsk consumer healthcare, and gsk investigators were involved in all stages of the study conduct and analysis. knowing how influenza virus is transmitted at home and in school is the key to preventing its spread. at the previous two meetings of this conference, 1,3 we introduced our study of household transmission of seasonal influenza and reported our conclusion that protracted survival of the virus even after treatment increases household transmission, and is a major factor in the transmission of the virus to infants. on the other hand, during the recent pandemic, many schoolchildren developed serious respiratory tract disorders, which again highlights the significance of schoolbased transmission of the disease. in this study, we compared transmission of a new influenza strain at home and in school with that of seasonal influenza and proposed countermeasures. the for the analysis of school-based transmission, the epidemic status of seasonal influenza in 4237 children at six elementary schools over the past two seasons (2002-2003 and 2003-2004 seasons) was compared with that of pdmh1 in 1913 children at two primary schools. using observational data of school-based transmission, we also constructed a model for influenza transmission 3, 4 and evaluated the effects of factors that could affect influenza transmission (e.g., antibody prevalence, transmission rate, non-infectious latent period, infectious latent period, school closure) through the use of 1000 simulations. in this study, a diagnosis of influenza was confirmed by rapid influenza antigen detection kit. we previously reported the high sensitivity of the kits, 5-7 not only for seasonal influenza, but also for h1n1 pandemic 2009 compared to virus isolation and pcr. serum antibody was not investigated. most of the index patients were treated with oseltamivir or zanamivir, and 30 patients were treated with amatadine. no treatment was done for 16 patients. no nai therapy was done as prophylaxis within the family. the incidence of households with an initial case patient who subsequently infected another member of the household was 21ae6% (275 of 1271 households) for seasonal influenza or 18ae3% (112 of 612 households) for pdmh1. thus, the household incidence of pdmh1 was lower than that of seasonal influenza. in addition, the percentage of family members in households who were infected by initial case patients (household transmission rate) was 10ae5% (375 of 3573 individuals) for seasonal influenza or 7ae9% (141 of 1792 individuals) for pdmh1. thus, the household transmission rate was also lower for pdmh1 than that for seasonal influenza. effect of family size on household incidence and household transmission rate an analysis of the effect of family size on household incidence showed that, in families consisting of 2-7 individuals, the incidence of seasonal influenza in order of increasing family size was 3ae5%, 17ae4%, 24ae5%, 24ae2%, 32ae4%, and 42ae9%, respectively, and the incidence of pdmh1 was 0ae0%, 13ae5%, 17ae8%, 24ae7%, 42ae9%, and 33ae3%, respectively, indicating that household incidence tends to increase with increasing family size. in contrast, no definite relationship was noted between household transmission rate and family size. transmission rates for seasonal influenza in order of increasing family size were 3ae5%, 10ae6%, 11ae8%, 8ae0%, 10ae3%, and 7ae1%, respectively, or 0ae0%, 6ae7%, 7ae7%, 8ae0%, 15ae2%, and 5ae6%, respectively, for pdmh1 (shown in table 1 ). effect of age cohort of initial case patient in household on household incidence and household transmission rate an analysis of the effect of the age cohort of the initial case patient in the household on household incidence and transmission rate showed that the household incidence of seasonal influenza in c1, c2, c3, and c4 was 26ae7% (112 of 419 households), 21ae4% (66 of 308 households), 16ae3% (21 of 129 households), 9ae9% (9 of 91 households), and for m and f was 19ae6% (37 of 189 households) and 22ae5% (29 of 129 households), respectively. therefore, household incidence was the highest in c1, followed by the parents. when the initial case patient was a child, the household incidence increased with decreasing patient age. in contrast, the household incidence of pdmh1 in c1, c2, c3, and c4 was 23ae1% (36 of 156 households), 17ae3% (42 of 243 households), 8ae1% (9 of 111 households), 10ae0% (4 of 40 households), and for m and f was 30ae8% (12 of 39 households) and 39ae1% (9 of 23 households), respectively. therefore, household incidence was higher when the initial case patient was a parent, rather than a child. the household transmission rates for seasonal influenza from c1 to f were 13ae4%, 10ae1%, 6ae0%, 3ae9%, 10ae3%, and 12ae6%, respectively. therefore, as for household incidence, the highest rate (13ae4%) was observed in c1. the corresponding household transmission rates for pdmh1 were 9ae9%, 6ae9%, 4ae0%, 4ae3%, 12ae7%, and 22ae1%, respectively, with the highest transmission rates observed for infections from parents (shown in table 1 ). if the rate of individuals with a secondary infection transmitted from the initial case patient in a household is presented as a percentage of the total number of affected individuals, the rates for seasonal influenza and pdmh1 were 22ae8% (375 of 1646 individuals) and 18ae9% (142 of 753 individuals), respectively. therefore, the rate of individuals with a secondary infection was lower for pdmh1 than that for seasonal influenza. by age cohort, the corresponding rates of individuals for seasonal influenza in c1, c2, c3, and c4 were 22ae3% (120 of 539 individuals), 11ae5% (40 of 348 individuals), 12ae2% (18 of 147 individuals), 7ae1% (7 of 98 individuals), and for m and f was 42ae7% (141 of 330 individuals) and 27ae1% (48 of 177 individuals), respectively. for pdmh1, the corresponding rates in c1, c2, c3, and c4 were 24ae3% (50 of 206 individuals), 9ae3% (25 of 268 individuals), 9ae0% (11 of 122 individuals), 4ae8% (2 of 42 individuals), and for m and f was 51ae9% (42 of 81 individuals) and 35ae3% (12 of 34 individuals), respectively. these findings indicate that, especially in the case of pdmh1, most secondary infections in parents tend to be transmitted from another household member. the mean annual prevalence of seasonal influenza and the new influenza strain at the elementary schools for the two seasons was 19ae6% and 10ae3%, respectively, whereas the prevalence determined 7 days after appearance of the first case in school was 3ae9% and 4ae5%, respectively. in the recent season at the same elementary schools, however, the prevalence was a high 43ae5%. since the prevalence at 7 days after the appearance of the first case in school was already 17ae5%, these data show that the influenza virus spread quickly throughout the schools. at the schools with high transmission rates in the early period of the pandemic, new infections were confirmed even 4 days after the school closure action was taken. these findings indicate that pdmh1, the current influenza virus, has a long latent period during which it becomes infectious and spreads from infected individuals to numerous others in their vicinity. we constructed a model for influenza transmission in schools and estimated the time course of changes in the number of expected cases and the expected prevalence during the season. in this model, school children were divided into six groups depending on the stage of infection: uninfected period with no immunity, non-infectious latent period, infectious latent period, onset, post-onset infectious period, and immune period. it was assumed that schoolbased transmission occurred during the infectious latent period prior to onset and that no infections occurred during the post-onset infectious period because children were absent from school. due to the long latent period of pdmh1, the distribution of the non-infectious latent period of pdmh1 was established as (day 1, day 2, day 3, day 4) = (10%, 40%, 40%, 10%) and the distribution of the infectious period as (day 0, day 1, day 2) = (10%, 80%, 10%). when simulations were performed under these conditions using the model for school-based transmission of influenza in which children from classes with an outbreak were kept at home for 3 days, the time course of changes in the number of affected individuals actually observed and the time course of changes in the number of expected cases were determined. the expected prevalence under these conditions was 35%. to evaluate the effect of school closure, simulations were performed based on the assumption that children from affected classes were not kept at home for 3 days. it was shown that there was an increase in the expected number of cases during the 3 days corresponding to the period of actual school closure and that the expected prevalence increased to 52%. based on these findings, it was concluded that keeping children home from classes with an outbreak is an effective means of controlling the transmission of influenza in schools (shown in figure 1 ). if the transmissibility of pdmh1 virus at home is estimated based on the speed of transmission and the degree to which pdmh1 is prevalent in schools, it would be expected that the household transmission of pdmh1 is also higher than that of seasonal influenza. in fact, the opposite is the case. this paradox can be explained in two ways. 1. the number of children aged 19 or more and parents with pdmh1 influenza as a percentage of the total number of affected individuals is lower than those with seasonal influenza (20ae8% versus 37ae2%). further, although the number of parents with a secondary infection was high at home, the percentage of the total number of individuals with pdmh1 was a low 3ae8% (29 of 754 individuals), compared to that for seasonal influenza (5ae5% [91 of 1640 individuals]). in other words, adults are less susceptible to pdmh1 infections and there was a correspondingly small number of affected individuals. therefore, it was considered that the transmission rate at home was lower than that at school for this reason. 2. the percentage of households with more than one affected individual within the same family was higher for pdmh1 at 28ae5% (157 of 550 households) than for seasonal influenza at 24ae4% (309 of 1265 households). in the patients secondarily infected with pdmh1, 26ae8% of them showed symptoms of infection 10 days or more after the onset in the first patient, suggesting that they were not infected at home, and the actual household transmission was 20ae9% (115 of 550 households). therefore, although the prevalence was higher for pdmh1, it seems that household transmission was lower because households with an affected individual implemented satisfactory control measures against infection. seasonal influenza differs greatly from pdmh1 influenza in its transmissibility at home and in school. in the household transmission of pdmh1 influenza, both the household incidence and household transmission rate of pdmh1 were low compared to those for seasonal influenza. although transmission of seasonal influenza from infants to parents was marked, in the case of pdmh1, the reverse was true with transmission from parents to children being predomi-nant. it should be noted that household transmission in mothers was common in all eight seasons, suggesting the need to reconsider control measures against infection when nursing unwell family members. in the case of school-based transmission, pdmh1 was more prevalent than seasonal influenza, indicating that the virus spread quickly throughout the schools. this difference was attributed to the long infectious latent period when pdmh1 rapidly became rampant in the schools. an analysis of school-based transmission using a model for influenza transmission showed that, when 10% of the student population is infected, schools should be closed for five consecutive days in order to minimize the spread of the disease. the effectiveness of seasonal influenza vaccine in preventing pandemic and seasonal influenza infection: a randomized controlled trial introduction household transmission has been estimated to account for one-third of all influenza transmission, 1,2 and children are at high risk of spreading the disease. with reference to previous evidence, 3-15 some vaccine deployment strategies target children to prevent them from infection and transmitting influenza. 16 nevertheless, few studies evaluated the effectiveness of vaccinating children in reducing household transmission. 10, 11 during 2008-09, a pilot randomized controlled trial was conducted to investigate such effect by studying households with school age children randomized to receive trivalent inactivated seasonal influenza vaccine (tiv). 17 the monovalent vaccine against pandemic influenza a (h1n1) (ph1n1) had yet been available until the end of the first wave. various conclusions have been made as to whether seasonal influenza vaccine might possibly protect against ph1n1. [18] [19] [20] [21] [22] [23] [24] [25] we report findings on the effectiveness of tiv against ph1n1 observed in our cohort. households were screened if they expressed interest after receiving invitation letters distributed via their children's school or an existing pediatric cohort study. 26 to be eligible, the household had to include at least one child aged 6-15 years who was not allergic or hypersensitive to any of the tiv components. children known to have immunosuppressive conditions or other contraindications against tiv were also excluded. written consent and assent were obtained from participants aged above 18 years and those aged 8-17 years, respectively. proxy written consent was obtained from legal guardians or parents for participants younger than 18 years. ethical approval was obtained from the institutional review board of the university of hong kong. consented households were allocated to the tiv and placebo group (in ratio 3:2) according a code generated by block randomization with random block sizes of 5, 10, and 15. an independent nurse prepared 0. one child (study subjects) from each household in the tiv group received a single dose of tiv with one child from each household in the placebo group receiving a single dose of saline placebo. parents and legal guardians were asked to report any adverse reactions 4 days following vaccination. all participants, study nurses, and other research staff were blinded to the allocation and administration of vaccine or placebo. the vaccine allocation sequence was only disclosed to the investigators at completion of the study. serum specimens were collected from subjects shortly before (november-december 2008), one month after vaccination (december 2008-january 2009), and after the winter (april 2009) and summer influenza seasons (august-october 2009). serum specimens were obtained from household contacts at baseline and after the winter and summer influenza seasons. all household members recorded any fever ‡37.8°c, chills, headache, sore throat, cough, presence of phlegm, coryza, or myalgia daily on a symptom diary. they were also invited to report to the study hotline immediately if they experienced at least 2 of the above signs or symptoms. as a response, the study nurse would visit the households with any sick members and collect nose and throat swab from all household members. the households were also telephoned monthly or increased to fortnightly during influenza seasons to monitor for signs and symptoms and remind them to report to the hotline. supermarket or book vouchers (for children) were given to the households including us$13 for each serum specimen collected, us$6.5 for each home visit, and us$65 for completion of the study. serologically-indicated influenza infection was the primary outcome of this study. it was define as a ‡4 fold rise in antibody titer within each influenza season. other study outcomes included rt-pcr confirmed influenza virus infection, acute respiratory illness (ari) (two of any of the above listed signs or symptoms), and influenza-like illness (ili) (fever ‡37.8°c with cough or sore throat). antibody titers against the vaccine strains were obtained by testing each serum specimens by haemagluttination inhibition (hai). viral microneutralization (vn) using standard methods was found to be more sensitive than hai in detecting antibody response against a ⁄ california ⁄ 04 ⁄ 2009(h1n1) in another study conducted by our group 27 and was, therefore, used in this study. the sera was initially diluted at 1 ⁄ 10 and further tested in serial doubling dilutions. nose and throat swabs were tested by reverse transcription polymerase chain reaction (rt-pcr) for influenza a and b viruses. technical details of the laboratory methods have been reported elsewhere. 27, 28 fisher's exact test and chi-squared tests were used to compare count data including occurrence of side effects, laboratory confirmed, and clinically defined influenza infections. wilcoxon signed-rank test were used to compare the serum antibody titers between groups. exact binomial method or the wald approximation was used to estimate 95% confidence intervals where appropriate. all analyses were carried out in r version 2.8.1 (r development core team, vienna, austria). twenty-five primary and secondary schools in the district of the study clinic were invited to participate. to parents of three schools that agreed to take part and another study cohort, 3690 invitation letters were sent and 105 households were enrolled. personal referrals were made from these parents to enroll 14 additional households. among 119 enrolled households, 1 subject with history of epileptic seizure was assessed to be contra-indicated against receiving the vaccine. blood taking failed in another subject, and both of them withdrew from the study. eleven households did not complete the study. table 1 shows subject and household contacts of the tiv and placebo group were similar in demographics and prior influenza vaccination history. antibody titers before vaccination were comparable between groups (data not shown). most study subjects who received tiv showed antibody titer ‡40 against the vaccine strains 1 month after receiving tiv, and the proportion was significantly higher than those who received placebo (a ⁄ h1n1 93% in tiv versus 68% in placebo group, p < 0.01; a ⁄ h3n2 97% versus 61%, p < 0.01; b 99% versus 91%, p = 0.15). none of the study subjects had antibody titer ‡40 against ph1n1 following receipt of seasonal tiv. no serious adverse reactions were reported, and only pain at injection sites was slightly higher in tiv group (data not shown). subjects who received tiv had lower rates of serologically confirmed seasonal influenza a(h1n1) (8% versus 21%, p = 0.10), a(h3n2) (7% versus 12%, p = 0.49) and b infection (3% versus 8%, p = 0.36, although the differences were not statistically significant (table 2) . study subjects had higher rate of serologically confirmed ph1n1 infection (32% versus 17%, p = 0.09), yet it was not statistically significant. after adjusting for potential cross reactive antibody response, 31% of subjects in tiv versus 12% in placebo groups showed ph1n1 infection confirmed by either serology or rt-pcr (p = 0.04). little differences were observed for rt-pcr confirmed infection, ari, and ili in results combining the winter and summer influenza seasons. during winter season when seasonal influenza predominated, study subjects who had received tiv showed a lower tendency to develop ili (15% versus 23%, p = 0.43) or ari (46% versus 54%, p = 0.52). an opposite tendency was seen (ili 27% versus 19%, p = 0.43; ari 49% versus 44%, p = 0.68) during summer when ph1n1 predominated. however, these differences were not statistically significant. rates of ili in subjects infected with ph1n1 did not differ statistical significantly between subject who received tiv and placebo (40% versus 25%, p = 0.30). the study was not powered to detect indirect benefits to household contacts of vaccines resulting from reduced household transmission. attack rates were found to be similar between household contacts of subjects received tiv and placebo (data not shown). to examine potential factors that might affect risk of laboratory confirmed ph1n1 infection, a multivariable logistic regression model was fitted to study all subjects and their household contacts. younger participants aged below 16 years were found to have a higher risk (<16 years or = 6.60, 95% ci 2.17, 20.13; 16-45 years or = 2.53, 95% ci 0.80, 7.99, >45 or = 1.00). after adjusting for age, sex, and date of study completion, receipt of tiv for the 2008-9 influenza season was not found to affect risk of ph1n1 infection. however, participants who had laboratory confirmed seasonal influenza infection during the study period had 65% lower risk of ph1n1 infection (infected with seasonal influenza or = 0.35, 95% ci 0.14, 0.87; not infected with seasonal influenza or = 1.00). as (see table s1 for winter and summer results separately). influenza-like illness (ili) defined as temperature ‡37.8°c plus cough or sore throat; acute respiratory illness (ari) defined at least any two of fever ‡37.8°c, chills, headache, sore throat, cough, presence of phlegm, nasal congestion, runny nose, muscle or joint pain. limited by the sample size, we were not able to differentiate between the protective effect of seasonal a(h1n1) and a(h3n2) infection against ph1n1. other details of the results from the study were published elsewhere. 17 discussion a non-significantly higher rate of ph1n1 infection was observed in study subjects who received tiv compared to placebo. results from a multivariable logistic regression suggested that such a pattern might be explained by more common seasonal influenza infection in placebo group prior to the pandemic, protecting the placebo group against ph1n1. seasonal influenza infection within 3-6 months observed in our study might have conferred better cross protection than tiv against ph1n1. this resembles similar previous findings on cross protection between influenza infections in human and animal studies. [29] [30] [31] [32] [33] [34] [35] [36] however, the same phenomenon has not been observed in some studies on seasonal influenza vaccine against ph1n1. 18, 21, [23] [24] [25] apart from differences in study design and vaccine used, we speculate that a short time interval between ph1n1 and most recent seasonal influenza peak activities might be crucial for the phenomenon. hong kong is a subtropical area where the 2009 pandemic was preceded immediately by summer seasonal influenza circulation and a few months apart from the winter 2008-9 influenza peak. if cross protection from seasonal influenza lasts for only a short period, it might have waned below partial cross protection from tiv over time from last seasonal influenza infection. the current study is limited by a small sample size, and further studies are required to confirm our hypothesis. while tiv is only effective against matching strains, a universal influenza vaccine could provide better protection against the ever evolving influenza viruses. introduction immunisation of healthy, as well as high risk, children has been the focus of much recent attention both in prevention of seasonal influenza and during the 2009 h1n1 pandemic. detailed information on reactogenicity, particularly for newer vaccine formulations that include adjuvants, is limited. we recently reported results of a head-to-head comparison of two 2009 h1n1 pandemic influenza vaccines in children in the uk. 1 here we present new, detailed analyses of reactogenicity data from that study, which has important potential implications for future paediatric influenza vaccine development and use. we compared the safety, reactogenicity, and immunogenicity of two h1n1 influenza vaccines, one as03 b (tocopherol based oil in water emulsion) adjuvanted egg culture derived split virion, the other non-adjuvanted cell culture derived whole virion, given as two dose schedules 21 days apart, in a randomised, open label trial as previously reported. the study was age stratified (6 months to under 3 years & 3-12 years) to ensure adequate data in young children. age appropriate safety data (simplified for under 5 year olds) were collected for 7 days after each vaccine dose and serum was collected at enrolment & 21 days after the second dose. nine hundred-thirty seven children received vaccines as per-protocol. when comparing the two vaccines, grade 3 ( ‡50 mm) local reactions were seen more frequently following the adjuvanted than the non-adjuvanted vaccine in both age groups, after both vaccine doses. in children over 5 years old, 7ae2% versus 1ae1%, p < 0ae001, after dose one; 8ae5% versus 1ae1%, p = 0ae002, after dose two, in children under 5 years old, 1ae5% versus 0ae0%, p = 0ae06, after dose one (non significant, ns); 5ae9% versus 0ae0%, p < 0ae001 after dose two. fever ‡38°c (axillary measurement) was seen more frequently following the second dose of the adjuvanted vaccine compared to the non-adjuvanted vaccine in <5 year olds (22ae4% versus 12ae5%; p < 0ae05). looking specifically at the adjuvanted vaccine in under 5 year olds, comparing the second dose with the first, there were significantly higher rates of fever ‡38°c (axillary measurement) (22ae4% versus 8ae9%, p < 0ae001), local grade 3 ( ‡50 mm) reactions (5ae9% versus 1ae5%, p = 0ae02), pain (39ae4% versus 31ae5, p = 0ae02), use of analgesia or antipyretic medication (43ae7% versus 31ae5%, p < 0ae001), and decreased activity (31ae9% versus 20ae4%, p < 0ae001). the adjuvanted vaccine was significantly more immunogenic, most notably in the younger children. in <3 year olds, haemagglutination inhibition (hi) seroconversion rates were 98ae2% versus 80ae1%, p < 0ae001. among all general and local reactions measured, only the maximum temperature measured during the 7 days after the second dose of the adjuvanted vaccine showed a significant (positive) association with post vaccination hi titres. for each 1°c rise in temperature there was a 27% increase in titre (p < 0ae001). these reactogenicity data demonstrate a step towards the future possibility of one-dose influenza immunisation programmes for young children associated with low rates of fever and other reactions. the occurrence of fever following adjuvanted vaccine, seen particularly after a second dose in younger children, was quantitatively associated with enhanced antibody titres. this association was not seen with unadjuvanted vaccine. this apparent difference between the relatedness of the pyrogenic and immunogenic effects of the two vaccines merits further investigation. novel adjuvants appear to have the potential to overcome the relatively poor immunogenicity previously experienced with inactivated influenza vaccines in infants and young children. however, careful adjustment may be needed to optimise the balance between high protection and acceptable reaction rates. tries causing sporadic human infections. vaccination has been used as an effective public health tool for influenza prophylaxis. the goal of this study was to evaluate live attenuated influenza vaccine (laiv) vaccine candidates for subtypes h1 and h5. the attenuated phenotype of h1 and h5 laiv candidates has been proven in experiments in ovo and in vivo. in randomized clinical trials among adult volunteers, no significant adverse reactions attributable to the live vaccine occurred. our results indicate that pandemic laiv candidates were well tolerated and elicited serum, local, and cellular immune responses. the emergence and spread of highly pathogenic avian influenza h5n1 viruses in avian populations and concurrent infections in humans since 1997 has prompted efforts to develop vaccines for use in the event of an influenza pandemic. in 2009, the world faced a new h1n1 pandemic. immunization with inactivated or live vaccines is the primary measure for preventing influenza. laivs appear to be safe and efficacious, and might possibly provide broader immune responses than inactivated vaccines. our study evaluated laiv pandemic candidates as part of the global influenza pandemic preparation project outlined by the who. capacity of the viruses to grow at optimum, low, and elevated temperatures (ca ⁄ ts phenotype) was evaluated by routine technique in embryonated hen eggs. laiv and placebo were supplied by microgen (irkutsk, russia). the monovalent laiv was produced from the pandemic vaccine candidates and formulated to contain 10 7 and 10 6ae9 eid 50 per dose (0ae5 ml) of a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 and a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 92, respectively. the vaccine or placebo was administered intranasally with a single-use dosing nasal sprayer. two doses were given at an interval of 21 days. one hundred-ninety healthy adults aged 18-60 years were randomly divided into groups to receive either pandemic vaccine candidates (204) or placebo (28) . subjects were informed about purposes and methods of the study and potential risks associated with participation. all participants had an hai antibody titer of £1:10 to a ⁄ california ⁄ 7 ⁄ 2009 (h1n1) pandemic virus. in all there were 47 and 42 vaccines and 19 and 8 participants who received placebo, and were further tested for immune responses to h1n1 or h5n2 pandemic vaccine, respectively. another 29 participants vaccinated with h1n1 laiv were children between 12 to18 years old. before the children were vaccinated, their parents were advised about study and their consent was required before any child was enrolled. on the advice of the national ethics committee, we did not include a placebo group in this study. individuals were not enrolled if they had an acute illness or fever at the beginning of the study or a history of egg allergy. immune responses of subjects were assessed by routine hai test (evaluation of serum igg antibodies), elisa (evaluation of iga antibodies eluted from the nasal swabs into steril pbs), and cytokine flow cytometry assay (evaluation of virus-specific cd3 + cd4 + ifnc + and cd3 + cd8 + ifnc + peripheral blood mononuclear cells). the results of phenotypic analysis in ovo showed that pandemic vaccine candidates retained the cold adapted-temperature sensitive (ca ⁄ ts) phenotype, typical of the coldadapted parental mdv. in contrast and as expected, a ⁄ california ⁄ 07 ⁄ 2009 and a ⁄ duck ⁄ potsdam ⁄ 1406-86 parental strains had the non-ts ⁄ non-ca phenotype typical of wt viruses. the h5n2 pandemic vaccine candidate demonstrated an attenuated phenotype in mice and in java macaques and did not infect chickens. the vaccine attenuation study confirmed the attenuated phenotype of a a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 pandemic laiv candidate in mouse, ferret, and guinea pig models. the phase i ⁄ ii randomized, controlled, double-blind clinical study safety evaluation of pandemic vaccine candidates in adults clinical examination of subjects who received two doses of pandemic vaccine candidates indicated that both vaccines were well tolerated. no fever reactions were observed after the first or second vaccination. after the first vaccination, 33ae0% and 40ae0% of reactogenicity events consisting of catarrhal symptoms, such as pharyngeal irritation or hyperemia, were observed for h1n1 and h5n2 vaccine candidates, respectively. after revaccination, subjects did not report local or systemic reactions. to determine whether a serological response occurred in the cohort of immunologically naïve subjects vaccinated with pandemic vaccine candidates, hai and elisa tests were used (table 1) . post-vaccination geometrical mean titers (gmt) among 20 subjects who received two doses of h5n2 vaccine were significantly higher than pre-vaccination titers. the frequency of ‡4 fold antibody rises was significantly higher (47ae1%) after revaccination than after one dose (5ae9%). the percentage of subjects with post-vaccination serum hai titers to h5n2 ‡ 1:20 was 47ae1% and for titers ‡1:40, it was 29ae4%. no seroconversions in the placebo group were detected. the virus-specific nasal iga antibody response to vaccination after two doses of the h5n2 vaccine candidate demonstrated significant increases of ‡4 fold rise iga antibodies (65%) compared to one dose. cumulative data of h5n2 vaccination (all applied tests) showed 35% and 80% of conversions after the first and the second vaccination, respectively. increasing h5n2 vaccine virus infectivity from 10 6ae9 to 10 8ae3 eid 50 ⁄ dose lead to an enhancement of post-vaccination hai titers in vaccinees after the first vaccination to homologous h5n2 antigen from 5ae9% to 31ae0% of ‡4 fold antibody rises. values of post-vaccination serum hai antibody titers in subjects vaccinated with another pandemic vaccine candidate, a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38, also proved to be rather low. after the primary vaccination, the percentage of subjects with hai protective antibody titers ‡1:40 were 2ae1%. after revaccination, this parameter increased to 17ae0%. four-fold increases in serum hai antibody titres were four-fold conversions after the first and the second vaccination was 23ae4% and 34ae4%, respectively. elisa antibodies in nasal swabs showed had an advantage in detecting induction of local iga as compared to serum hai antibodies. after revaccination four-fold serum hai antibody conversions were 34ae4% vs. 63ae8% of iga conversions in nasal swabs, respectively. taking into account cumulative data of h1n1 vaccination (hai and elisa data), the obtained results were here and in the 42ae5% and 70ae2% of conversions after the first and the second vaccination, respectively. fourty-seven subjects were vaccinated with h1n1 laiv, and 19 who received a placebo were chosen for evaluation of cellular immune response by cytokine assays. after revaccination, the mean increases of both cd4 + and cd8 + memory cells were significantly higher in vaccinated subjects compared to the placebo group. interestingly, the same effect of vaccination was observed in vaccinees without detectable conversions of hai antibody titers. even after a single vaccination, the rate of subjects with significant increases of these cells in the blood was 37ae5% (cd8 + ) and 75% (cd4 + ). after the revaccination, the percentage of subjects with significant increases in cd8 + and in cd4 + cells was 68ae8%. immunogenicity of h1n1 pandemic vaccine candidate in children hai antibody results among children aged 12 to 18 years proved to be significantly higher when compared to adult subjects: after the first vaccination, 41ae4% of the children seroconverted; after revaccination, seroconversions reached 83ae3% ( table 2 ). the gmt rise to h1n1 vaccine with primary vaccination was 3:1; after revaccination it increased to 6:7. benefits of vaccination with laiv to aid in the control of influenza outbreaks are acknowledged by the who. 1 many years of laiv seasonal trials have shown excellent tolerability and low reactogenicity. [2] [3] [4] indeed, data showed that live influenza vaccines cause minimal systemic, local, and thermal reactions, generally from 0 to 3%. a different situation was observed in the cohort of immunologically naïve volunteers vaccinated with pandemic vaccines. the rate of local reactions to a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 and a ⁄ 17 ⁄ duck ⁄ potsdam ⁄ 86 ⁄ 9 vaccine candidates increased to 33ae0% and 40ae0%, respectively. after revaccination no significant local and systemic reactions were observed. this confirms, indirectly, the development of a sufficiently high level of protection after the first vaccination with pandemic laiv. the most important criterion for assessing the quality of vaccines is their estimated safety, epidemic effectiveness, and immunogenicity. however, current regulatory documentation 5 mandates that induction of serum antibodies, measured by hai, as the only criterion for a laiv immunogenicity evaluation. in addition to the standard hai assay, we determined serum (igg) and local (iga) antibodies in adult subjects vaccinated with an h1n1 pandemic vaccine candidate. evaluation of overall results obtained in these additional serological tests, as well as those from the hai assay, showed an immune response to the vaccine in the majority of subjects (42ae5% of ab seroconversions after the single vaccination and 70ae2% after revaccination, respectively). these data show that methods used to routinely measure laiv immunogenicity should be revised to include a number of additional immunological methods such as igg and iga elisa, and cytokine assays consistent with the recently updated who recommendations on laiv monitoring. these clinical studies clearly demonstrated that pandemic laiv candidates are effective at generating pandemic specific influenza immunity. a key finding from this study is that it may be practical to give the vaccine as a single dose to both children and adults. evaluation of our laiv pandemic vaccine candidates was performed as part of the global influenza pandemic preparation project outlined by the who. 6 it was considered that laiv could be produced in greater quantities and more rapidly than inactivated vaccines. together with the generation of herd immunity by laiv, this suggests that laiv implementation during the first wave of a pandemic may provide significant social, economic, and health benefits to the community. authors are thankful to path for the financial support of h1n1 pandemic vaccine study. we are grateful for the the main evolutionary mechanism of influenza viruses during inter-pandemic period is the antigenic drift, but the epidemiological picture of circulating viruses is complicated by a high level of heterogeneity of strains, even though drift does not occur, due to co-circulation of drifted and old strains or to co-circulation of viruses belonging to the same type ⁄ subtype but with different antigenic patterns. [1] [2] [3] [4] [5] [6] lack of data exists on the impact of the wide heterogeneity of circulating strains on the seroprotection and on-field effectiveness of influenza vaccine: in particular, little is known about the ability of influenza vaccine to elicit an effective immune response against isolates with few amino acid mutations with respect to vaccine strains that represent the majority of circulating viruses. mf59-adjuvanted vaccines, which are currently used for the prevention of seasonal influenza epidemics in elderly, are showed to confer higher seroprotection against homologous and drifted a(h3n2) strains than non-adjuvanted vaccines. [7] [8] [9] the broader immune response showed by mf59-adjuvanted vaccine was measured using hi and nt assays against egg-grown drifted strains representing vaccine composition changes during the following seasons, but its ability to elicit a broader immune response against circulating viruses belonging to vaccine cluster and presenting amino acid mutations onto antigenic sites or against on-field isolates not-antigenically distant from vaccine strains has not yet been investigated. showing amino acid changes onto antigenic sites in position 145 (n145k), 189 (n189k), and 227 (p227s) with respect to a ⁄ california ⁄ 7 ⁄ 04. in particular, a ⁄ genoa ⁄ 13 ⁄ 04 and a ⁄ genoa ⁄ 27 ⁄ 04 presents n126d amino acid mutation detected in clade 2 a ⁄ wyoming ⁄ 3 ⁄ 03-like viruses. 1 the ha sequences of a ⁄ genoa ⁄ 59 ⁄ 04, a ⁄ genoa ⁄ 2 ⁄ 05, genoa ⁄ 11 ⁄ 05, a ⁄ genoa ⁄ 47 ⁄ 05, and a ⁄ genoa ⁄ 62 ⁄ 05 fell within the clade represented by the ha of a ⁄ califor-nia ⁄ 7 ⁄ 04; among these isolates, a ⁄ genoa ⁄ 2 ⁄ 05 and a ⁄ genoa ⁄ 11 ⁄ 05 showed antigenic site sequences very close to that of the 2005 ⁄ 06 vaccine strain, whereas ha sequences of a ⁄ genoa ⁄ 62 ⁄ 05, a ⁄ genoa ⁄ ⁄ 47 ⁄ 05 and a ⁄ genoa ⁄ 59 ⁄ 05 posses amino changes onto antigenic site a(r142k), c(g50e) and d(r208k), respectively. the ha sequences of more recent isolates fell within the clade represented by the ha of a ⁄ brisbane ⁄ 10 ⁄ 07 and characterized by the amino acid changes, relative to the ha of a a ⁄ california ⁄ 7 ⁄ 04, g50e and k140i, with the exception of a ⁄ genoa ⁄ 3 ⁄ 07, showing r142g and l157s amino acid changes present in viruses belonging to a ⁄ nepal ⁄ 921 ⁄ 06 clade. measure of genetic distance between vaccine and circulating strains was calculated as previously described by gupta. 10 two blood samples were collected from each subject, just before and 22 ± 2 day post-vaccination. all sera were stored at )20°c. all samples were tested at the laboratory of health sciences department, university of genoa, by haemagglutination-inhibition (hi) and neutralization (nt) assays, performed following the who criteria and standardised method in our laboratory, respectively. [11] [12] [13] guinea pig red blood cells were used for hi assay. all samples were assayed twice for hi and for nt. the obtained antibody titre was expressed as the reciprocal of the last sera haemagglutinating or inhibiting virus dilution. immunogenicity was determined by: geometric mean titre (gmt); mean-fold increase (mfi; ratio of post-to pre-vaccination titre); seroprotection rate (the percentage of subjects achieving an hi and nt titre ‡40 iu); and seroconversion rate (percentage of subjects with a fourfold increase in hi or nt antibody titers, providing a minimal post vaccination titer of 1:40). post-vaccination gmt was reported as ratio, with the corresponding 95% confidence interval, of gmts after vaccination with mf59-adjuvanted vaccine and with non-adjuvanted subunit vaccine. seroprotection and seroconversion rate 95% confidence interval was calculated using modified wald method. comparisons of seroconversion and seroprotection rates between subunit and mf59-adjuvanted vaccine groups have been analyzed by fischer's exact test. the results were evaluated against the committee for medicinal products for human use (chmp) criteria for approval of influenza vaccines in the elderly, which require that at least one of the following criteria be met: mfi >2; seroprotection rate >60%, or seroconversion rate >30%. furthermore, hi titres were also transformed into binary logarithms, corrected for pre-vaccination status, as described by beyer et al. 14 and were expressed as median titres, with the corresponding 25°-75°i nter-quantile range. comparisons of corrected post-vaccination titers between subunit and mf59-adjuvanted vaccine groups were analyzed by wilcoxon test. difference in immunogenicity profile between vaccine groups, expressed by ratio of different parameters, was correlated with genetic and antigenic distance between vaccine and viruses used in the study using spearman test. pre-vaccination titres were not significantly different between vaccine groups, for all 15 strains (data not shown). post-vaccination gmt ratios between mf59-adjuvanted and non-adjuvanted vaccine groups determined using hi and nt assays, with the corresponding 95% confidence interval, according to viral strain are shown in figure 1 . both vaccines met chmp requirements for mfi (>2), seroconversion (>30%), and seroprotection rate (>60%) against a ⁄ wyoming ⁄ 3 ⁄ 03-like, with the exception of a ⁄ genoa ⁄ 13 ⁄ 04 and a ⁄ california ⁄ 7 ⁄ 04-like circulating viruses and against egg-grown a ⁄ wyoming ⁄ 3 ⁄ 03, a ⁄ california ⁄ 7 ⁄ 04, and a ⁄ wisconsin ⁄ 67 ⁄ 05 strains; the immune response against a ⁄ genoa ⁄ 13 ⁄ 04 met the requirements for mfi and seroprotection rate only in mf59-adjuvanted vaccine group. requirements for mfi, seroconversion, and seroprotection rate against the a ⁄ brisbane ⁄ 10 ⁄ 07-like virus a ⁄ genoa ⁄ 2 ⁄ 07 and the a ⁄ nepal ⁄ 921 ⁄ 06-like genoa ⁄ 3 ⁄ 07 viruses and against egg-grown a ⁄ brisbane ⁄ 10 ⁄ 07 strain were reached only in subjects vaccinated with the mf59adjuvanted vaccine. a similar pattern emerged from the analysis of mfi, seroconversion and seroprotection rates using nt assays. subjects vaccinated with the mf59-adjuvanted vaccine showed significantly higher post-vaccination hi gmts against a ⁄ wyoming ⁄ 3 ⁄ 03-like, a ⁄ california ⁄ 7 ⁄ 04-like, a ⁄ nepal ⁄ 921 ⁄ 06-like and a ⁄ brisbane ⁄ 10 ⁄ 07like viruses, with the exception of a ⁄ genoa ⁄ 3 ⁄ 06, and against egg-grown a ⁄ california ⁄ 7 ⁄ 04, a ⁄ wisconsin ⁄ 67 ⁄ 05, and a ⁄ brisbane ⁄ 10 ⁄ 07 strains, compared with individuals immunized with the non-adjuvanted vaccine ( figure 1 ). the mf59-adjuvanted vaccine also induced significantly higher seroconversion and seroprotection rates against following correction for pre-vaccination status, hi titres were significantly higher for the mf59-adjuvanted vaccine group when evaluated against a ⁄ wyoming ⁄ 3 ⁄ 03-like viruses, a ⁄ brisbane ⁄ 10 ⁄ 07-like a ⁄ genoa ⁄ 2 ⁄ 07, and a ⁄ nepal ⁄ 921 ⁄ 06-like a ⁄ genoa ⁄ 3 ⁄ 07 strain ( figure 2 ). pre-vaccination titre corrected response was higher in subjects vaccinated with mf59 adjuvanted vaccine also against egg-grown a ⁄ wyoming ⁄ 3 ⁄ 03, a ⁄ california ⁄ ⁄ 7 ⁄ 04, a ⁄ wisconsin ⁄ 67 ⁄ 05, and a ⁄ brisbane ⁄ 10 ⁄ 07. among viruses more closely related to a ⁄ california ⁄ 7 ⁄ 04, subjects immunized with mf59-adjuvanted vaccine showed a significantly higher corrected titres against a ⁄ genoa ⁄ 59 ⁄ 04, a ⁄ genoa ⁄ 47 ⁄ 05, and a ⁄ genoa ⁄ 62 ⁄ 05 strains compared with the non-adjuvanted vaccine ( figure 2) . spearman test showed a clear correlation between the distances and the advantage offered by mf59 expressed by ratio between mfi, post-vaccination gmts, corrected post-vaccination median, seroconversion, and seroprotection rates calculated using hi test in the two vaccine groups. similarly, ratio between mfi, seroconversion, and seroprotection rates calculated with nt test correlated with the genetic and antigenic distance between vaccine and viruses used for the study. the ability of mf59 to enhance the immunogenicity and to elicit a broader immune response against drifted strains than non-adjuvanted vaccine is consistent with other findings reported during the last decade. [7] [8] [9] 15 in subjects vaccinated with the mf59-adjuvanted vaccine containing a ⁄ california ⁄ 7 ⁄ 04, the immune response, expressed by a number of parameters, such as crude and corrected postvaccination titers, seroconversion, and seroprotection rates calculated using hi and nt assays, is higher than that observed in individuals immunized with subunit vaccine when it is evaluated against a drifted strains, such as a ⁄ brisbane ⁄ 10 ⁄ 07-like and a ⁄ nepal ⁄ 921 ⁄ 06-like strains, and against egg-grown a ⁄ brisbane ⁄ 10 ⁄ 07 virus. for the first time in this study, the impact of heterogeneity of circulating strains antigenically close to the vaccine on the antibody response elicited by mf59-and non-adiuvanted vaccines is evaluated. immune response against viruses isolated during the 2004 ⁄ 05 season, that appear more phylogenetically close to 2004 ⁄ 05 vaccine strain a ⁄ wyoming ⁄ 3 ⁄ 03, was higher in subjects vaccinated with mf59-adiuvanted vaccine as demonstrated by higher crude and corrected post-vaccination hi titres and higher postvaccination nt titres, with the exception of a ⁄ genoa ⁄ 27 ⁄ 04, against whom the nt post-vaccination gmt is identical in mf59 and subunit vaccine groups. furthermore, hi seroconversion and seroprotection rates were higher in mf59 vaccine group when evaluated against a ⁄ genoa ⁄ 13 ⁄ 04 and a ⁄ genoa ⁄ 27 ⁄ 04. as far as the immune response against a ⁄ california ⁄ 7 ⁄ 04-like viruses, the small number of enrolled subjects did not allow appreciating differences using qualitative response indicators, but crude post-vaccination hi titres were higher in mf59 vaccine group for all the strains. interestingly, a ⁄ california ⁄ 7 ⁄ 04-like viruses with at least one amino acid change onto antigenic sites, i.e. a ⁄ genoa ⁄ 59 ⁄ 04, a ⁄ genoa ⁄ 47 ⁄ 05, and a ⁄ genoa ⁄ 62 ⁄ 05, showed a more marked difference in terms of response between the two vaccine groups. individuals immunized with mf59-adiuvanted vaccine showed higher corrected post-vaccination hi titres and post-vaccination nt titres in comparison with subjects vaccinated with plain vaccine. these response indicators were similar in the two vaccine groups when the response was evaluated against a ⁄ genoa ⁄ 2 ⁄ 05 and a ⁄ genoa ⁄ 11 ⁄ 04, which present no amino acid changes onto antigenic sites and identical hi titers respect with a ⁄ california ⁄ 7 ⁄ 04 at molecular and antigenic characterization, respectively. thus, the advantage offered by mf59 in terms of higher immunogenicity expressed by higher post-vaccination hi titres is observable also against viruses showing antigenic and molecular pattern undistinguishable from vaccine strain, but it became even more evident as the antigenic and molecular distance between vaccine and circulating strains grew. as emerged for a ⁄ genoa ⁄ 59 ⁄ 04, a ⁄ genoa ⁄ 47 ⁄ 05, and a ⁄ genoa ⁄ 62 ⁄ 05, one amino acid was a sufficient change in antigenic sites for 2-fold decrease of hi titre against homologous vaccine strain to observe 2-fold higher post-vaccination nt titers (mf59 ⁄ subunit postvaccination gmt ratio range between 1ae79 and 2ae45, figure 1) and one-dilution higher corrected post-vaccination hi titers in mf59 vaccine group ( figure 2) . finally, the correlation between the distance and the improvement offered by mf59 in terms of higher immunogenicity clearly emerged by spearman correlation analysis: it remains wellfounded both using a number of different response parameters obtained from hi and nt assays and calculating the distance by serological and genetic methods. outbreaks of h1n1pdm in pigs in commercial swine operations have been reported in several countries. in all incidents, epidemiological investigations have linked humans as the possible source of the infection to pigs. experimentally, it was established that the virus is pathogenic and transmits readily in pigs. 1 the natural outbreaks of h1n1pdm and laboratory studies underscore the threat that the virus poses to the swine industry and highlight the need for developing effective control strategies. in the united states, a trivalent live attenuated influenza vaccine (flumistò) has been licensed for use in humans since 2003. 2 in swine medicine, however, temperature-sensitive laivs are not available. currently, only inactivated vaccines are available for pigs, but they provide limited protection against antigenically diverse influenza viruses. additionally, the use of inactivated vaccines has been associated with enhanced pneumonia when immunized pigs were challenged with divergent viruses. 3 thus, the development of laivs has the potential to circumvent the drawbacks associated with commercial vaccines. with the aim of developing laiv temperature-sensitive influenza vaccines against the h1n1pdm virus, we have used reverse genetics to introduce attenuation markers in the polymerase genes of a swine-like tr h3n2 influenza virus, a ⁄ turkey ⁄ ohio ⁄ 313053 ⁄ 04 (h3n2) (ty ⁄ 04). 4 we chose this isolate because it grows well in both eggs and cell culturebased substrates, displays a broad host range, and has internal genes similar to the h1n1pdm virus. safety and efficacy studies of the ty ⁄ 04 att vaccine candidates in pigs demonstrated that this vaccine backbone is attenuated in swine and conferred sterilizing immunity upon an aggressive intratracheal challenge of pigs with the 2009 h1n1 pandemic virus. thus, introduction of genetic signatures for att in the backbone of a swine-like tr influenza virus resulted in highly attenuated and efficacious live influenza vaccines with promising applications veterinary medicine. 293-t cells and mdck cells were maintained as previously described. 5 a ⁄ turkey ⁄ ohio ⁄ 313053 ⁄ 04 (h3n2) (ty ⁄ 04) has options for the control of influenza vii ª 2011 blackwell publishing ltd, influenza and other respiratory viruses, 5 (suppl. 1), 328-394 been previously described and it was kindly provided by yehia saif, ohio state university. 4 a ⁄ california ⁄ 04 ⁄ 09 (h1n1) (ca ⁄ 04) was kindly provided by the centers for disease control and prevention (cdc). generation of recombinant viruses by reverse genetics (rg) was done using a previously described method. 6 the genetic signatures for attenuation 5 were introduced into the pb2 and pb1 genes of ty ⁄ 04. ny 2:6 ty ⁄ 04 att is a 2:6 reassortant with the surface genes from the a ⁄ new york ⁄ 18 ⁄ 2009 (h1n1) virus and the ty ⁄ 04 att internal genes. all viruses were amplified in mdck cells to produce viral stocks. twenty-five pigs were divided into five groups (n = 5) and intranasally inoculated with 10 5 tcid 50 ⁄ animal of either h3n2:6ty ⁄ 04 att or with ny(h1n1)2:6ty ⁄ 04 att vaccines diluted in 2 ml of mem. two other groups were similarly inoculated with h3n2:6ty ⁄ 04 wt and h3n2:6ty ⁄ 04 rg and served as controls, whereas a fifth group was mockvaccinated with pbs alone. clinical observations were performed as previously described. 1, 7 efficacy of h1n1 ty ⁄ 04 att vaccine in pigs fourty pigs were divided in four groups (n = 10)( table 1) . group 1 was vaccinated with 10 5 tcid 50 ⁄ animal of ny(h1n1)2:6ty ⁄ 04 att through intranasal route, whereas group 2 was vaccinated intramuscularly with 2 ml of an adjuvanted uv-inactivated ca ⁄ 04 vaccine (uvadj-ca ⁄ 04). 7 group 3, non-vaccinated and challenged (nv+ca ⁄ 04), and group 4, non-vaccinated, mock-challenged (nv+mock), were also included. pigs were boosted two weeks later. fourteen days post boost (dpb), pigs from groups 1-3 were challenged intratracheally with 2 ml of 1 · 10 5 tcid 50 of ca ⁄ 04. following challenge, pigs were monitored using methods as previously described. 7 all statistical analyses were performed using graphpad prism software version 5ae00 (graphpad software inc., san diego, ca). the differences were considered statistically significant at p < 0ae05. the ty ⁄ 04 att-based vaccines are attenuated in swine pigs inoculated with wt ty ⁄ 04 viruses developed fever (>40°c) that peaked 24 hpi ( figure 1a) and shed large amounts of in nasal secretions ( figure 1b) . similarly, viral titers in bronchoalveolar lavage fluid (balf) collected at 3 dpi ranged from 10 5 to 10 6 tcid 50 ⁄ ml ( figure 1c ). at necropsy, the lungs from animals inoculated with these viruses had severe pneumonia ( figure 1d ). in contrast, none of the animals inoculated with h3n2 or h1n1 ty ⁄ 04 att viruses developed clinical signs following vaccination, indicating that the ty ⁄ 04 att viruses were safe for administration to pigs ( figure 1a) . correspondingly, there was 100-1000 fold less virus shedding from the nose of pigs vaccinated with ty ⁄ 04 att viruses as compared to unmodified ty ⁄ 04 viruses. in general, ny(h1n1)2:6ty ⁄ 04 att -vaccinated pigs shed less virus than h3n2:6 ty ⁄ 04 att inoculated pigs ( figure 1b ). in addition, viral titers in balf were significantly reduced (p < 0ae01) in ty ⁄ 04 attvaccinated pigs as compared to ty ⁄ 04 wt-infected pigs ( figure 1c ). although both vaccines caused mild gross and microscopic lesions in the lungs, the percentage of lung 0ae2 ± 0ae1* 0 ± 0* 0 ± 0* 0 ± 0* balf, bronchoalveolar lavage fluid, uvadj-ca ⁄ 04, uv-inactivated ca ⁄ 04 vaccine; nv+ca ⁄ 04, non-vaccinated, challenged positive control group; nv+mock, non-vaccinated, non-challenged negative control group. *significantly different from nv+ca ⁄ 04 control group at p < 0ae05. geometric mean hi titer against ca ⁄ 04 at the day of challenge. à percentage of macroscopic lung lesions given as mean score ± sem. § average viral titer (log 10 ) measure as tcid 50 per ml. -average viral titer (log 10 ) in balf at 5 dpc. involvement was not significantly different from mock-vaccinated pigs, corroborating the clinical findings that these vaccines are sufficiently attenuated in pigs ( figure 1d, e) . histopathologically, nasal turbinates and trachea obtained from pigs immunized with either vaccine were similar to control animals, as opposed to the wt-inoculated pigs ( figure 1e ). vaccination with h1n1 ty ⁄ 04 att-based vaccines provides sterilizing immunity against h1n1pdm in pigs the clinical performance in pigs of the h1n1 vaccines is summarized in table 1 . nv+ca ⁄ 04 animals had macroscopic pneumonia, viral replication in balf and shedding in the nose. uvadj-ca ⁄ 04 vaccine provided satisfactory protection, but this protection was not sterilizing. remarkably, animals vaccinated with ny(h1n1)2:6ty ⁄ 04 att had sterilizing immunity. in both vaccine groups there was a significant reduction (p < 0ae001) in the percentage of macroscopic lung pathology compared to the nv+ca ⁄ 04 group. control pigs had neither significant macroscopic nor microscopic lesions in the lungs. hi antibody titers measured at the day of challenge in both vaccine groups were approximately the same (table 1 ). in the present study, we developed for the first time, temperature-sensitive laiv for use in pigs. data from our safety studies showed that both the h3n2 and h1n1 ty ⁄ 04 att vaccines were attenuated in pigs. although the ty ⁄ 04 att vaccines were detected in balf samples, the level of viral replication was significantly reduced in comparison to unmodified virus and, more importantly, caused no overt clinical signs. a minimal amount of replication is likely beneficial for eliciting t-cell responses to internal genes that may provide heterologous cross-protection. one of the most challenging tasks in producing effective live attenuated vaccines is to achieve an adequate balance between safety and efficacy. by introducing the att modifications into the polymerase genes of a swine-like tr strain, this desirable balance was achieved. the vaccines were histopathologic scores of nasal turbinates, trachea and lungs at 3 dpi. ny(h1n1)2:6ty ⁄ 04 att (a virus that carries the surface genes of a ⁄ new york ⁄ 18 ⁄ 09 (h1n1) and ty ⁄ 04 att internal genes). all h3n2 viruses have their surface genes derived from ty ⁄ 04. values are shown as the mean ± sem. * p < 0ae05; **p < 0ae01; *** p < 0ae001. options for the control of influenza vii ª 2011 blackwell publishing ltd, influenza and other respiratory viruses, 5 (suppl. 1), 328-394 attenuated in pigs and, more importantly, provided sterilizing immunity upon an aggressive challenge with pandemic h1n1 as opposed to an experimental ca ⁄ 04 inactivated vaccine, which elicited protective but not sterilizing immunity in all animals. in the face of influenza pandemics that have the ability to overcome the species barriers such as the 2009 h1n1, the supply of vaccines for use in agriculture could be jeopardized. our cell culture-based live att h1n1 vaccines could be an attractive alternative for this possible pandemic vaccine shortage. because the ty ⁄ 04 att live vaccines developed here are efficacious in swine, are easier to manufacture than inactivated vaccines, and do not require adjuvants, our study represents a major advance in vaccine development for the 2009 h1n1 pandemic. in conclusion, our second generation of live att influenza vaccines based on modifications of the pb2 and pb1 genes of ty ⁄ 04 retains its safety properties in vivo and can induce excellent protection against aggressive h1n1 challenges in the swine host. influenza virus is one of the most important respiratory pathogens worldwide. 1, 2 type a influenza causes an acute disease of the upper airways, and affects 20-40 million persons yearly. moreover, the threat of human influenza epidemic and pandemic has dramatically increased in recent years. 3 vaccination is one of the crucial interventions for reducing the spread and impact of influenza. the generally used parenteral inactivated influenza vaccines induce mainly systemic antibody responses and only weak cell-mediated immunity and low levels if any mucosal immunity. on the other hand, intranasal immunization with live virus can induces a broad spectrum of both systemic and mucosal antibodies, and the immune response localized in the mucosa blocks the virus even during the first phase of infection. unfortunately, the use of live vaccines is always associated with a certain risk. the development of a crossprotective vaccine against potentially pandemic strains is an essential part of the strategy to control and prevent a pandemic outbreak. we induced intrasubtypic and intersubtypic cross-protection in balb ⁄ c mice by intratracheal (it) immunization with inactivated influenza viruses together with dead delipidated bacillus firmus (dbf) as an adjuvant. ten days after the 2nd immunization dose, the mice were infected with live influenza virus b ⁄ lee ⁄ 40 lethal for mice (total infection dose corresponded to 5 · ld 50 ) or a⁄ pr 8 ⁄ 34 (total infection dose corresponded to 0ae5-5 ld 50 ). dbf adjuvant markedly increased both systemic and mucosal anti-viral antibody formation when applied together with inactivated influenza a or b viruses. 4 protective significance was tested in vivo. mice were preimmunized with 1) pbs (controls), 2) dbf alone, 3) virus alone, and 4) vir-us+dbf. influenza b virus strains b ⁄ lee and b ⁄ yamanashi 166 ⁄ 98 (58 years phylogenetically distant and antigenically substantially different, especially in terms of the main protective antigen -surface haemagglutinin) or two different influenza a subtypes -a ⁄ pr 8 ⁄ 34 (h 1 n 1 ) and a ⁄ california 7 ⁄ 04 (h 3 n 2 ) -were used (figures 1 and 2) . the mice were challenged with 5 · ld 50 of either b ⁄ lee ⁄ 40 or a ⁄ pr 8 ⁄ 34 as appropriate. all controls died. the mice treated with dbf alone died with a delay or survived, which could be explained by stimulation of innate immunity. the animals immunized with virus alone were protected against homologous strains. adjuvant immunization was cross-protective: the mice immunized with a heterologous b strain (figure 2 ) fell ill (pronounced body mass loss), but almost all survived and recovered. 5 the mice immunized with a heterologous a subtype were excellently protected (negligible weight loss and zero mortality). 6 intratracheal dbf (500 lg per mouse) given to non-immunized mice 24 hour before influenza infection eliminated the lethal effect in 40-100% of infected animals depending on infection dose (0ae5-5 ld50); in mice infected with lower than lethal doses (0ae5 ld50), weight loss was minimized or did not occur. the current mode of vaccination-induced immunity is mostly effective against a homologous strain of the virus used for vaccination. the attention is therefore focused on vaccines that are able to induce cross-protection and could be effective also in case of sudden appearance of a new virus variant. inactivated influenza viruses are known to be often insufficiently effective when used for mucosal immunization and for induction of cross-protection against drifted influenza viruses or novel subtypes. the drawback of vaccination with dead virus can be overcome by using a suitable adjuvant. mouse models were successfully immunized with vaccine containing inactivated virus in combination with cholera toxin or the escheria coli heat-labile toxin (lt). [7] [8] [9] the use of cholera toxin in humans is precluded because of its high toxicity; a number of lt mutants that retain their adjuvant activity have been prepared; these mutants were likewise tested on the mouse model and should not cause any serious side effect in humans. for this reason, current studies aim at finding a suitable and safe mucosal and systemic immune response. dbf has been shown to be a very efficient adjuvant for mucosal immunization stimulating both innate and adaptive immunity. intratracheal immunization with inactivated influenza viruses and dbf as adjuvant induced efficient and even heterosubtypic cross-protection. dbf given 24 hour before infection provided partial protection probably because of its strong stimulatory effect on the innate immunity. temperature-sensitive and cold-adapted candidates for live attenuated influenza vaccine with genomic composition of 7:1 based on highly pathogenic influenza a ⁄ h5n1 viruses with pandemic potential were generated by the replacement of six internal genes from the influenza a ⁄ puerto rico ⁄ 8 ⁄ 34 (pr8) virus from pr8-based rg-candidates for inactivated vaccine with appropriate internal genes of influenza a ⁄ leningrad ⁄ 134 ⁄ 17 ⁄ 57 (h2n2) master donor virus (mdv) for russian laiv by methods of classical reassortment. all attempts to capture avian n1 neuraminidase into the genome of the mdv laiv production were ineffective. 6:2 reassortants were not generated. step by step co-infection of triple reassortants (h5n1-h1n1-h2n2) with h2n2 mdv in some cases was the only possibility to generate influenza a ⁄ h5n2 cold-adapted vaccine reassortants. difficulties in generating 6:2 reassortants could be explained by a substantial gene constellation in the genome of pr8based h5n1 reassortant viruses. strong coupling of pb2 ⁄ pr8 and avian n1 genes in a ⁄ h5n1-pr8-rg reassortants was revealed. annually updated laiv strains are generated by classical reassortment of circulating influenza viruses with well characterized, attenuated, ts ⁄ ca mdvs. resulting attenuated reassortants inherit the relevant ha and na of wild type parental virus and six internal genes of the mdv. 1 candidates for inactivated influenza vaccines based upon avian influenza viruses with pandemic potential are generally generated by reverse genetics methods. 2 in these cases, like with laiv, vaccine strains are 6:2 reassortants which possess the modified ha and na from potentially pandemic virus and six internal genes from the pr8 virus. the pr8 virus is considered to be of low virulence, i.e. attenuated, for humans, yet offers properties of high seed virus growth for influenza vaccine production. the ha of avian h5 influenza viruses with pandemic potential is engineered to remove four basic amino acid codons from the cleavage site of ha, resulting in a virus that is considered attenuated for natural hosts and safe for people. the objective of this study was to safely generate vaccine candidates for a laiv using highly pathogenic avian influenza viruses by the replacement of six internal pr8 genes in the genome of candidates for inactivated vaccine subtype h5n1 (a ⁄ h5n1-pr8-rg) with internal genes of the laiv mdv by methods of classical reassortment. len17-mdv and a ⁄ h5n1-pr8-rg virus were co-infected in embryonated chicken eggs. five rounds of selective propagation were performed, three of which were at low temperature (25°c). the production and selection of reassortants were carried out in the presence of rabbit antiserum to len17-mdv. cloning by endpoint dilution was performed in each of the last three passages. a virus sample in an open petri dish was rocked gently for 20 sec while being irradiated with a ge 15 watt germicidal lamp at a distance of 20 cm from the dish. the residual infection titer was measured by titration in embryonated chicken eggs. genome composition of reassortant viruses was monitored by rflp analysis. 3 in addition, capacity of reassortant viruses to grow at optimum, low, and elevated temperatures (ca ⁄ ts phenotype) for influenza viruses was determined by virus titration in chicken eggs. reassortment of the mdv with the vn-pr or indo-pr viruses either resulted in reassortants that contained six internal genes from len17-mdv. however, all generated clones contained the na from the mdv. of ten such 7:1 reassortants based on vn-pr three reassortants had the pa gene from pr8 and one had ns gene from pr8. 6:2 reassortants from the targeted h5n1 composition were not generated. after repeated attempts, 7:1 temperature sensitive and cold adapted reassortants based on vn-pr and indo-pr viruses were obtained, but again, none had inherited the avian n1 neuraminidase (table 1) . in contrast, nibrg-23 didn't reassort with the mdv at all. twelve unsucsessful attempts to develop 6:2 or 7:1 reassortants of nibrg-23 with mdv showed that the classical reassortment procedure (cloning by limited dilutions in the presence of anti-mdv serum, followed by co-infection of equal doses of two parental viruses in eggs and two selective passages at 25°c) did not work for this virus pair. to disharmonize the incredibly strong gene constellation of nibrg-23, various modifications of the co-infection step were studied, such as: altering the nibrg-23 to mdv ratio (from 1:1 to 1: tions of anti-mdv serum alone or together with anti-pr8 serum. it was noted that even if the h5n1 to mdv ratio was 1:10 6 , the clones obtained were presumably parental h5n1 viruses without the transfer of any mdv-genes into genome of nibrg-23. in all, 234 clones were isolated, and 209 of them were identical to nibrg-23 parental virus. in nine clones, only the pa gene from mdv was included, whereas in three clones only the 'cold' ns gene was included (data not shown). using uv inactivation of nibrg-23 prior co-infection was more encouraging. after the first round of co-infection of partially uv-inactivated nibrg-23 with mdv (at ratio 1:10 2 ), reassortants that inherited several internal genes of mdv were obtained in the context of the nibrg-23 background (b3, c2, c4, d1) ( table 2 ). some of them (c2, c4, d1) were chosen for the next round of co-infection. after the second round of co-infection, c2, c4, and d1 'intermediate' reassortants with mdv (at ratio 1:1 or 1:10) 7:1 vaccine reassortants finally were obtained. live attenuated influenza vaccine is considered as one of the most promising pandemic vaccines. according to the who there is evidence that laiv might be more effective than inactivated vaccines. 4 this study attempted the safe development of laiv for potential pandemic highly pathogenic avian a ⁄ h5n1 viruses on the base of rg-reassortants for inactivated vaccine with modified h5 hemagglutinin and mdv for laiv. replacement of pr8based internal genes into genome of vn-pr and indo-pr reassortants with appropriate genes of mdv was realized by the classical reassortment procedure. difficulties were encountered in obtaining 6:2 reassortants that contained both the ha and na from the wild type avian h5n1 parental virus. in attempts to reassort the nibrg-23 with mdv, the classical reassortment procedure was unsuccessful. the challenge faced was to break an incredibly strong gene constellation of the nibrg-23 virus. partial uv-inactivation of nibrg-23 was encouraged in replacement of some pr8 internal genes with mdv genes 5 in some cases avian-human reassortant viruses with gull h13n6 and human influenza h1n1 genes were difficult to generate, and reassortants with the desired genotype of six gull virus genes with human influenza a h1 and n1 genes were not isolated despite repeated attempts. the gull pb2, np, and ns genes were not present in any of the gull-human h1n1 reassortants generated. 6 it is difficult to fully understand potential reasons for observed difficulties to reassort some avian viruses with human strains. unsuccessful attempts to develop 6:2 vaccine reassortants may be caused by an observed strong connection of pb2 and na genes in the genome of a ⁄ h5n1-pr8-rg viruses. in our attempts, each reassortant that possessed avian n1 neuraminidase inheritied pb2 gene of pr8 as well. and vice versa, the 'cold' pb2 gene always appeared to be coupled with the n2 neuraminidase of the mdv. in some cases, step by step co-infection of triple reassortants (h5n1-h1n1-h2n2) with h2n2 mdv may be the only possibility to generate a cold-adapted vaccine reassortant. our studies demonstrate unique and significant challenges that are faced in the development of influenza vaccines for avian influenza viruses with pandemic potential. such challenges must be further studied to identify methodologies to allow for rapid development and response to emerging viruses in a crisis. it is imperative that these studies be continued and expanded to identify either mechanisms of such tight gene constellations in influenza viruses produced by rg-derived vaccine strains or inability some genes of human h2n2 and avian h5n1 viruses to cross. in addition, further studies to improve the efficiency of classical reassortment processes will be conducted. during the period from 1997 to 2009, avian influenza outbreaks among humans have been registered in 15 countries of asia, europe, and africa. morbidity and mortality of humans followed the global spread of avian influenza h5n1 among wild and domestic birds, which caused great economic loss to the poultry industry in many regions including some highly developed countries. the global threat from avian influenza forced scientists to develop technologies for the production of a ⁄ h5n1 human vaccine. the development of ai a ⁄ h5n1 vaccines using strains isolated in kazakhstan and the organization of local production and creation of strategic stockpiles of effective vaccines is the an important issue for public health protection in the republic of kazakhstan. to address this, a scientific program 'influenza a ⁄ h5n1 vaccine development for public health protection in kazakhstan' was approved and financed from 2008 to 2010. in this article we give basic results of the development of a recombinant ai a ⁄ h5n1 inactivated whole virion vaccine with aluminium hydroxide as adjuvant for public health protection in kazakhstan. [1] [2] [3] the development of vaccine technology was conducted with the use of a ⁄ astanarg ⁄ 6:2 ⁄ 2009(a ⁄ h5n1) recombinant strain made of a ⁄ chicken ⁄ astana ⁄ 6 ⁄ 05(h5n1) and a ⁄ pr ⁄ 8 ⁄ 34(h1n1) strains by the reverse genetics. inactivation of virus containing allantoic fluid was carried out with the use of formalin in different concentrations. complete-ness of the virus inactivation was tested by 3-fold virus passaging in embryos. 4, 5 purification and concentration of the inactivated viruscontaining allantoic fluid was conducted with the use of ultra filtration in tangential flow, which was followed by gel filtration. then we evaluated the content of total protein, hemagglutinin, and ovalbumin in purified and concentrated material. 6 vaccine was composed of clarified and inactivated virus concentrate with the known ha dose containment, and 0ae4% aluminum hydroxide was added in 1:1 proportions. composition components and quality control of finished vaccine was determined in the stages of semi-finished product and finished biopreparation. determination of quantitative ovalbumin content was conducted by elisa applying a strip test-system chicken egg ovalbumin elisa kit cat. n 6050 (alpha diagnostic international, usa). vaccine immunogenicity was evaluated by hai micro test in u-bottom 96-well plates produced by 'costar' (usa). 7 vaccine apyrogenicity was evaluated after intravenous injection of the studied preparation to rabbits. 8, 9 for confirmation of the results vaccine series were tested for bacterial endotoxins with the use of limulus amebocyte lysate produced by charles river laboratories, inc. usa. 8 the vaccine toxicity was evaluated in white mice with body weight 18-20 gm and in rats with body weight 180-210 g both males and females according to glp principles. 9 allergenic characteristics of the inactivated vaccine was determined in white outbred mice and guinea-pigs both males and females according to 'methodic guideline for evaluation of allergenic characteristics of pharmacological substances'. 10 in the first series of experiments, we conducted work for obtaining influenza a(h5n1) recombinant strain. bidirectional expression plasmid phw_b754 with full-length sequences of ha and na gene segments of the strain a ⁄ chicken ⁄ astana ⁄ 6 ⁄ 05 (h5n1) isolated in kazakhstan were synthesized in geneart ag, (regensburg, germany). ha gene was modified by deleting the region encoding multiple basic amino acid rrrk motif in ha cleavage site. moreover, to prevent recovery of repeating basic amino acids motif due to polymerase slide, we inserted replacements g fi t and k fi t. thus the ha cleavage site consists of the following sequence ntpqgerrrkkrglfgai ntpqtetrglfgai. the basic amino acid motif of highly pathogenic strain a ⁄ chicken ⁄ astana ⁄ 6 ⁄ 05 (h5n1) was replaced by the sequence tetr ⁄ glf, which is characteristic of low pathogenic strains of influenza h5n1. sequence of gene coding na in the strain a ⁄ chicken ⁄ astana ⁄ 6 ⁄ 05 (h5n1) was cloned without modifications. the other segments pb1, pb2, pa, np, m and ns were obtained from influenza virus ivr-116 and synthesized and cloned in two-forked expression plasmid phw_b754 in geneart ag company, germany. the origin of genetic segments of vaccine strain a ⁄ astanarg ⁄ 6:2 ⁄ 2009 (h5n1) is presented in table 1 . vero cell culture (134 passage) (who) was received from european cell culture collection (salisbury, wiltshire sp4 0jg, great britain). the cell culture was grown in dmem ⁄ f12 medium with the addition of 10% of fetal bovine serum and 2 mm l-glutamine. to obtain reassortant virus a ⁄ astana ⁄ 6 ⁄ 05r-6:2, vero cells were infected with correlative plasmids by way of electroporation using nucleofector ii (amaxa) equipment. infected cells were placed in 6-well plates. after 6 hour, dmem ⁄ f12 medium was changed into 4 ml of opti-pro sfm (gibco) medium adding 2 mm l-glutamine and 1lg ⁄ ml trypsin. two days after cytopathic effect appearance supernatant was collected and used for infection of spf-eggs. the virus a ⁄ astanarg ⁄ 6 ⁄ 05-6:2 was grown in chicken embryos, and then virus titer was determined in chicken embryos and madine-darby canine kidney (mdck) cell culture. the titer of two final a ⁄ astanarg6 ⁄ 05-6:2 virus stocks was 9ae1 log 10 eid 50 ⁄ ml (chicken embryos); 1ae8 log 10 tcid 50 ⁄ ml (mdck cells); ha titer 1:512. a ⁄ chicken ⁄ astana ⁄ 6 ⁄ 05 (h5n1) virus contains motif of repeating basic amino acids in ha cleavage site. it is known that this sequence is the main determinant of ai virus pathogenicity. that is why this site was deleted in vaccine candidate strain. sequence results confirmed that influenza virus a ⁄ astanarg ⁄ 6 ⁄ 05r-6:2 strain ha gene sequence contains modified ha cleavage site and keeps mutations inserted for prevention of return to virus wild type. to confirm stability of modified ha gene sequence, five additional passages of recombinant strain a ⁄ astana rg ⁄ 6 ⁄ 05-6:2 were conducted in chicken embryos. sequencing and following phylogenetic analysis of the recombinant strain a ⁄ astana rg ⁄ 6 ⁄ 05-6:2 ha gene sequence proved the presence of modification in ha cleavage site. deletion of pathogenicity site of the obtained virus was confirmed by lethality test for chicken embryos, intravenous pathogenicity test in chicken, and in plaque-forming test with trypsin. pathogenicity test in chicken embryos showed that recombinant strain a ⁄ astanarg6 ⁄ 05-6:2 is capable of growing up to high titers without causing embryos' death. a ⁄ astanarg6 ⁄ 05-6:2 strain pathogenicity evaluation was conducted in 5-6 week-age white leghorns chicken, and this study proved that the strain a ⁄ astanarg6 ⁄ 05-6:2 (h5n1) is not virus pathogenicity inductor in chickens, which got intravenous injections of this virus (pathogenicity index is equal to 0). h5n1 strain ha cleavage site modification provides its cleavage capability only with tripsin-like proteases, which shows low level of pathogenicity. aiming at confirmation of ha cleavage site modification, we experimentally studied virus replication ability both with trypsin and without this enzyme. and we got the following results. in the plaque-forming test, a ⁄ astanarg6 ⁄ 05-6:2 strain produced plaques in mdck cells only with trypsin, proving the trypsin-dependent phenotype characteristic of low pathogenic avian influenza viruses. to prove the ha subtype antigenic analyses of a ⁄ astana ⁄ 6 ⁄ 05r-6:2 strain was conducted by means of serological methods in hemagglutinin inghibition test with the use of postinfection antisera of rabbits and rats (influenza research institute swd rams), standard serum received from cdc, atlanta, usa. hai test proved that a ⁄ astana ⁄ 6 ⁄ 05r-6:2 strain belongs to h5 subtype. furthermore, toxicity of vaccine candidate strain was evaluated by way of subcutaneous injection of viral material to balb mice. the strain appeared to be non-toxic for white mice getting subcutaneous injection of 0ae5 ml of the preparation. the conducted research showed that according to all tested characteristics, a ⁄ astana ⁄ 6 ⁄ 05r-6:2 strain can be used for influenza a ⁄ h5n1 inactivated vaccine production. according to its genetic characteristics, this strain belongs to the group of vaccine strains recommended by who for the development of influenza pre-pandemic inactivated vaccines. we determined basic cultivation parameters of the recombinant strain a ⁄ astanarg6 ⁄ 05-6:2 in 10-11 day chicken embryos. the determined parameters are the following: infection dose, 1000-10 000 eid 50 ; cultivation period, 72 hour; incubation temperature, 33°c. these cultivation parameters allow obtaining virus containing material with biological eid and hemagglutinating activity of 8ae5-9ae0 log 10 eid 50 ⁄ cm 3 and 1:512 ha titre and even higher. in the next series of experiments, we conducted research on the determination of optimal sequence of technological stages of virus clarification, concentration, and inactivation in the order of vaccine production. samples of viral material were subjected to inactivation before and after clarification and concentration. the regimen of virus inactivation by formaldehyde with final concentration of 0ae05%, period of inactivation of 3 days, temperature of inactivation medium of 4-6°c, ph of inactivation medium of 7-7ae5. on the basis of the conducted experiments we determined that the selected regimen of inactivation provides complete and irreversible inactivation of viral suspensions of the hpai strain irrespective of the kind of inactivated material. we did not observe reduction of ha activity in non-clarified viral suspensions. however, when we inactivated clarified and concentrated material, ha activity reduced by an order of magnitude. comparison of forms and sizes of virion structural elements in native (non-clarified) and formalin inactivated preparations did not reveal any significant differences. concentration of virus particles in the studied preparations was similar. the selected inactivation regimen provides obtaining completely avirulent viral suspension of the strain a ⁄ astanarg6 ⁄ 05-6:2, and it does not influence the structure of the virus. on the basis of the experiments results, we selected method of viral allantoic fluid inactivation without preliminary clarification. during further research, we tried to get highly clarified viral concentrate. this study resulted in the combined scheme, which includes clarification of inactivated viral allantoic fluid by low speed centrifugation at 4000 circulations per min for 30 minutes, filtration through membrane filters with pore diameter of 0ae45 lm, ultrafilatration ⁄ diafiltration, gel filtration in 6b sepharose, and sterilization of viral suspension through membrane filters with pore diameter of 0ae22 lm. the experiments resulted in the development of production technology of embryonic inactivated vaccine based on recombinant strain a ⁄ astanarg ⁄ 6:2 ⁄ 2009 (h5n1) contain-ing aluminium hydroxide as adjuvant. the developed influenza a ⁄ h5n1 human vaccine has the trade name kazfluvacò. its composition components are presented in table 2 . preclinical testing of the vaccine kazfluvacò was conducted according to the following parameters: general health condition of animals, change of body weight and temperature of immunised animals (for ferrets), presence of post vaccination antibodies response in sera, forming protective immune response against reassortant viruses of h5 subtype, study of acute and chronic toxicity of three experimental vaccine series in different doses and semi-finished vaccine product applying different ways of injection, study of allergic and immunotoxic characteristics of the vaccine, as well as study of pyrogenic reaction and analysis for bacterial endotoxins presence. [11] [12] [13] [14] preclinical tests of kazfluvacò vaccine safety showed that this vaccine does not have toxic effect on organisms of warm-blooded laboratory animals. double intramuscular injection of kazfluvacò vaccine in inoculative dose does not effect appearance, general health condition, behaviour of animals, their muscular strength and physical activity, does not have negative effect on biochemical parameters of blood and basic physical functions of animals organism, and does not cause pathomorphological changes. this shows the safety of the vaccine. local irritation action was not observed. the results of the vaccine allergic action study showed that the vaccine does not have allergic effect at the intravenous injection. the research also showed that the vaccine does not have negative effect on immune system of laboratory animals. research conducted on mice and ferrets showed high immunogenic activity of the vaccine at one-and two-dose regimen of injection. the research showed 100% of protective effect of kazfluvacò vaccine at two-dose injection regimen in ferrets infected by homological strain of influenza virus. the devised inactivated influenza a ⁄ h5n1 vaccine kaz-fluvacò is a safe and immunogenic biopreparation that is not worse than the overseas analogues in its immunobiological characteristics. [15] [16] [17] [18] to date the whole-virion inactivated influenza a ⁄ h5n1 vaccines of the producers such as omnivest (hungary), biken, denka seiken, kitasato institute, kaketsuken (japan), gsk biologicals (belgium), sinovac biotech (china) are registered. all of them are produced on the basis of chicken embryos and aluminum is used as an adjuvant. kazfluvacò differs from its analogues in the flowchart of the virus purification and concentration that makes possible to produce a safer preparation. 19, 20 the results of the conducted research and preclinical testing allow starting work towards implementation of phase i preclinical tests on volunteers. it is planned to conduct a randomized blind placebo-controlled phase i study on double application of kazfluvacò vaccine in increasing doses. the preparation will be administered to volunteers aged 18-60 years for assessment of its safety and immunogenicity in doses of 7ae5 and 15ae0 lg of ha. when the world health organization (who) announced the sixth phase of a ⁄ h1n1v influenza pandemic, scientists all over the world started investigation to develop technology for production of prophylactic means against the disease. having taken into consideration the threat of a pandemic for kazakhstan, the ministry of education and science of the republic of kazakhstan launched the program ''monitoring, study, and development of diagnostic, prophylactic, and therapeutic means for influenza a ⁄ h1n1.'' this paper presents the experimental data obtained at the ribsp in the course of the studies towards the development of technology for production of an inactivated a ⁄ h1n1 influenza vaccine, as well as the results of pre-clinical testing of the developed vaccine. the development of vaccine production technology was conducted with the use of who recommended vaccine strain nibrg-121xp constructed by the method of reverse genetics in the national institute for biological standards and control (nibsc, great britain). the virus was inactivated with formalin at different final concentrations, and the extent of inactivation was evaluated via threefold virus passages in developing chicken embryos. 1 the inactivated virus was purified and concentrated by the method of ultrafiltration in tangential flow followed by gel filtration. the purified and concentrated material was evaluated judging on the total protein, hemagglutinin (ha), and ovalbumin. the vaccine was prepared by pooling the purified and concentrated virus material with the certain weight content of ha and the work solution of aluminum hydroxide (0ae4%) in the ratio 1:1. the ovalbumin content was quantified in elisa with the use of the strip test system chicken egg ovalbumin elisa kit (cat. no. 6050 alpha diagnostic international, san antonio, texas, usa). weight content of the virus ha was determined according to sominina, burtseva. 2 the content of the residual formaldehyde, aluminum (al +3 ) ions, and thiomersal in the vaccine was measured according to the operating instructions. 3 the vaccine immunogenicity was assessed in the hemagglutination inhibition test, which was carried out as a microassay in 96-welled u-bottomed plates (''costar'', new york, usa). 3, 4 apyrogenicity of the vaccine was assessed post intravenous administration of the tested preparation to rabbits. 5, 6 to confirm the obtained results the vaccine batches were tested for bacterial endotoxins with use of the limulus amebocyte lysate (charles river laboratories, inc., wilmington, ma, usa). 7 the toxicity of the vaccine was assayed in white mice weighing 18-20 g and in rats weighing 180-210 g (male and female) in compliance with the principles of good laboratory practice. 8 allergenic properties of the inactivated vaccine were determined according to the ''operating instructions on assessment of allergenic properties of pharmaceutical substances'' 9 in white outbred laboratory mice and guinea-pigs of both sexes. the first step in the course of developing technology for vaccine production was to determine the major conditions for influenza virus cultivation: usage of 10-days embryonated chicken eggs at the infectious dose within 1000-10 000 eid 50 , incubation temperature (34 ± 0ae5)°c, and duration of the incubation period 72 hours. the established parameters for virus cultivation made it possible to produce virus-containing materials of infectious activity within 8ae5-9ae0 log eid 50 ⁄ cm 3 and hemagglutinating activity 1:256 and higher. in the subsequent experiments, an optimal method for virus inactivation was selected. on the basis of the experimental findings, the following conditions for inactivation of the native virus-containing material were elected: formalin of 0ae05% final concentration as an inactivating agent; inactivation period of 72 hours at temperature (4 ± 2)°c. these conditions provide the complete inactivation of the virus (nibrg-121xp strain) material, did not impact distinctly the structural organization of the virus, and did not reduce the antigenic activity. as it is well known, virus purification and concentration means very much in the development of technology for production of an inactivated whole-virion influenza vaccine. the investigation into optimization of the technological step of purification and concentration of the recombinant influenza virus nibrg-121xp strain resulted in selection of an optimal pattern including such steps as clarification of the virus suspension by filtration through membranes with pore size 0ae45 lm, virus concentration by ultrafiltration in a tangential flow, dialysis filtration in a tangential flow, gel filtration on sepharose 6b, and sterilization of the viral suspension through membrane filters with pore size 0ae22 lm. the studies conducted by the ribsp specialists resulted in the development of technology for production of the first domestic whole-virion inactivated a ⁄ h1n1 influenza vaccine with aluminum hydroxide as adjuvant and with the brand name refluvac ò . the key processing characteristics of the whole-virion inactivated a ⁄ h1n1 influenza vaccine vaccine refluvac ò are shown in table 1 . simultaneous with the performance of all process operations, the parameters such as sterility, inactivation extent, ph, vaccine specificity, total protein content, weight content of has, aluminum and formalin contents, content of thiomersal, and ovalbumin, pyrogenicity of the vaccine and its immunogenicity for mice, were optimized. the key qualitative characteristics of the designed influenza a ⁄ h1n1 vaccine refluvac ò are shown in table 2 . before implementation of phase i clinical trials on volunteers, preclinical testing of three experimental batches of refluvac for immunogenic activity and safety was carried out. it was conducted in three laboratory bases of research institutions: the toxicology institute ⁄ federal medicobiological agency, russia (st petersburg), the research institute for biological safety problems (republic of kazakhstan), and the influenza research institute ⁄ north-western branch of the russian academy of medical sciences (st petersburg), with use of different animal models (mice, rats, chinchilla rabbits, guinea-pigs, ferrets). the results of the preclinical testing are as follows: • electron microscopy of the preparation has shown that the viral particles are well dispersed and do not aggregate. the portion of whole (intact) particles is over 95%, which is evidence of virion integrity; • assessment of polypeptide composition of the vaccine refluvac by electrophoresis in 10% polyacrylamide gel with sodium dodecyl sulfate has shown the vaccine to contain both surface antigens (ha, na) and highly purified inner virion proteins (np, m1) that are typespecific antigens, so the vaccine is a preparation of full immunological value; • judging on the parameters of acute and chronic toxicity for white mice and rats of both sexes, the vaccine is a non-toxic and safe preparation; • under conditions of a chronic experiment on white mice and rats, it was found that refluvac does not produce changes in behavior, somatic, or vegetative responses; • assay of hematological and biochemical blood characteristics of white mice and rats following vaccine administration did not reveal any significant differences as compared to the animals of the control group; • refluvac does not cause allergenic and immunotoxic impact; • the vaccine refluvac does not cause local irritative effect; • refluvac is apyrogenic for laboratory animals; • the pathomorphological and hystopathological analysis did not reveal any changes due to immunization in animal organs; • testing of immunogenic characteristics of the vaccine on mice and ferrets has shown formation of hemagglutinating antibodies in animals after single administration; • refluvac induces 100% protection in immunized ferrets at their challenge with the wild-type influenza virus a ⁄ california ⁄ 07 ⁄ 2009 (h1n1v). the results of the performed preclinical testing have allowed concluding that refluvac, an inactivated whole-virion vaccine with aluminum hydroxide as adjuvant, is a safe and highly effective preparation against influenza a ⁄ h1n1v. the implemented study resulted in development of technology for production of the first domestic inactivated allantoic whole-virion influenza a ⁄ h1n1 vaccine with aluminum hydroxide as an adjuvant under the brand name refluvac ò based on the recombinant strain nibrg-121xp. the devised pandemic vaccine meets who requirements as well as requirements concerning safety and immunogenicity of the national pharmacopeias of the republic of kazakhstan and russian federation. [9] [10] [11] [12] the devised technology for vaccine production differs from the previous technologies for production of allantoic whole-virion influenza a ⁄ h1n1 vaccines in its processdependent parameters. presence of an adjuvant (aluminum hydroxide) increases significantly the vaccine immunogenicity and allows maximal reduction of the dose of the administered antigen that, in turn, results in diminished reactogenicity of the vaccine. aluminum hydroxide is an adjuvant that is most frequently used in clinical practice. 13 to date the results of the double-centered randomized study of the europe-licensed vaccine fluval p [monovalent inactivated whole-virion influenza vaccine with aluminum phosphate based on strain a ⁄ california ⁄ 07 ⁄ 2009 (h1n1) nymc x-179a (omninvest, pilisborosjeno, hungary)] that is similar to the refluvac preparation are published. the data of this research are an evidence of safety and high immunological effectiveness of the vaccine in dose 6 lg ha at single administration both in adults and elderly persons. 14 the results of the pre-clinical tests allow recommending carrying out phase 1 clinical testing of the refluvac ò vaccine for safety and immunogenicity. single immunization of volunteers with refluvac ò in doses 3ae75, 7ae50, and 15ae00 lg of ha are planned. mid 50, respectively. the study results confirm that new h1n1 laiv and h7n3 laiv candidates are safe and immunogenic and confer protection from homologues influenza virus infection in mice. the recent emergence of a new pandemic h1n1 virus and the threat of transmission of avian viruses to humans had stimulated research and development of live attenuated cold-adapted influenza vaccines against newly appeared influenza viruses. formulations of live attenuated influenza a vaccine (laiv) against pandemic influenza strains, including h1n1, h5n1, h9n2, and h7n3 are currently being tested in preclinical and phase i clinical studies. 1 the following paper describes the preclinical study of new h1n1 and h7n3 laiv candidates in mice. the study addressed the following three objectives: (i) to demonstrate that cold-adapted (ca) reassortant influenza a(h1n1) and a(h7n3) vaccine candidates are indistinguishable from the parental a ⁄ leningrad ⁄ 134 ⁄ 17 ⁄ 57 (h2n2) master donor strain (mds) virus with regard to replication efficiency in upper and lower respiratory tract of mice; (ii) to demonstrate the immunogenicity of different doses of cold-adapted (ca) reassortant influenza a(h1n1) and a(h7n3) vaccine candidates in mice; and (iii) to demonstrate the protective efficacy of cold-adapted (ca) reassortant influenza a(h5n1) and a(h7n3) vaccine candidates in mice against a homologous wild-type virus challenge. the a ⁄ 17 ⁄ mallard ⁄ netherlands ⁄ 00 ⁄ 95 (h7n3) reassortant containing the ha and na genes from a ⁄ mallard ⁄ netherlands ⁄ 00 (h7n3) and six other genes from mds, the a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) reassortant containing the ha and na genes from a ⁄ california ⁄ 7 ⁄ 2009 (h1n1) and six other genes from a ⁄ leningrad ⁄ 134 ⁄ 17 ⁄ 57 (h2n2) were generated by classical genetic reassortment in embryonated chicken eggs (ec). viruses were propagated in 10days old eggs (34°c, 48 hours). fifty percent egg infectious dose (eid 50 ) titers were determined by serial titration of viruses in eggs. titers were calculated by the method of reed and muench. 2 female balb ⁄ c mice, 6-8 weeks of age were used in all experiments. mice were lightly anesthetized with ether and then inoculated intranasally (i.n.) with 50 ll of infectious virus diluted in phosphate-buffered saline (pbs). mice were inoculated with 100 mid 50 (50% mouse infectious dose) of a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1), a ⁄ 17 ⁄ mallard ⁄ netherlands ⁄ 00 ⁄ 95 (h7n3), and a ⁄ leningrad ⁄ 134 ⁄ 17 ⁄ 57 (h2n2) mds. viral loads were measured in respiratory and brain tissues collected at 3 and 6 days post-infection (dpi). tissue homogenates prepared using a disruptor and clarified supernatants were titrated on eggs at permissive temperature to determine infectious concentrations. groups of animals were inoculated with 1000 mid 50 or 100 mid 50 of either h1n1 laiv or h7n3 laiv intranasally after collecting a pre-immunization blood sample. a second blood sample was collected at 28 dpi. on the same day, the animals received a second intranasal inoculation with the same virus that was used for priming at 0 dpi. to assess protection, all animals were infected 42 dpi with either 100 mid 50 of a ⁄ california ⁄ 7 ⁄ 2009 (h1n1) or 100 mid 50 a ⁄ mallard ⁄ netherlands ⁄ 00 (h7n3) virus by the intranasal route. four animals from each group were euthanized at 45 dpi, and the respiratory and systemic organs were harvested for virus titration. a forth blood sample was collected at 56 dpi from the remaining animals. hi antibody titers were determined for individual serum samples collected on days 0, 28, 42, and 56. body weights were taken daily following challenge through day 14 postchallenge. sera were tested for hi against homologous h1n1 and h7n3 viruses. the h1n1 laiv, h7n3 laiv and h2n2 mds influenza viruses replicate in mice lungs at level 2ae1-2ae3 lgeid 50 ⁄ ml at 3 dpi (figure 1 ). at 6 dpi, replication of the viruses in the lungs decreased to 1ae6-2ae0 lgeid 50 ⁄ ml (data not shown). in contrast, the wild-type virus a ⁄ mallard ⁄ netherlands ⁄ 00 (h7n3) demonstrated high level replication in lungs -6ae4 lgeid 50 ⁄ ml. the levels of replication of studied viruses in nasal turbinates were 2ae5-3ae7 lg eid 50 ⁄ ml at 3 dpi (figure 1) , and 2ae0-2ae2 lgeid 50 ⁄ ml at 6 dpi (data not shown). there were no significant differences between the viruses in regard to replication in upper respiratory tract of mice. thus, it was shown that a ⁄ 17 ⁄ mallard ⁄ netherlands ⁄ 00 ⁄ 95 (h7n3) and a ⁄ 17 ⁄ california ⁄ 2009 ⁄ 38 (h1n1) vaccine candidates was indistinguishable from parental a ⁄ leningrad ⁄ 134 ⁄ 17 ⁄ 57 (h2n2) in terms of replication in the lungs and noses of mice at 3 and 6 dpi. no virus was found in the brain tissue of immunized mice at 3 and 6 dpi (in undiluted samples tested). thus, it was shown that a ⁄ 17 ⁄ mallard ⁄ netherlands ⁄ 00 ⁄ 95 (h7n3), a ⁄ 17 ⁄ cali-fornia ⁄ 2009 ⁄ 38 (h1n1) vaccine candidates are identical to a ⁄ leningrad ⁄ 134 ⁄ 17 ⁄ 57 (h2n2) in lacking neuroivasive capacity, and all three viruses similarly fail to replicate in the brain. it was shown that all immunized animals survived after challenge with wild-type a ⁄ mallard ⁄ netherlands ⁄ 12 ⁄ 00 (h7n3) virus. the mice in vaccine groups showed no signs of morbidity. average weight changes were tracked from day 2 to day 6 in all study groups, but the changes did not exceed 5%. as shown in figure 2 , the challenge virus actively replicated in respiratory tissue taken from mock immunized animals (5ae7 lgeid 50 in the lung and 4ae2 lgeid 50 in the nose), but failed to infect the brain and spleen. on the other hand, in both h7n3 laiv vaccinated groups, all tested organs were free from presence of challenge virus. thus, immunization of mice with either 1000 mid 50 or 100 mid 50 h7n3 laiv protected the animals from the subsequent challenge infection with a homologous with wild-type h7n3 virus. both h1n1 and h7n3 laiv candidates were found to be immunogenic. after one dose of 100 mid 50 of h1n1 laiv, gmt of hi antibodies were 17ae6. one dose of 1000 mid 50 or 100 mid 50 h7n3 laiv elicited hi antibody level with gmt of 8ae7 and 6ae6, respectively. the second dose of h7n3 laiv further stimulated serum hi antibody levels to gmt 40ae0 and 23ae3, for 1000 mid 50 or 100 mid 50, respectively (data not shown). the mouse model is widely used to better understand the pathogenicity of avian influenza viruses for mammalian species, to be able to predict the pandemic potential of such viruses, and to develop improved methods for the prevention and control of the virus in a potential pandemic. 3 a subset of the h7 viruses was evaluated for the ability to replicate and cause disease in balb ⁄ c mice following intranasal administration. h7 subtype viruses were able to infect mice without adaptation and manifested different levels of lethality and kinetics of replication. 4 there is limited preclinical information available for laiv. thus, live monovalent vaccine against pandemic influenza virus h1n1 (influvir) was tested for acute toxicity and its effect on the systems and organs of laboratory animals. according to toxicology and necroscopy results, the live monovalent influenza vaccine influvir, when applied intranasally, was safe and was well tolerated. 5 in our current study we demonstrate that a(h1n1) and a(h7n3) laiv are indistinguishable from the parental mds virus with regards to replication kinetics in the upper and lower respiratory tract of mice. both h1n1 and h7n3 laiv candidates were immunogenic and protect mice against subsequent a challenge with the wild-type virus. live attenuated cold-adapted (ca) influenza vaccines are an effective means for the control of influenza, most likely due to their ability to induce both humoral and cellular immune responses. in our study we confirm that new h1n1 laiv and h7n3 laiv candidates are safe, immunogenic, and confer protection from influenza infection in mice. health organization (who) declared a pandemic by raising the worldwide pandemic alert level to phase 6. therefore, h1n1 inactivated monovalent vaccine formulated with our proprietary oil-in-water emulsion based adjuvant was evaluated in ferrets for its potential to induce with low antigen dose efficient, robust, and rapid protective immunity against a wild type challenge virus (a ⁄ netherlands ⁄ 602 ⁄ 2009). this adjuvant was also tested in ferrets in a h5n1 avian influenza model for its ability to induce a cross-clade immunity and cross-protection. two independent studies (a&b) were carried out with male and female outbred ferrets (musleta putorius furo) in compliance with ''guide for the care and use of laboratory animals,'' ilar recommendations and aaalac standards. ferrets used in both studies were influenza seronegative by anti-nucleoprotein elisa and by hi assay against the pandemic and seasonal strains. in study a, four groups of seven ferrets aged approximately of 6 months received one or two im vaccinations 3 weeks apart of either af03-adjuvanted (3ae8 lg of ha with af03) or unadjuvanted ( body weight loss was monitored as an indicator of disease and a mean body weight loss of 20% was recorded in the control group at day of necropsy. body weight loss was reduced to £10% and £8% in animals that had received 1 and 2 doses of either unadjuvanted or af03-adjuvanted vaccine, respectively. viral lung titration showed high levels of virus replication ( ‡4ae7 tcid50 ⁄ g tissue) in the lungs of all control ferrets 4 days after challenge. one or two administrations of unadjuvanted vaccine reduced lung viral load by 2 and 3 log 10 , respectively. interestingly, ferrets that received either one or two doses of af03-adjuvanted h1n1 2009 vaccine, showed significantly greater reduction of lung viral loads (>4 log 10 ). no virus was detected in the lungs of 6 ⁄ 7 (86%) animals immunized with a single injection of the af03-adjuvanted vaccine and in 100% of ferrets vaccinated twice. assessment of viral shedding from the upper respiratory tract showed that the af03-adjuvanted a ⁄ h1n1 monovalent vaccine was able to reduce the viral load in the nose and in the throat by 3ae8 and 3ae1 log 10 , respectively, as compared to the control group. conversely, viral loads were only slightly reduced in the nose and mostly unchanged in the throat in ferrets immunized with either one or two doses of unadjuvanted a ⁄ h1n1 monovalent vaccine. gross pathology and histology examinations revealed lung lesions consistent with influenza a ⁄ h1n1 virus infechowever, a second dose of af03-adjuvanted vaccine strongly increased hi and mn titers, which persisted for 3 months (table 2 ). antibody responses cross-reactive to heterologous clade 2.2 strain were elicited ferrets vaccinated with the af03-adjuvanted clade 2.1 vaccine. hi antibody titers ‡40 crossreactive to clade 2.2 and persistent up to d110 were observed in vaccinated animals. an inter-clade low crossreactive hi response to a clade 1 strain was only detected in a few ferrets that had been vaccinated with the af03-adjuvanted clade 2.1. all af03-adjuvanted clade 2.1 antigen vaccinated animals survived challenge either with the homologous or heterologous virus until euthanized day 5. after challenge, mean body temperature and mean body weights were monitored as indicators of disease. in the control ferrets, mean body temperature increased by 2-3°c (depending on the challenge virus strain) 24 h post challenge, with an accompanying mean body weight loss ranging from 15ae4% to 20ae7%. ferrets vaccinated with the af03-adjuvanted clade 2.1 vaccine showed a lower and delayed fever compared to control ferrets that received the same viral challenge, whereas no significant differences were observed between vaccinated animals and their respective controls upon challenge with clade 2.2 or clade 0 viruses. body weight loss was reduced in all vaccinated animals when compared to controls after challenge with either the homologous clade 2.1 strain or with one of the heterologous strains. lung virus titration showed high levels of virus replication in all control animals 5 days after homologous challenge with the clade 2.1 virus. lung viral loads of all ferrets immunized with the af03-adjuvanted clade 2.1 vaccine were reduced more than 4 log 10 . vaccination resulted in complete viral clearance from the lungs of 80% of animals assessed 5 days after challenge. as compared to controls, a reduction of the mean viral load of about 2 log 10 was observed in ferrets vaccinated with the af03-adjuvanted clade 2.1 vaccine after heterologous challenge with either the clade 0 or clade 1 virus. conversely, vaccination with af03-adjuvanted clade 2.1 vaccine did not result in reduction of lung viral loads after challenge with the clade 2.2 heterologous virus strain. titration of pharyngeal swabs showed high levels of viral shedding in all control ferrets after challenge with clade 2.1 strain, whereas virus was not detected in any vaccinated animal. similarly, 5 log 10 reduction of viral shedding was seen in vaccinated versus control ferrets following clade 1 heterologous challenge. lower reductions in viral shedding were observed after clade 2.2 challenge (2ae6 log 10 ) and clade 0 challenge (1ae4 log 10 ). gross pathology and histology revealed lung lesions consistent with influenza a ⁄ h5n1 virus infection all control animals challenged with the clade 2.1, clade 2.2 or clade 0 strains. mild to moderate lung lesions were observed in control animals following challenge with clade 1 virus. macroscopic evaluation (percentage of affected lung parenchyma) and histopathological analysis (extent and severity of alveolitis, alveolar oedema and hemorrhage) showed that lung lesions were significantly reduced in af03-adjuvanted clade 2.1 vaccinated animals after challenge with the homologous clade 2.1 virus strain as compared to controls. similarly, a reduction of the macroscopic and microscopic lung lesions was observed in vaccinated animals upon heterologous challenge with clade 2.2 and clade 0 virus strains, whereas no differences were observed between control and vaccinated animals after challenge with clade 1 virus. the results of these ferret challenge studies demonstrated that low doses of pandemic influenza vaccines formulated with an oil-in-water emulsion adjuvant, af03, elicited strong antibody responses specific to the immunizing strain. importantly, these vaccines provided protection after homologous challenge with complete virus clearance in ferret lungs and reduced viral shedding from the upper respiratory tract suggesting an ability to reduce virus transmission. moreover, af03-adjuvanted h5n1 vaccine can provide cross-protection upon challenge with different h5n1clades by preventing mortality and reducing the viral burden in the lower and the upper respiratory tract. in conclusion, the results of these studies highlighted the ability of af03-adjuvanted influenza vaccines to induce potent immune responses and full protection in ferrets against homologous challenge and suggested that protection may be mediated, at least in part, by antigenspecific humoral immunity. since 1993, outbreaks of h9n2 influenza virus infection in poultry have occurred in eurasian countries. phylogenetic and antigenic analysis of h9n2 isolates revealed that there are three sublineages, consisting of g1, g9, and korean, among ha genes of the eurasian h9n2 viruses. h9n2 viruses do not cause severe disease in poultry, but co-infection of h9n2 viruses with bacteria such as staphylococcus aureus, haemophilus paragallinarum, or attenuated coronavirus vaccine may exacerbate the disease. 1,2 h9n2 viruses were isolated from domestic pigs in china and korea and from humans with febrile respiratory illness in hong kong in 1998 kong in , 1999 kong in , and 2003 it is, thus, postulated that in the present study, h9 virus strains were analyzed antigenically and phylogenetically to select a proper h9n2 vaccine strain. inactivated whole virus particle vaccine was prepared, and its potency against h9 virus challenge was assessed in mice. viral rnas were extracted from the allantoic fluid of chicken embryos infected with viruses by using a commercial kit (trizol ls reagent; invitrogen, california, usa) and reverse-transcribed with the uni12 primer 6 and m-mlv reverse transcriptase (invitrogen). the primers used for the ha gene amplification were h9-101f 7 and h9-1341r. for phylogenetic analysis, sequence data of the genes together with those from public database were analyzed by the neighbor-joining method. 8 h9 influenza viruses were analyzed by hemagglutinationinhibition (hi) test. 9 chicken hyperimmunized antisera against seven h9 viruses were prepared according to previous report. 10 virus replication and pathogenicity against embryonated chicken eggs viruses were inoculated into 10-day-old embryonated chicken eggs and incubated for 48 hours at 35°c. ha titers and 50% egg infectious dose (eid 50 ) were measured every 12 hours post-inoculation. pathogenicity of dk ⁄ hok ⁄ 49 ⁄ 98 against embryonated chicken eggs was evaluated by mean death time (mdt) as described previously. 11 dk ⁄ hok ⁄ 49 ⁄ 98 was injected into the allantoic cavities of 10-day-old embryonated chicken eggs and propagated at 35°c for 48 hours. the virus in the allantoic fluids (512ha) was purified by differential centrifugation and sedimentation through a sucrose gradient according to previous report. 12 the concentration of protein was measured by od using ultrospec 3100 pro (amersham biosciences, tokyo, japan). the purified virus was inactivated with 0ae1% formalin at 4°c for 7 days. immunization of mice and challenge of immunized mice with hk ⁄ 1073 ⁄ 99 four-week-old female balb ⁄ c mice were purchased from japan slc, inc. (shizuoka, japan). the mice were injected subcutaneously with 10, 2, 0ae4, or 0ae08 lg proteins of inactivated dk ⁄ hok ⁄ 49 ⁄ 98 whole virus vaccine. two weeks later, the mice were boosted by subcutaneous injection with the same dose of the vaccine. control mice were injected with pbs. serum samples were tested by enzyme-linked immunosorbent assay (elisa) according to previous report. 10 one week after the second vaccination, 10 mice in each group were challenged intranasally with 30 ll of 10 6ae5 eid 50 of hk ⁄ 1073 ⁄ 99 under anesthesia. on 3 days postinfection, five mice in each group were sacrificed, and the lungs were separately homogenized to make a 10% (w ⁄ v) suspension with minimal essential medium (nissui, tokyo, japan). the virus titers of the supernatants of lung tissue homogenates were calculated in 10-day-old embryonated chicken eggs and expressed as the eid 50 ⁄ gram of tissue. the other five mice in each group were monitored for body weight for 14 days after challenge. the ha genes of 22 h9 viruses were sequenced and analyzed by the neighbor-joining method. all of the 22 h9 viruses were classified into the eurasian lineage ( figure 1) . eleven, seven, and four strains were classified in the korean, g9, and g1 sublineages, respectively. the h9 viruses of the korean and g9 sublineages were isolated from waterfowl, poultry, pigs, and humans in the east asian countries, and those of the g1 sublineage were isolated from poultry in the west asian countries. the cross-reactivity between these antisera and h9n2 viruses were analyzed by hi test. the antisera against h9 viruses belonging to the korean sublineage were broadly cross-reacted to h9 viruses belonging to the g9 and g1 sublineages. h9 viruses belonging to the korean lineage were reacted to the antisera against h9 viruses belonging to the g9 and g1 sublineage compared with h9 viruses belonging to the other sublineage (data not shown). thus, it was suggested that h9 vaccine strain should be selected from the viruses of korean sublineage to prepare for the vaccine strain of h9 viruses. dk ⁄ hok ⁄ 49 ⁄ 98 replicated efficiently in 10-day-old embryonated chicken eggs (data not shown). pathogenicity of dk ⁄ hok ⁄ 49 ⁄ 98 against embryonated chicken eggs was determined by mdt. dk ⁄ hok ⁄ 49 ⁄ 98 was low pathogenic against embryonated chicken eggs (data not shown) and was selected as an h9 vaccine strain. to assess the potency of the vaccine against h9 virus infection, mice vaccinated subcutaneously with inactivated dk ⁄ hok ⁄ 49 ⁄ 98 were challenged intra-nasally with hk ⁄ 1073 ⁄ 99. immunogenicity of the inactivated vaccine was assessed by measuring the igg antibodies in mouse sera by elisa. antibody was detected in the group of mice injected 10 lg protein after the first immunization and detected in the group of mice injected 2 lg protein after the second immunization. thus, potency of the present inactivated whole virus vaccine was demonstrated in mice. next, to assess the protective immunity of the inactivated vaccine in mice, viral titers in the lungs was determined. the virus titers in the lungs were 10 1ae5 -10 3ae7 eid 50 ⁄ g in the groups of mice injected 10, and 2 lg protein, and 10 5ae3 -10 6ae0 eid 50 ⁄ g in the other vaccinated groups. body weight reduction of mice were observed in the group of mice injected 0ae4, 0ae08 lg protein, and control groups from 3 dpi, and reached to 10% body weight loss from 4-to 6-day post-infection ( figure 2 ). this result correlates with antibody titer in mouse sera and viral titers in the lungs. these results suggest that the test h9 inactivated whole vaccine confers prevent of weight loss and reduction of virus replication against h9 influenza virus infection in mice. recently, h9n2 viruses of all of three sublineage have been isolated from wild birds and poultry in worldwide. h9n2 viruses were isolated from pigs and humans in china 3 and korea, suggesting that h9n2 virus would be a potential for a pandemic influenza virus in human population. h9n2 viruses were isolated from pigs in china and korea and were classified into the g9 and korean sublineage. in human cases, all h9n2 virus isolated from humans in china was classified into the g1 sublineage. it was suggested that h9n2 viruses isolated from pigs and humans vary in antigenicity of isolates between the korean, g9, and g1 sublineages. therefore, it is important for the preparedness of influenza pandemic to develop h9 influenza virus vaccine, which could broadly cross-react to antisera of all sublineage viruses. so, we selected the vaccine candidate strain, dk ⁄ hok ⁄ 49 ⁄ 98, which could broadly cross-react to antisera of all sublineage viruses, and which could replicate in this study, it was suggested that the test vaccine has potency to protect against challenge with h9 virus using mice for mammalian model. the challenge virus, hk ⁄ 1073 ⁄ 99, was isolated from human, replicates efficiently in mice, and shows pathogenicity in mice. the test vaccine inhibited viral replication and body weight loss in mice. whole inactivated vaccine produced protective immunity, supporting our approach of using whole virus particles for vaccine development. furthermore, whole particle virus vaccine could induce igg and mucosal iga levels after intranasal vaccination with whole particle vaccine. 13 the present results may facilitate the studies of the vaccine for future pandemic caused by h9 influenza virus in humans. tants to attempt to improve growth. to determine whether wild type h1n1pdm grew better in the novartis mdck suspension cell line (mdck33016pf) than in eggs, isolations from h1n1pdm positive clinical samples were attempted in both substrates. the isolation rate of h1n1pdm viruses was higher in mdck33016pf cells (89%) (31 ⁄ 35) compared to allantoically inoculated eggs (66%) (23 ⁄ 35) . however the yields were lower than observed with seasonal viruses. little improvement in virus yield was seen with extra passaging or dilutions of h1n1pdm viruses isolated in mdck33016pf cells. with the emergence of the swine-origin pandemic h1n1 (h1n1pdm) influenza in april 2009, 1 the need for efficient production of a suitable vaccine was a high priority. 2 virus isolates were distributed by the who for the urgent development of suitable vaccine strains early in the pandemic. vaccine viruses can be grown in embryonated chicken eggs 2 or in certified mammalian cells. 3, 4 unfortunately wildtype h1n1pdm virus strains distributed by the who grew poorly in cell lines and eggs, requiring the generation of a series of 3 conventional and 8 reverse genetics 5 derived reassortants to attempt to improve growth. from these reassortants, only the conventional egg derived reassortants nymc-x-179a and nymc-x-181 (both based on one of the earliest known viruses a ⁄ california ⁄ 7 ⁄ 2009) showed high enough growth and yield in eggs and cell culture to make them suitable for vaccine manufacture. these reassortants, while acceptable, still only gave haemagglutinin (ha) yields of approximately 60% that of seasonal h1n1 reassortants. to determine if more recent wild type h1n1pdm viruses grew better in the novartis mdck suspension cell line (mdck33016pf), h1n1pdm positive clinical samples were cultured in mdck33016pf cells and also in embryonated hen's eggs. in addition, to improve virus yields from mdck33016pf isolates, extended passaging of three wild type h1n1pdm influenza viruses was performed using various virus dilutions at each passage level. the results were assessed using various serological and molecular biology techniques and compared to viruses isolated in eggs and conventional mdck cells. h1n1pdm viruses were received at the centre from who national influenza centres, who influenza collaborating centres and other regional laboratories and hospitals in australia, new zealand, and the asia ⁄ pacific region. viruses were received as original clinical specimens consisting of nasal swabs, throat swabs, nasopharyngeal aspirates, or nasal washes that had previously been shown to be h1n1pdm positive by real time rt-pcr. these specimens were then cultured in mdck33016pf cells with serum free medium containing trypzean (optaflu) 3 and also independently inoculated into the allantoic cavity of 11 day-old embryonated hen's eggs. virus cultures in mdck33016pf cells were sampled at 48 and 72 hour and evaluated by various means including ha titres. at 72 hour, virus cultures were further passaged at varying dilutions ranging from 10 )5 to 10 )8 up to a total of 10 passages. embryonated hen's eggs were incubated at 35°c for 3 days and allantoic fluid was harvested and ha titres performed to determine whether a further passage was required in order to improve growth. the conventional reassortants were produced by a mixed infection of eggs or mdck 33016pf cells with the wild type virus and a donor virus carrying the internal genes of the a ⁄ puerto rico ⁄ 8 ⁄ 34 virus. the reassortants were obtained by sequential passages using immuno-selective antisera against the surface antigen of the donor virus to remove virus populations carrying the ha and na protein of the donor strain. 6 the reverse genetics viruses were rescued in vero cells using the 12 plasmid system. 7 both types of reassortants were generated and supplied by who collaborating centres and essential regulatory laboratories except the nvd-c-07 strain, which was produced by novartis. in this small study with recent h1n1pdm viruses, the isolation rate was higher in mdck33016pf cells (89%) (31 ⁄ 35) compared to allantoically inoculated eggs (66%) (23 ⁄ 35) . assessment of ha titres, however, showed higher ha titres in egg-isolated viruses compared to viruses isolated in mdck33016pf cells after two passages. egg generated or cell generated reassortant viruses gave higher ha titres compared to the homologous wild type viruses (table 1) . no amino acid changes were observed in mdck33016pf isolated influenza viruses compared to original specimens or viruses isolated in conventional atcc derived mdck cells, unlike egg isolated viruses which showed a number of amino acid changes, many consistent with egg adaptation mutations (table 1) . 10 viruses isolated in mdck33016pf cells grouped phylogenetically with viruses isolated in conventional atcc derived mdck cells or viruses sequenced from original clinical samples, while egg isolated viruses grouped slightly differently (data not shown). as a result of the poor growth of h1n1pdm viruses in mdck33016pf cells, serial dilutions were performed over a number of passages ( figure 1 ). based on the results obtained from the virus isolates, a ⁄ victoria ⁄ 2081 ⁄ 2009, a ⁄ wellington ⁄ 188 ⁄ 2009, and a ⁄ darwin ⁄ 2131 ⁄ 2009, a supplemental protocol was developed and used in the isolation of a ⁄ brisbane ⁄ 10 ⁄ 2010 (figure 1 ). only small differences in ha titer were seen between different dilutions, and copy number showed a similar trend to ha titer at each passage ( figure 1 ). following the supplemental protocol for the isolation of a ⁄ brisbane ⁄ 10 ⁄ 2010 results showed slightly higher ha titres with little variation between passages. the egg derived reassortants nymc-x-179a and nymc-x-181 were also assessed for growth in mdck33016pf cells and were found to be superior by ha titer to other conventional reassortants (egg or cell derived), reverse genetics derived reassortants, or wild type viruses (table 1) . two methods were used to determine the ratio of ha to other viral proteins: densiometric analysis using sds-page and reversed-phase hplc using a subtype specific standard. 11 ha content in different vaccine seeds of influenza a subtypes demonstrated that the ha content per total virus protein from the nymc h1n1pdm reassortants was significantly different to the seasonal influenza a subtypes. for the seasonal h1n1 the ratio of ha to p4 p5 p6 p7 p8 p9 p10 p3 p4 p5 p6 p7 p8 p9 p10 p3 n and m1 was ‡30%, for the h3n2 the ratio of ha to n and m1 was £30%, while for the pandemic a ⁄ h1n1, the ratio of ha to n and m1 was much lower at £20% (data not shown). the results of this study has observed the growth of a series of 2009-2010 h1n1pdm viruses in vaccine suitable mdck33016pf cells to be generally lower than what has been seen with other seasonal influenza viruses. 11 little improvement in virus yield was seen with extra passaging of h1n1pdm viruses isolated and passaged in mdck33016pf cells. passaging up to 10 times in mdck33016pf cells using dilutions ranging from 10 )5 to 10 )8 resulted in supernatants with viral ha titres ranging from 8 ha ⁄ 25 ll to 512 ha ⁄ 25 ll. the isolation rate of h1n1pdm viruses was higher in mdck33016pf cells (89%) compared to allantoically inoculated (and passaged) eggs (66%), a trend also seen in previous work with seasonal influenza viruses. 12 in contrast a study by hussain and colleagues 13 found similar rates of isolation and replication of seasonal influenza viruses in mdck cells and eggs. the virus load as determined by matrix gene copy number showed a similar trend to ha titers. two of the isolates exhibited small rises and falls in ha titer during passaging, while a third, a ⁄ victoria ⁄ 2081 ⁄ 2009 gave consistently higher titers. interestingly this virus was unable to be isolated in eggs. the ha sequences of all strains were assessed at p1, p2, p3, p4, p10 and when available compared to the original clinical sample ha sequence. mdck33016pf-isolated viruses had few if any changes in their ha amino acid sequence, while the majority of egg isolates showed 1-2 amino acid changes compared to the clinical sample, with an egg adaption change (l191i) evident in a number of them. the ha sequence of one of the better growing viruses, a ⁄ victoria ⁄ 2081 ⁄ 2009, was found to have a g155e change compared to the a ⁄ california ⁄ 7 ⁄ 2009 reference virus. this change was also seen in the virus isolated in conventional, adherent mdck cells. these viruses with g155e change when tested by hai have shown reduced reactivity with ferret antisera to a ⁄ california ⁄ 7 ⁄ 2009-like viruses, but normal reactivity with ferret antisera to h1n1pdm a ⁄ bayern ⁄ 69 ⁄ 2009-like viruses. despite this mutation all mdck33016pf derived viruses appeared to be a ⁄ california ⁄ 7 ⁄ 2009-like by hai. the h1n1pdm egg-derived reassortants (nymc x-179a and nymc x-181) when grown in mdck33016pf cells were superior to wild type h1n1pdm viruses, reverse genetics derived reassortants, and other egg-derived reassortants. the yields of haemagglutinin from the nymc h1n1pdm reassortants were still below those seen with sea-sonal h1n1 reassortants as was also seen in eggs. this trend has also been noted in other studies. 10 in summary, attempts to improve growth and yield of the h1n1pdm wild types for mdck 331016pf cells by extended passaging were not successful, and reassortants did not perform as well as seasonal h1n1 reassortants have in the past. however, using higher dilutions for the passaging of h1n1pdm viruses in mdck33016pf cells did result in higher ha titres (a ⁄ brisbane ⁄ 10 ⁄ 2010). further work is therefore required to generate pandemic h1n1 seed viruses that grow well in a variety of cell culture and egg based vaccine production systems. the aim of this study is to evaluate antibody response to influenza virus neuraminidase (na) following immunization with live attenuated influenza vaccine (laiv). we adjusted the peroxidase-linked lectin micro-procedure previously reported by lambre, et al. (1990) to assay neuraminidase inhibition (ni) antibody in sera taken from immunized mice and from human subjects in a clinical trial. for the assay, we prepared the a(h7n1) reassortant virus containing the na of a ⁄ california ⁄ 07 ⁄ 2009 (h1n1) and the hemagglutinin (ha) of a ⁄ equine ⁄ prague ⁄ 1 ⁄ 56(h7n7). in addition, we used an na-specific igg elisa assay to test sera from immunized mice and volunteers. in mice, one dose of laiv induced ni antibody of a geometric mean titer (gmt) of 31ae7, compared to 10ae6 in the control group. gmt of ni from human subjects who received two doses of pandemic a(h1n1) were significantly higher than pre-vaccination titers. in unvaccinated human subjects, na-specific cross-reactive antibodies to pandemic a(h1n1) were detected more often than cross-reactive antibodies to ha. antibody response to influenza virus na contributes to the overall immune response to influenza and may provide partial protection against influenza infection and reduce severity of disease in the host. 1 a number of preclinical studies using purified or recombinant na have shown that various two-dose vaccine regimens in mice may significantly reduce pulmonary virus titers following viral challenge. [2] [3] [4] a plasmid dna-vaccine model demonstrated cross-reactive antibodies to human n1 in mice could provide partial protection against a lethal challenge against h5n1 or recombinant pr8 bearing the avian n1. 4 immunogenicity of current influenza vaccines, including laivs, is measured primarily as a level of strain-specific hemagglutination inhibition (hi) antibodies. 5 however, the who meeting on the role of na in inducing protective immunity against influenza infection (2008) specified a need to develop suitable assays for anti-na antibody detection to enhance influenza vaccine evaluation in preclinical and clinical studies. 6 the aim of the current study was to evaluate anti-na antibodies to pandemic a(h1n1) 2009 influenza virus following laiv immunization. the rn1 ⁄ 09-swine a(h7n1) reassortant influenza virus containing the na of a ⁄ california ⁄ 07 ⁄ 2009(h1n1) and the ha of a ⁄ equine ⁄ prague ⁄ 1 ⁄ 56(h7n7) generated by classical genetic reassortment in embryonated chicken eggs (ce). parental a ⁄ equine ⁄ prague ⁄ 1 ⁄ 56(h7n7) influenza virus was obtained from the center for disease control and prevention, atlanta, ga, usa. viruses were propagated in 10 day old ce and purified by sedimentation out of the allantoic fluid, followed by ultracentrifugation on 30-60% sucrose step gradient. for the mouse studies, 10 week old cba mice were inoculated intranasally with one dose 10 7 eid 50 ⁄ 0ae05 ml of a ⁄ 17 ⁄ california ⁄ 09 ⁄ 38(h1n1) vaccine strain or received 0ae05 ml pbs. blood samples were collected on day 15 post inoculation. healthy young adults were immunized twice, 10 or 21 days apart in the fall 2009 with a ⁄ 17 ⁄ california ⁄ 09 ⁄ 38(h1n1) laiv manufactured by microgen, irkutsk, russia. for the human studies, peripheral blood specimens were collected from volunteers before vaccination, 21 days after the first vaccination, and 21 days after the second dose of vaccine. sera from five subjects diagnosed with influenza a(h1n1) were collected in december 2009, 3 to 4 weeks post infection and kindly provided by e. vo ıtsekhovskaia from biotechnology laboratory, institute of influenza, rams. also, sera obtained in 2005 from unvaccinated vol-unteers were tested for presence of cross-reactive antibodies to a ⁄ california ⁄ 07 ⁄ 2009 (h1n1). sera were treated with a receptor-destroying enzyme from vibrio cholera (denka-seiken, tokyo, japan) and then were tested in duplicates for hemagglutination-inhibition (hi) h1 specific antibodies by standard procedures 7 using a ⁄ 17 ⁄ california ⁄ 09 ⁄ 38(h1n1) test antigen. the peroxidase-linked lectin micro-procedure previously reported by lambre, et al. 8 was adjusted to assay ni antibody. briefly, 96-well plates (sarstedt, inc., nümbrecht, germany) were coated overnight with 150 ll of 50 lg ⁄ ml fetuin. the purified a(h7n1) reassortant virus was diluted in pbs with 1% bsa and 10 mm ca 2+ to give a four times higher optical density at 450 nm (od 450 ) compared to control wells not containing virus. fifty-microliter volumes of serially diluted serum samples were incubated with an equal volume of prediluted virus for 1 hour at 37°c. after incubation, the plates were washed and neuraminidase activity was measured by subsequently adding peroxidase-labeled lectin (2 lg ⁄ ml; sigma, st. louis, mo, usa), incubating for 1 hour at room temperature, washing the plates, and adding 100 ll of peroxidase substrate (tmb). the reaction was stopped after 5 minute by adding 100 ll of 1n sulfuric acid. od values were measured at 450 nm using the universal microplate reader (el x 800; bio-tek instruments, inc., winooski, vt, usa). the ni titers were expressed as the reciprocal dilution that gave 50% od of positive control (virus, no serum control). in addition we used an igg elisa assay 9 with 0ae5 lg ⁄ ml of purified na from a ⁄ california ⁄ 07 ⁄ 09(h1n1) to test sera from immunized mice and volunteers. data were analyzed with statistica software (version 6ae0) (statsoft, inc. tulsa, oklahoma, usa). geometric mean titers (gmt) were calculated and used to represent the antibody response. the comparisons were made within groups between pre-and postvaccinated titers (expressed as log 2 ) after first and second vaccination using wilcoxon matched pairs test. to compare multiple independent groups we used a kruskal-wallis anova with subsequent multiple pairwise comparison based on kruskal-wallis' sums of ranks. a p-value of <0ae05 was considered to be statistically significant. in mice, one dose of laiv induced antibody responses to both ha and na components of the a ⁄ california ⁄ 07 ⁄ 2009(h1n1) influenza virus vaccine (table 1) . geometric mean titers of ni antibody levels from vaccinated mice were 31ae7 and were significantly higher compared to those in unvaccinated control animals (p < 0ae02). elisa igg titers expressed as log 10 were 1ae7 compared to 1ae0 in control group. there was good correlation between antibody rises obtained using ni or elisa tests (r = 0ae7). in a study during the fall of 2009, 85% of 60 examined unvaccinated subjects were negative to pandemic a(h1n1) (hi titers £1:10). serum hi antibody titers to pandemic a(h1n1) ‡1:40 were considered to be protective against *the postvaccination gmts of hi antibodies after revaccination were higher than respective prevaccination titers (p = 0ae04) **the postvaccination gmts of ni antibodies after revaccination were higher than respective prevaccination titers (p = 0ae02) serum hi and ni antibodies to a ⁄ california ⁄ 07 ⁄ 09(h1n1) after one or two doses of pandemic laiv were evaluated in subjects who had pre-vaccination hi titers £1:10 ( table 2) . post-vaccination gmts of a(h1n1)-specific antibodies were significantly higher than pre-vaccination titers only among subjects who received two doses of laiv ( table 2 ). the frequency of subjects with ‡ fourfold rises in hi antibody titers was higher after two doses (32ae3%) compared to responses after one dose (7ae1%) although the differences were not statistically significant ( table 2 ). the highest antibody titers of hi and ni antibodies were achieved after natural infection (p < 0ae01 compared to all post-vaccination groups). all five subjects with confirmed influenza also had high levels of n1-specific igg measured by elisa using purified na as the coating antigen (data not shown). influenza ha and na surface proteins are primary targets of neutralizing antibodies that provide protection against influenza infection. the correlation of strain-specific hi antibody titers ‡1:40 to protection of 50% of the subjects against influenza infection is based on a number of reports published in 1980s. 10 serum antibodies against viral na as result of influenza infection or vaccination also can neutralize the virus from infecting cells; however, little is known about protective levels of such antibodies. to evaluate ni antibodies directed against pandemic a(h1n1) we used the reassortant a(h7n1) influenza virus with mismatched ha to avoid non-specific inhibition. we demonstrated laiv immunization effectively increased levels of ni antibody, although in smaller amounts compared to influenza infection. our data suggest that an antibody to neuraminidase, resulting from an earlier infection of the circulating seasonal influenza a(h1n1), evidently cross-reacted with the n1 of pandemic influenza virus, perhaps due to the previously reported 17% of conserved na epitopes in pandemic a(h1n1). 11 the peroxidase-linked lectin test using the reassortant a(h7n1) influenza virus was shown to be a sensitive and time effective means of revealing homologous and cross-reactive anti-na antibodies after laiv immunization or influenza infection. this could be a useful method for influenza vaccine evaluation. significant levels of anti-na antibodies detected in peripheral serum from subjects infected with wildtype h1n1 virus or with h1n1 laiv. and the cross-antibody response to ph1n1. for calculation of geometric mean titer (gmt), a titer of <10 was assigned a value of 5. statistical significance was determined by paired t-test. cross-reactive antibody response to ph1n1 in vaccinated populations of seasonal influenza virus table 1 shows the antibody response to seasonal influenza viruses and ph1n1 of participants. before vaccination, no or little antibody response to ph1n1 had been detected in all age groups. vaccination with seasonal influenza vaccines resulted in seroresponse in over 70% of subjects, except children aged 0-7 years (68%) and subjects aged of 18-59 years (69%) vaccinated with 2007-2008 season influenza vaccine and adults aged ‡60 years (59%) vaccinated with 2008-2009 season influenza vaccine. seroconversion was detected in over 40% of subjects of all ages. postvaccination to prevaccination gmt ratios for response to seasonal influenza viruses was more than 2ae5-fold. in contrast, seroresponse to a ⁄ california ⁄ 07 ⁄ 2009 after vaccination with 2007-2008 and 2008-2009 seasonal influenza vaccines were detected in only 3% and 4% of those aged 0-7 years, 3% of those aged 8-17 years, 7% and 3% of those aged 18-59 years, 7% and 5% of those aged ‡60 years, respectively. seroconversion in all participants ranged from 2% to 4%, and postvaccination to prevaccination gmt ratios were <2ae5-fold. preexisting antibody response to ph1n1 among subjects born before 1920s in china according to a recent report, people who were born from 1910 to 1929 had a preexisting immunity to ph1n1. 7 although only a very low level of cross-reactive antibody response to ph1n1 had been observed among older subjects aged more than 60 years old in china, we further analyzed these data by different age distribution of subjects, which can trace back to the previous infection that is genetically and antigenically more closely related to this new ph1n1 influenza virus. the proportion of seroresponse to ph1n1 with the titer of 40, 80, and 160 (highest titer detected from participants of all ages in this study) and the value of gmt were analyzed according to the birth decade of subjects from 1915. similarly, a peak of antibody response and the value of gmt occurred both in subjects born from 1915 to 1925 and sharply decreased afterward ( figure 1 ). the seroresponse of subjects born in and before 1925 is significantly higher than subjects born afterward (p < 0ae05). similar to recent studies in some asia countries (guangxi province of china and singapore), limited antibody response to ph1n1 had been detected in children and adults. 8, 9 but, some other studies from european countries (finland, germany, the united kingdom) and the united states reported a high proportion of older individuals aged >60 years with pre-existing cross-reactive antibodies to ph1n1, which may possibly ba a result of previous exposure to antigenically related h1n1 influenza viruses circulating in earlier decades or a lifetime of exposure to influenza a, which has resulted in broad heterosubtypic immunity among older individuals in those countries. previous infection and vaccination with a ⁄ new jersey ⁄ 76 may also contribute to the high level of cross-reactive antibody response to ph1n1 among adults older than 60 years in the us. 10, 11 the peak of the antibody response to ph1n1 in subjects born between 1915 and 1925, which is consistent with recent reports, may suggest the previous viral infections of 1918 spanish flu or closely related influenza viruses, which is before and little after the year of 1918. recent antigenic report of new ph1n1 viruses indicated that they are antigenically homogeneous among historical viruses, which are most similar to classical swine a(h1n1) viruses. a number of reviews [12] [13] [14] [15] confirmed that the 1918 virus is the likely ancestor of all four of the human and swine h1n1 and h3n2 lineages, as well as the 'extinct' h2n2 lineage. in 1930, a(h1n1) influenza viruses were first isolated from swine. they have been shown to be antigenically highly similar to the recently reconstructed human 1918 a(h1n1) virus. 16 the cellular responses may contribute to the sustaining and long term antibody response. probably, boosting by persisting antigenically related viruses in the early decades of the 20th century, may have contributed to the ability of these subjects to sustain memory b cells, 17 and it is well established that a subset of plasma cells is long-lived, and these cells contribute to durable humoral immune responses, 18 such as that observed after childhood smallpox vaccination. furthermore, t cells that recognize cross-reactive epitopes are preserved and might be enriched in the memory population; the course of each infection is influenced by the t-cell memory pool that has been laid down by a host's history of previous successive infections. 19 our study indicated that wide transmission of this new virus or any antigenically close related influenza a(h1n1) viruses may not have circulated among populations in china before the outbreak of ph1n1. our data also suggests the need for vaccination with ph1n1 vaccine in all age groups. hypo-and agammaglobulinemia patients have an impaired immune system and are particularly susceptible to bacterial infections that are normally defended against by antibodies. therefore, patients routinely receive replacement therapy with immunoglobulins isolated from healthy blood donors. 1 these patients are also prone to get viral infections, possibly due to defects in toll-like receptors 8 and 9. 2 because these patients lack an antigen specific humoral immune response, they are rarely vaccinated. the ability of hypogammaglobulinemic patients to produce a specific cell-mediated immune response upon vaccination has only been sparsely investigated. in contrast to local mucosal antibodies, vaccine-induced cell-mediated immunity is not believed to protect against pathogen entry per se, but may be sufficient to provide protection against severe disease and death following transmission of some microbes. 3, 4 the aim of this pilot study was to investigate if influenza vaccination of hypogammaglobulinemic patients can induce an influenza-specific cell-mediated immune response. we therefore vaccinated hypogammaglobulinemic patients and healthy controls with pandemic h1n1 virus vaccine and subsequently investigated the bcell and t-cell responses. the percentages of ifn-c, il-2, and tnf-a cytokine producing cd4+ th1-cells were determined, as these cytokines are important indicators of cell-mediated immunity. five a-or hypogammaglobulinemic patients were classified based on the freiburg classification 5 : patient #1 is diagnosed with x-linked agammaglobulinemia, patient #2 and #4 are in group ia, patient #3 is in group ib and patient #5 is in group ii. the monovalent egg grown split virus vaccine adjuvanted with as03 was manufactured by glaxosmithkline (gsk), belgium. the vaccine strain was produced by reassortment between influenza a ⁄ california ⁄ 7 ⁄ 2009 (h1n1) and a ⁄ pr ⁄ 8 ⁄ 34 (h1n1) to produce a ⁄ california ⁄ 7 ⁄ 2009-like virus (x179a). the vaccine was mixed with adjuvant to contain 7ae5 lg haemagglutinin (ha) of a ⁄ california ⁄ 7 ⁄ 2009-like virus (h1n1), squalene (21ae38 mg), dl-atocopherol (23ae72 mg), and polysorbate 8 (9ae72 mg) per ml. healthy controls and hypogammaglobulinemia patients were vaccinated by intramuscular (im) injection. hypogammaglobulinemia patients received one or two vaccine doses 21 days apart. the intention was to vaccinate the hypogammaglobulinemic patients with two doses of 3ae75 lg ha, but 7ae5 lg ha was inadvertently administered to the patients as the first dose. for patient #4 this was the second dose as he had received an initial dose of 3ae75 lg ha 3 months prior to the study. patient #1, #2, and #5 received a second dose of 3ae75 lg ha. four healthy controls were immunised with one dose of 3ae75 lg ha according to norwegian national guidelines. peripheral blood mononuclear cells (pbmcs) were harvested and washed in pbs with 10% fbs. the pbmcs were resuspended in lymphocyte medium (rpmi 1640 with l-glutamine, 0ae1 mm non-essential amino acids, 10 mm hepes ph 7ae4, 1 mm sodium pyruvate, 100 iu ⁄ ml penicillin, 100 lg ⁄ ml streptomycin, 0ae25 lg ⁄ ml fungizone and 10% fbs) prior to use in the enzyme-linked immunospot (elispot) and influenza-specific cd4+ t-cell assays. serum haemagglutination inhibition antibodies were tested by a standard method using 8 ha units and 0ae7% turkey erythrocytes. all samples were tested in duplicate and the test was repeated at least two times. titres <10 were assigned a value of 5 for calculation purposes. for numeration of antibody-secreting cells (asc), an eli-spot assay was conducted as previously described 6 with the following modifications. ninety-six well elispot plates were coated with 2 lg ⁄ ml of a ⁄ california ⁄ 7 ⁄ 2009like (x179a) h1n1 virus diluted in pbs overnight at 4°c. after blocking with rpmi (10% fbs), 10 5 pbmcs were added and incubated (37°c, 5% co 2 ) for 16 hour. secreted antibodies were detected with biotinylated goat anti-human igg, iga and igm specific antibody (southern biotech, birmingham, alabama, usa), incubated for 2 hour at room temperature and developed with extravidin peroxidase and aec substrate. the numbers of spots were counted using an elispot reader (immunoscanô) and immunospot ò software. the influenza-specific cd4+ th1-cell response was measured by intracellular cytokine production of ifn-c, il-2, and tnf-a. peripheral blood mononuclear cells (10 6 per well) were incubated for 16 hour (37°c, 5% co 2 ) in 200 ll lymphocyte medium containing 1 lg ⁄ ml anti-cd28, 1 lg ⁄ ml anti-cd49d, 0ae7 lg ⁄ ml monensin, 1 lg ⁄ ml brefeldin a, (bd biosciences, franklin lakes, new jersey, usa), and the h1n1 influenza split virus vaccine x179a (either 2ae5 lg ⁄ ml or 10 lg ⁄ ml ha). basal cytokine production was determined by incubating pbmcs in lymphocyte medium without influenza virus, and the percentage of cytokine positive cells without influenza stimulation were subtracted from influenza-stimulated cells. cells were stained for cd3, cd4, cd8, ifn-c, il-2, and tnf-a (bd biosciences) as previously described. 7 finally, cells were resuspended in pbs containing 5% fbs and 0ae1% sodium azide and analysed by bd facscanto flow cytometer (300 000-500 000 cells acquired). flowjo v8ae8ae6 (tree star, ashland, oregon, usa) was used for data analysis. five to six fold lower gmts were found in the patient group as compared to the healthy controls throughout the study ( figure 1a) . the lowest hi titres were obtained in patients #1, #2, and #3, whilst patients #4 and #5 and all healthy controls fulfilled two of three european medicines agency committee for medicinal products for human use (chmp) 8 seasonal influenza vaccine licensing criteria, by obtaining an hi titre >40 and a mean geometric increase of 2ae5 between pre-and post-vaccination. thus, the hi data indicate that two vaccine doses was sufficient to induce a protective hi antibody response in two out of five of the hypogammaglobulinemia patients tested in this study. the numbers of influenza-specific iga, igg, and igm asc were tested pre-vaccination and 7 days post-vaccination with the h x179a virus. few or no ascs were detected pre-vaccination (data not shown). at 7 days post-vaccination the patient's iga, igg, and igm asc levels were significantly lower (p < 0ae01) compared to the healthy controls ( figure 1b) . but, the post-vaccination asc numbers in the patients were generally higher than at pre-vaccination stage (0-2 ascs). patient #3 had the highest iga and igg asc numbers, followed by patients #2 and #5, whilst patient #4 and #1 had few or no asc's. these results confirm that the patients are indeed hypogammaglobulinemic and that some of the patients (#1 and #4) could be agammaglobulinemic in the context of producing influenza-specific antibodies. the asc levels of patients #2, #3, and #5 were lower than those of the healthy controls, but could possibly be adequate for reducing the severity of influenza disease. the influenza-specific th1-cell response was evaluated by stimulating pbmcs with the influenza x179a virus 21, 42, and 90 days post-vaccination. stimulation of healthy control pbmcs with x179a 21 days after vaccination, induced ifn-c, il-2, and tnf-a production by an average of 0ae10%, 0ae09%, and 0ae03% cd4+ t-cells, respectively. patient #1 and #2 had higher responses than the healthy controls and stimulation with x179a induced 0ae45%, 0ae34%, and 0ae05% of t-cells from patient #2 to produce ifn-c, il-2, and tnf-a, respectively (figure 2a) . the response of patient #2 was further boosted by a second vaccine dose, which resulted in 0ae73%, 0ae54%, and 0ae25% cd4 t-cells producing ifn-c, il-2, and tnf-a, respectively at day 42 ( figure 2b ). these results show that the hypogammaglobulinemia patients studied here did not have a common impaired influenza-specific cd4+ th1 cytokine response. rather, there was a tendency towards increased responses, suggesting that the diminished antigen specific b-cell responses could induce a compensatory antigen specific th1-cell response. the results from this pilot study suggest that some hypogammaglobulinemia patients may benefit from influenza vaccination. we found very different patient responses to influenza vaccination, but some of the patients (patient #2 and #3) did mount low influenza-specific asc responses. in addition, the vaccine-induced hi antibody titres above the protective level in patient #4 and #5. these results are in accordance with previous publications, which described that polypeptide vaccines induce humoral responses in subgroups of common variable immunodeficiency patients. [9] [10] [11] in this study, we also investigated cell-mediated immunity and found the percentages of homologous and cross-reactive influenza-specific cd4+ th1-cells to be in the same range (for patient #3, #4, and #5) or higher (for patient #1 and #2) in the a-or hypogammaglobulinemic patients compared to the healthy controls. the higher response is probably due to the patients having received a vaccine dose of 7ae5 lg ha, whilst the controls received 3ae75 lg ha. in addition, the patients received a second booster dose, which influences the day 42 and 3 months responses. nonetheless, these results are the first to demonstrate that proliferation of pandemic influenza antigen specific th1cells can be induced in hypogammaglobulinemic patients. in addition, vaccination induced influenza-specific asc's in some patients. the findings are promising and provide hope that hypogammaglobulineamic patients could be vaccinated against influenza and other diseases preventable by figure 2 . peripheral blood mononuclear cells s from patients and healthy controls were isolated at day 21 (a), 42 (b), and day 90 (c) and stimulated for 17 hour with x179a virus before staining and flow cytometric analysis. the figure shows the mean ± sd frequency of influenza-specific cd4+ cytokine producing cells (%) where the basal cytokine production from unstimulated cells has been subtracted. data for the hypogammaglobulinemia patients are additionally shown as a number for each patient. **significantly higher frequency of il-2 producing cd4+ t-cells in the patients compared to the healthy controls (students t-test p < 0ae05). titres are presented as the geometric mean titre ± 95% confidence interval. elispot data (b) are presented as the mean number of influenza-specific iga, igg, and igm ascs per 100 000 peripheral blood mononuclear cells ± sem. data for the hypogammaglobulinemia patients are additionally presented by a number for each patient. *significantly higher numbers of ascs were detected in the healthy controls as compared with the hypogammaglobulinemia group (students t-test, p < 0ae01). vaccination. however, this hypothesis should be tested in larger clinical studies. the influenza virus undergoes antigenic evolution under intense immune selection pressure from herd immunity in humans through the process called antigenic drift and shift. 1,2 because of antigenic drift, yearly updating of vaccine strain is needed. a mismatch between the circulating strains and the vaccine strain in the subsequent season is often encountered, resulting in reduction of vaccine effectiveness and lack of protection from the circulating strain. 3 in order to address this, a universal influenza vaccine based on a more conserved part of the influenza virus, which is not affected by antigenic change and that is conserved across all strains, remains the ultimate goal to afford cross-protection to drifted strains as well as to other subtypes of influenza which may arise from antigenic shift. 4, 5 previous studies have investigated the potential of the m2e. 6, 7 m2e has remained highly conserved since it was first isolated in 1933. 6 several studies have examined the use of m2e as a vaccine component, using various approaches including proteins, peptides, dna vectors, and attenuated viral vectors. 6, [8] [9] [10] [11] [12] [13] although m2e is a weak antigen, by linking the protein to a carrier hepatitis b virus core particle, protection against influenza has been achieved in mice particularly when administered with an adjuvant. 3 some articles found that vaccination with m2e coupled to hbc induces protective antibodies, whereas the contribution of t cells to protection was negligible. protection induced by vaccination with m2e-hbc was weak overall and failed to prevent weight loss in vaccinated infected animals, and mice succumbed to high dose infection. 14 we aimed to address the poor immunogenicity of m2e-hbc by using igv as adjuvant. igv domain is common and conserved in the tim family. ligand binding sites of t cell immunoglobulin mucin (tim) located at igv domain. [15] [16] [17] tim function is done by anti tim antibody which recognized the ligand binding sites of igv domain. 18 tim family members share a common motif, including an igv domain. they are differentially expressed on th1 cells and th2 cells with the ability to regulate the immune system. 19, 20 the igv domain of human b7-2 is sufficient to co-stimulate t lymphocytes and induce cytokine secretion. 21 soo hoo et al. vaccinated with tim-1 antibody and inactivated influenza and found enhanced vaccine-specific immune response. 22 we report here for the first time the use of igv recombinant protein as adjuvant to immunize mice with influenza m2e-hbc. results indicated that igv can induce the strong cellular immune response and cross reaction with different subtype influenza virus antigen. target igv may be used to develop the new method for vaccination strategies. expression and purification of recombinant igv protein rna was extracted from healthy human pbmc. one-step rt-pcr (qiagen, valencia, ca, usa) was done for the amplification igv gene. the pcr product was purified and cloned into pet32a (novagen, madison, germany). the resultant construct pet32a-igv has a histidine (his) tag (6his) at the n terminus. dna sequence of the insert was determined by sequencing. igv. recombinant protein was expressed in escherichia coli and was purified on a ni column (novagen). the purified protein was examined by sds-page and western blotting. six-eight weeks female balb ⁄ c mice (institute of zoology chinese academy of sciences, china) was used for the study. mice were immunized twice intradermally with 10 ug m2e-hbc (provided by cnic, china) combined with different doses of recombinant igv protein 5, 10, 50 ug, respectively, or without igv as control. the area proximal to the tibialis anterior muscle was sterilized with 70% ethanol and different groups of mice were injected bilaterally with 5, 10, 50 ug igv plus 10 ug m2e-hbc in 50 ul phosphate buffer saline per mouse using a 1 ml syringe with attached 1 ⁄ 2¢¢ 30g needle. the immunization was given at 3 weeks intervals. four blood samples were obtained from every mouse: before immunization, after the first and second immunization, and after virus challenge by retro-orbital plexus puncture. after clotting and centrifugation, serum samples were collected and stored at )80°c prior to use for assays. mouse-adapted a ⁄ pr ⁄ 8 ⁄ 34 (h1n1), a ⁄ brisbane ⁄ 1 ⁄ 10 (h3n2), a ⁄ xinjiang ⁄ 1 ⁄ 2006 (h5n1), and a ⁄ guangzhou ⁄ 333 ⁄ 1999 (h9n2) were provided by chinese national influenza centre. nine to eleven days old embroynated specific pathogen free (spf) chicken eggs were inoculated with virus, and the eggs were incubated at 35°c for 2-3 days. the allantoic fluid was collected and purified by sucrose density gradient centrifugation, and the virus was inactivated by formaldehyde at 4°c overnight. to identify igg, igg1, igg2a against m2e, elisa assays were used. in brief, 96-well (nunc, brunei, denmark) were coated with 100 ul ⁄ well of m2e recombinant protein (provided by gene lab of ivdc, xuanwu district, beijing, china) in carbonate buffer (ph 9ae6) overnight at 4°c. immediately before use, the coated plates were incubated with blocking solution (2% bsa in pbs) for 2 h at 37°c and washed four times with pbs containing 0ae05% tween 20 (pbs-t). the serum samples were serially diluted and added in the plates. the detection color was developed by adding hrp-labeled goat anti-mouse igg, igg1, or igg2a ( figure 1) . no cross-strain response was observed in the control group. the igv adjuvented groups show splenocytes stimulation with seasonal h1n1, h3n2, h5n1, and h9n2 antigens. m2e-hbc immunization without igv showed splenocyte stimulation, but the extent was lower than animals immunized in the presence of the igv adjuvent. these data suggested that igv had enhanced effect on priming against the conserved viral antigen matrix protein and generation cross-strain immune response. influenza is a respiratory disease causing epidemics every year. h5n1 viruses and swine-origin h1n1 have also infected humans in recent years. seasonal influenza vaccine cannot cope with significant antigenic drift or with the emergence of pandemic viruses of different subtypes not contained in the vaccine. the high extent of conservation of the m2e makes it a promising immunogen. a vaccine based on coupling of the m2e peptide to an appropriate carrier may provide a universal vaccine with effectiveness and safety. m2e based vaccination induces protective antibodies not only in mice, but also in ferrets and monkeys. the carrier hepatitis b core as carrier with m2e forms a virus like particle (vlp). vaccination with m2e coupled to hbc induces protective antibody, 11 whereas the contribution of t cell protection was negligible. protection induced by vaccination with m2 coupled to hbc was weak overall. in order to improve the vaccination effect of m2e-hbc, new adjuvant igv was evaluated in combination with the m2e-hbc. the tim molecules are a recently discovered class of proteins with the ability to regulate the immune system. crystal structures of the tim molecules has revealed a unique, conserved structure with ligand-binding sites in the igv domain. to determine the potential immunostimulatory molecular properties of igv, we have evaluated immune response of the igv in combination with m2e-hbc vlp. previous papers reported that vlp immunized mice can induce the th1 and th2 immune response. 23 different adjuvant combined vlp can produce biased immune response th1 ⁄ th2 mixed immune response, or th1-preferred th1 ⁄ th2 profile. 24 thus, the response following the use of igv as a new adjuvant combined with m2e-hbc vlp needs to be evaluated. results indicated that igv combined groups showed th1 biased immune response and enhanced cross reactive t cell immune responses. this may show that igv immunized the mice and antiigv antibody can cross link the igv on t cells and enhance the cell figure 1 . t cell proliferation assay. mice were immunized twice with 50, 10, 5, 0 ug ⁄ ml igv plus m2e-hbc, respectively, and naive group was immunized with pbs. three weeks after a boosting immunization, spleens were harvested from immunized and naive mice. different subtypes of inactivated virus antigen (a) h1n1, (b) h3n2, (c) h5n1, (d) h9n2 were added and cocultured with different group splenocytes for 96 h. quick cell proliferation assay kit was used to detect the cell proliferation. the 420-480 nm absorbance was read on a plate reader. data were showed were shown as mean values. the difference between naive group and different doses igv plus m2e groups was determined using the student's t-test. all significance level is p < 0ae05. response. we also evaluated the cross-protection produced by igv combined m2e-hbc. we challenge with mouseadapted strain pr8 and prove the cross protection via reaction between the cells from the immunized animal and different subtypes of virus antigen. some subtypes of virus cannot infect the mice naturally, and therefore, virus challenge cannot be used to evaluate the effect. we co-cultured the t cells with inactivated antigen h1, h3, h5, and h9, and t cell proliferation was measured. results indicated that after immunization with igv plus m2e-hbc, the t cells show cross-protection with other subtypes. this provides evidence that igv can enhance the cross protection across subtypes. the results of this study demonstrated that recombinant igv can be useful as an adjuvant and polarize the m2e-hbc vlp immune response to a th1 profile. igv induced the m2e-hbc vlp to induce t cell proliferation and cross-reactive responses to different influenza virus subtypes. this finding represents a new direction for the promotion of cell mediated immunity in m2e based vaccine against influenza. a core european protocol, i-move, describing the methods to estimate influenza vaccine effectiveness (ive) was proposed by the european centre for disease prevention and control (ecdc) and epiconcept for the 2009-2010 season. it includes a case control method for pooled analysis based on a randomized ''systematic'' sample of swabs. 1, 2 collection of swabs using a non randomized, i.e., ''ad hoc,'' sampling strategy, left at the appreciation of sentinel practitioners, provides a greater number of cases and con-trols for ive estimation more easily than using a systematic randomized sampling strategy. the french grog (groupes régionaux d'observation de la grippe) early warning network collects more than 5000 specimens yearly from cases of acute respiratory illness (ari), using both sampling methods. 3, 4 during the circulation of pandemic influenza 2009 viruses in france, it gave an opportunity to compare ive estimates using systematic randomized versus non systematic ''ad hoc'' sampling. influenza vaccine effectiveness was estimated by a casecontrol methodology according to ecdc i-move protocol, using on the one hand a systematic random sampling, on the other hand ''ad hoc'' non random sampling. the study was proposed to 608 primary care practitioners of the grog network (493 general practitioners and 115 pediatricians) trained to collect data and swabs. the study population was patients from the community of all ages consulting a grog practitioner for an influenza like illness (ili) and having a nasal or throat swab taken within an interval of <5 days after symptom onset. ili was defined according to the european union (eu) case definition as sudden onset of symptoms with at least one of the following four systemic symptoms: fever or feverishness, malaise, headache, myalgia; and at least one of the following three respiratory symptoms: cough, sore throat, shortness of breath. swabs were performed through usual surveillance. no ethical approval was needed, but an oral informed consent was requested. cases were excluded if they refused to participate in the study or if they were unable to give informed consent or to follow the interview in native language because of aphasia, reduced consciousness, or other reasons. an individual was considered as vaccinated against pandemic influenza if he or she reported having received a pandemic influenza vaccination during the current season, and if at least one vaccine dose occurred more than 14 days before ili onset. the study period started with the initiation of active influenza surveillance by the grog network, i.e., 15 days after the beginning of the influenza vaccination campaign, and finished at the end of the influenza period defined as the last week with at least one swab positive for influenza within the grog network. ''ad hoc'' sampling patients from which swabs were taken were selected by the grog practitioners during the study period. systematic random sampling during the same period, patients were selected at random as follows. an age-group 0-4 years (gps and pediatricians); 5-14 years (gps and pediatricians); 15-64 years (gps); 65 years or more (gps) was assigned to each practitioner, who was requested to swab the first patient of the week presenting with an ili within the pre-assigned age-group. swabs were collected in appropriate transport medium (virocult ò , viralpack ò , utm copan ò ) and sent by post to the laboratory in triple packaging following the international guidelines for the transport of infectious substances (category b, classification un 3373). laboratory confirmation of influenza was by rt-pcr to detect currently circulating influenza a (subtypes h3, seasonal and pandemic h1) and b viruses. an influenza case was defined as an ili case with a respiratory sample positive for influenza during the study period. controls were cases of ili having a swab negative for influenza during the study period. the outcome of interest is laboratory confirmed influenza. confounding factors and effects modifiers identified during the i-move preliminary study 5 were registered: risk factors, chronic diseases, severity of underlying conditions, smoking history, former vaccinations, and functional status. data on cases and controls were collected by the practitioners using a standardized questionnaire adapted from the i-move study. questionnaires were sent by the practitioners with the swab to the virology laboratory, and sent to the grog national coordination. data entry and validation were ensured by open rome through the vircases computing tool. validation steps included control of exhaustiveness of centralization of questionnaires, comparison of data entered by the labs and the national grog coordination, coherence control, and identification of missing data. analysis was done for the two sampling groups (systematic and ad hoc) on cases ⁄ controls following the european method proposed by epiconcept, using excel 2007 ª (microsoft corp. redmond, washington, usa) and stata ª . baseline characteristics of cases and controls in unmatched studies were compared using the chi-square test, fisher's exact test, or the mann-whitney test (depending on the nature of the variable and the sample size). the association between vaccination status and baseline characteristics was assessed for both case and control groups. the vaccine effectiveness was computed as ive = 1)or (odds ratio). an exact 95% confidence interval (ci) was computed around the point estimate. analysis was stratified according to age groups, time (month of onset), presence or absence of chronic disease, and previous influenza vaccination. effect modification was assessed comparing the or across the strata of the baseline characteristics. confounding factors were assessed by comparing crude and adjusted or for each baseline characteristic. a multivariable logistic regression analysis was conducted to control for negative and positive confounding factors using a complete case analysis (with records with missing data dropped) and using multiple imputation with chained equations. the complete model included age group, number of gp visits, onset week, seasonal vaccination, previous seasonal influenza vaccination, presence of chronic disease and associated hospitalizations in the previous 12 months, gender, and smoking status. variables were tested for multi-colinearity. interactions were tested using the likelihood ratio test (or wald test) and included in the model if significant at 5% level. a model with fewer variables (age group, number of gp visit, onset week, and seasonal vaccination) was also tested. several models were applied to both the ''ad hoc'' and systematic sampling groups of cases and controls. as shown in table 1 , whatever the analysis method used, the ''ad hoc'' sampling strategy led to a slightly lower estimate of ive. the ci were extended when data were missing and reduced when using multiple imputations with chained equations. however, from a statistical point of view, comparison of ''ad hoc'' versus systematic strategies is not straightforward, because ''ad hoc'' sampling is not randomized and does not allow comparisons with statistical tests using statistical distribution laws. there are more missing data with the ad hoc sampling method. this is mainly due to our validation procedure: in the case of missing data in the systematic sampling group, as required by the i-move study protocol, queries were sent to sentinel practitioners using mail and phone calls. this specific heavy workload is not usually performed during routine surveillance and has not been achieved for the ''ad hoc'' sampling group given the great number of cases and controls (2690). within the framework of the i-move study, several items were added to the grog's usual clinical form accompanying swabs (hospitalizations, number of gp visits, smoking status, help needed for bathing or walking). in 2009-2010, gps explained that this added workload was not compatible with their daily additional workload due to the pandemic situation. therefore, many of them refused to fill these new items systematically and threatened to leave the network. we thus obtained that the ''i-move items'' would be filled in for the clinical forms linked to systematic sampling, but were not in a position to obtain that for ''ad hoc'' sampling. the weekly distribution of systematic swabbing is not similar to that of ad hoc swabbing. the percentage of ad hoc swabs was higher than systematic swabs during the pandemic wave (mid-november to end of december) during which time the percentage of swabs positive for influenza was also higher ( figure 1 ). this could explain the higher rate of positive swabs within the ''ad hoc'' samples. the vaccination campaign was launched by the ministry of health on october 20, 2009, 6 and vaccination coverage increased during the surveillance period. in february, the vaccination coverage was 10ae2% in patients swabbed in the systematic group (9ae6% on imputed data) and 11ae5% [4ae4-23ae4] in the ad hoc group (11ae9% on imputed data). at the national level, vaccine coverage is estimated at 7ae95%. 7 due to the over-mediatisation of pandemic vaccination and to rumors about its poor effectiveness, overconsultation of vaccinated patients and over-swabbing of vaccinated patients in the ad hoc group are not surprising. age distribution is significantly different between our two samples (p < 0ae0001): the rate of 5-14 years old is lower in the systematic sampling group (25ae5%) than in the ad hoc sampling group (35ae7%). this can be explained by the fact that for the systematic sampling procedure, each grog practitioner had to swab the first ili patient in his assigned age group, whereas for ''ad hoc'' sampling, every grog practitioner could swab any ili patient irrespective of age. given the emphasis by health authorities and media on the burden of pandemic influenza among children and teenagers, one can hypothesize that when they were able to, sentinel practitioners focused on these age groups. gps in the ad hoc sampling scheme seem to have been more likely to select cases and further, to select vaccinated cases. those patients may have consulted earlier with specific symptoms (strong headache being more prevalent among cases). over-swabbing of patients having these symptoms in the ad hoc group is likely. the 2009-2010 pandemic influenza season was markedly different from previous ones: vaccination rate increased during and mainly after the pandemic peak; behaviors were strongly modified by unusual media hype; clinical features and risk factors might be different. it will be necessary to see if similar results are observed during a regular influenza season during which the vaccination rate increases before the epidemic peak with usual messages about vaccination and usual clinical influenza features. influenza early warning networks can estimate ive, taking into account many covariates. from a stakeholders and patients point of view, during the 2009-2010 influenza pandemic wave, there were no major discrepancies between ive estimated with an ad hoc sampling strategy, based on sentinel practitioners instinct, and ive estimated with a systematic random sampling strategy whatever the multivariable analysis methodology. although from a statistical point of view, comparison of the two strategies is not readily feasible because of the non random nature of ad hoc sampling. this latter strategy seems to result in slightly lower ive estimates, which could potentially be attributed to sentinel practitioners swabbing behavior. the ability to avoid missing data is a key point to decide which sampling method must be adopted, because ci extent depends greatly on the proportion of missing data among covariates. to match ive evaluation to surveillance networks practicality, selection of only those data essential for the study endpoint and easily collected by sentinel practitioners is paramount. it will be necessary to determine if results similar to those observed during the 2009-2010 pandemic season are found during a regular influenza season. influenza a viruses are important pathogens which remain a major cause of morbidity and mortality worldwide, and large numbers of the human population are affected every year. the first influenza pandemic in this century broke out in humans in march 2009, and it was declared to be pandemic by mid-june. as of 1 august jul 2010, the pandemic virus had caused more than 18 449 deaths worldwide, according to the world health organization (http:// www.who.int/csr/don/2010_08_06/en/index.html). the infection and spread of the pandemic influenza was reduced in part due to the use of vaccines. however, the lack of h1n1pdm vaccine early in the pandemic illustrates the need to improve vaccine production and to generate vaccines that induce stronger cross-protection. inactivated split vaccines or live attenuated influenza virus vaccines (laivs) against h1n1pdm viruses were approved for human use by the united states food and drug administration. both the inactivated vaccines and laivs are produced by creating reassortant viruses that generally contain six vrnas (pb2, pb1, pa, np, m, and ns) from a master donor strain, plus the two glycoprotein vrnas (ha and na) from a virus that antigenically matches the strain predicted to circulate in upcoming influenza season (e.g. a ⁄ ca ⁄ 07 ⁄ 2009). the reference viruses containing inactivated split virus vaccines are produced in embryonated chicken eggs, and primarily result in the production of antibodies that recognize the viral glycoproteins. both of these vaccine approaches require significant lead time for vaccine production, and modern approaches to speed preparation of vaccines and improve their efficacy is a global priority. 1, 2 the ns1 protein of influenza a virus is a multifunctional protein that plays important roles in virus replication and as potent type i ifn antagonist. 3, 4 mutations and ⁄ or deletions in ns1 typically induce stronger ifn responses by the host; those in turn suppress the replication of influenza virus 3-7 and can enhance immune recognition. [8] [9] [10] [11] in this study, we created a panel of experimental h1n1pdm ns-laiv candidates that have different deletions in the ns vrna and analyzed the vaccine potential of each ns-laiv in mice and ferrets to identify the best candidate(s). wt h1n1pdm influenza a virus a ⁄ new york ⁄ 1682 ⁄ 2009 (ny1682) was created by reverse-genetics directly from a human swab specimen collected in new york state in april 2009. 12 deletions were introduced into the ny1682 ns plasmid to create three mutant ns segments: ns1-73, ns1-126, and nsd5. nucleotides 246-482 (cdna of ns segment) and 405-482 were replaced by stop codons to generate ns1-73 and ns1-126; nucleotides 612-626 were deleted to generate nsd5, whose open reading frames for ns1 and nep were maintained. recombinant viruses were generated by co-transfection of eight reverse-genetics plasmids carrying the cdna of each gene segment into 293t ⁄ mdck cocultured monolayer adapted from hoffmann et al. 12, 13 mouse studies experiments were performed in a biosafety level 3 laboratories approved by the u.s. centers for disease control and prevention and the u.s. department of agriculture, and were conducted under approved animal care and use protocols. groups of 6-week-old female balb ⁄ cj (jackson laboratory, bar harbor, me, usa) were anesthetized with isoflurane and inoculated intranasally with 10 5 tcid 50 of each recombinant virus in 20 ll of pbs diluent, or pbs as controls. body weights and clinical symptoms of the mice were monitored daily for 10 days. nine mice in each group were euthanized on 1, 2, and 5 days post inoculation (dpi), and nasal washes and lungs were collected for virus titration by tcid 50 assay in mdck cells. at 30 dpi, 12 mice per group were challenged intranasally with 5 · 10 4 tcid 50 (100 ld 50 in 6-week-old mice) of a mouse-adapted variant of ny1682 (a ⁄ ny ⁄ 1682 ⁄ 2009-ma7) (accepted, journal of virology). disease symptoms and weights of the vaccinated mice were monitored for 10 days, and four mice from each virus group were euthanized at 3 and 6 days post challenge. lungs were removed and homogenized for virus titration by tcid 50 assay. the mice that became moribund or lost >25% of their starting body weight were euthanized for humane reasons. male fitch ferrets (triple f farms, sayre, pa, usa), 7-10 months of age and serologically negative by hemagglutination inhibition (hi) assay for currently circulating influenza viruses were used in this study. groups of 3 or 4 ferrets were inoculated intranasally with 10 6ae5 tcid 50 of one of the viruses: ny1682 wt (n = 4), ns1-73 (n = 3), ns1-126 (n = 4), or nsd5 (n = 4). ferrets were monitored for clinical signs through 14 dpi as previously described. 14 nasal washes were collected on 1, 3, 5, and 7 dpi and were titrated in mdck cells by tcid 50 assay. serum was isolated from blood collected 6ae5 weeks after immunization and used for neutralization assays. the ferrets were challenged with 10 6 pfu of a ⁄ mexico ⁄ 4482 ⁄ 2009 15 6ae5 weeks postimmunization and monitored for clinical signs of disease through 14 dpi. nasal washes were collected on 1, 3, 5, and 7 dpi, and were titrated in mdck cells by plaque assay. using reverse genetics, we created three laiv candidates weight loss of wt virus inoculated mice became evident at 3 dpi, and the mice did not recover until 8 dpi (figure 1a) . in contrast, mice inoculated with any one of the vaccine candidates had no clinical signs of disease and continued to gain weight at the same rate as did the mockinoculated mice ( figure 1a ). viral titers in the lungs of ns1-73, and ns1-126 infected mice were $100-fold lower than titers from wt virus-infected mice at all the time points analyzed (1, 2, and 5 dpi) ( figure 1b) . notably, the nsd5 laiv was cleared from the mouse lungs very rapidly, and the mean titers were $100-fold and 50 000-fold lower than the titers of the wt virus at 1 and 2 dpi, respectively ( figure 1b) . the vaccinated mice were challenged with a mouseadapted variant of ny1682 (accepted, journal of virology) on 30 dpi. no disease symptoms were observed in the mice immunized by any of the ns-laiv candidates or the wt control. in contrast, disease symptoms including ruffled fur, hunched posture, and weight loss were observed in the mock-immunized mice as early as 2 days post challenge (dpc); the symptoms progressed to severe disease, and the animals showed dramatic weight loss, became moribund, and succumbed to infection by 5 dpc (figure 1c ). high titers of virus ($10 7 tcid 50 ⁄ ml) were present in the mock-immunized mice at 3 dpc and at 6 dpc ( figure 1d ). in contrast, virus was not detected in the lungs of immunized mice ( figure 1d ). this challenge data demonstrates that all of the ns-laiv candidates, including the highly attenuated nsd5, induced sterilizing immunity that protected mice from a lethal ny1682 h1n1pdm variant. groups of ferrets were intranasally immunized with 10 6ae5 tcid 50 of each vaccine candidate or the wt virus. the titer of viruses recovered from nasal washes ranged from 10 6ae8 to 10 4ae5 tcid 50 ⁄ ml through day 5 in the wt virusinfected group, while the ns-laivs showed various degrees of attenuation (figure 2a) . the viral titer of all of the ns-laivs is at least 100-fold lower than that of wt in the nasal wash collected at 1 dpi. the ns1-73 laiv was the least attenuated in ferrets, and its replication was similar to that observed in mice. relative to the wt virus, the ns1-126 laiv showed 100-fold reduction in titer, and the nsd5 laiv was below the limit of detection (at least 1000fold reduction) at 3 dpi. sera from blood collected 6ae5 weeks after immunization was analyzed for the presence of neutralizing antibodies by micro-neutralization assays. the ns-laiv candidates all induced very strong neutralizing antibody responses (1280-5120) that were similar to the titer elicited by wt virus infection ( figure 2b ). the ferrets were challenged with 10 6 pfu of a ⁄ mexico ⁄ 4482 ⁄ 2009 (h1n1pdm) 6ae5 weeks post immunization. little disease or weight loss were observed in the naïve ferrets, and the ferrets immunized by infection with wt virus or the ns-laiv candidates didn't show any disease symptoms or weight loss. in contrast to the high titer of virus detected in the naïve ferrets through 5 dpc, the ns-laiv immunized ferrets had very low levels of a ⁄ mexico ⁄ 4482 ⁄ 2009 in their nasal washes at 1 dpc ( figure 2c ). the ferrets immunized with the ny1682 ns-laivs had $10 000-to 100 000-fold lower viral titers than did the naïve animals ( figure 2d ). in summary, the ns-laiv candidates dramatically inhibited initial replication of the h1n1pdm virus under stringent challenge conditions (10 6 pfu), and that the vaccinated animals rapidly cleared the infection (to below the limit of detection, by 3 dpc). our results demonstrate that all of the ns-laiv candidates are attenuated compared to the wt h1n1pdm virus, and the degree of attenuation is dependent on the specific ns mutation. ns1-73 was the least attenuated and does not represent a good vaccine candidate; whereas, nsd5 and ns1-126 were highly attenuated in both the mouse and ferret models. although they were markedly attenuated, they elicited strong neutralizing antibody responses and protected mice and ferrets from subsequent challenge. nsd5 has a subtle in-frame deletion (15 nt) that affects both the ns1 (residues 196-200) and nep (residues 39-43), and is analogous to a naturally attenuated variant of a normally highly pathogenic h5n1 virus (a ⁄ sw ⁄ fj ⁄ 03). 16 the analogous ns1 deletion in a ⁄ sw ⁄ fj ⁄ 03 (residues 191-195) was shown to reduce binding to host cleavage and polyadenylation specificity factor (cpsf), reduce ns1 protein stability, and enhance the type i ifn response of this h5n1 virus. 16 our study indicates that deletion of these 15nt in the ns vrna of the h1n1pdm also stimulates the host ifn response, specifically, ifn-ß, ifn-k1, ip10, and mxa (data not shown). the role of the deletion of residues 39-43 from nep has not been elucidated, but the induction of ifn and isgs by nsd5 was similar to, or slightly lower than, their induction by ns-126, suggesting that the nep mutation also has an attenuating effect that warrants future investigation. in summary, we have generated a panel of laivs directly from a swab specimen containing a new pandemic virus and analyzed their attenuation and immunogenicity in two animal models. our study demonstrates that nsd5 is a novel ns-laiv that could be used to create laivs for diverse influenza a viruses. this study also validates the use of ns-laiv candidates, which are not only highly attenuated, but they also elicit strong innate and adaptive immune responses, resulting in protection of mice from subsequent challenge with a lethal mouse-adapted variant of ny1682, and ferrets from challenge with a ⁄ mexico ⁄ 4482 ⁄ 2009 (h1n1pdm). currently, a total of approximately 300 million doses of inactivated influenza vaccine are being produced worldwide each year. one of the limitations in vaccine production is poor growth of human isolates in embryonated chicken eggs. this is essential to develop high yield seed viruses for large scale production of influenza vaccines. influenza a vaccine production utilizes high yield reassortants carrying ha and na genes from a wild type (wt) strain with generally 5-6 internal genes from the a ⁄ pr ⁄ 8 ⁄ 34 (pr8) strain, an highly egg adapted high growth donor strain. 1 influenza b vaccines, however, have been produced directly from wt strains, partly because no high yield donor analogous to pr8 has been identified. in recent years, reverse genetics has been used as an alternative means of developing high growth vaccine viruses. 2, 3 since in this plasmid-based technology, a 6:2 reassortant (six internal genes from a donor strain and two surface antigen genes from wild type strain) can be directly rescued, reverse genetics-derived reassortant viruses were expected to grow as efficiently as those derived from classical reassortment. however, reverse genetics reassortants have not produced the expected high growth for several reasons: (i) the 6:2 configuration is not always the best for virus yield, (ii) there is no process included for positive selection of adaptive mutants from quasispecies, and (iii) cell-derived viruses are not readily adapted to grow efficiently in eggs. our laboratory at new york medical college has been preparing b reassortants for several years by classical reassortment using b ⁄ lee ⁄ 40 as a donor. it has been possible to develop b reassortants, which produce higher virus yields than wt strains in eggs, and it was found that the np gene of b ⁄ lee ⁄ 40 was important in producing high yield b reassortants. 4 however, b ⁄ lee ⁄ 40 is inconsistent in providing high yield properties to b reassortants. in this study, in an attempt to find an alternative donor, we investigated the usefulness of b ⁄ panama ⁄ 45 ⁄ 90 for developing high yield b reassortants. as a wt strain, b ⁄ brisbane ⁄ 60 ⁄ 2008 was used, which is one of the recommended influenza b virus vaccine strains for the 2009 ⁄ 10 and 2010 ⁄ 11 seasons. we found that b ⁄ panama ⁄ 45 ⁄ 90 is a useful donor, and some of the resultant reassortants were considered as vaccine candidates. 5 b reassortant viruses were prepared by the classical reassortment method described by kilbourne. 1 the antiserum to b ⁄ panama ⁄ 45 ⁄ 90 hemagglutinin and neuraminidase (hana) was raised in this study by immunizing rabbits with hana isolated from b ⁄ panama ⁄ 45 ⁄ 90; purified igg was used for antibody selection. the yields of the reassortants and their corresponding parent viruses were assessed by hemagglutination assay. viral rna was extracted directly from the allantoic fluid and amplified by rt-pcr to produce cdna for analyzing the gene composition. restriction fragment length polymorphism (rflp) analyses were performed to determine the origin of each gene segment of the high yield reassortants. restriction enzyme sets for each gene segment are available upon request. in this study we investigated the usefulness of b ⁄ panama ⁄ 45 ⁄ 90 as a donor for transferring high yield phenotype. b ⁄ panama ⁄ 45 is a yamagata lineage strain with high growth phenotype (ha titer: 1024-2048). b ⁄ panama ⁄ 45 ⁄ 90 itself was a recommended b virus vaccine strain for 1989 ⁄ 90-1994 ⁄ 95 seasons. as a wt virus, a victoria lineage strain, b ⁄ brisbane ⁄ 60 ⁄ 2008, was used, which is a recommended b virus vaccine strain for use in the 2009 ⁄ 10 and 2010 ⁄ 11 seasons. reassortants were prepared according to classical reassortment protocol. after co-infection of b ⁄ panama ⁄ 45 ⁄ 90 and b ⁄ brisbane ⁄ 60 ⁄ 2008, progeny viruses carrying surface antigens (ha and na) of the vaccine strain were negatively selected by anti-b ⁄ panama hana antibodies, followed by passages without antibodies for positive selection of eggadapted viruses and finally limited dilution cloning. nymc bx-33, bx-33b, bx-33d, and r-3a are representative of resultant reassortants, which have significantly higher ha titers than the wt strain. the complete gene compositions of these reassortants were determined by rt-pcr ⁄ rflp analyses. as shown in table 1 , all of these reassortants contained the pb2 of b ⁄ panama ⁄ 45 ⁄ 90. other genes of b ⁄ panama ⁄ 45 ⁄ 90 (np of bx-33, m of bx-33b, and pb1 of bx-33d) may not be involved in the high virus yield, since no significant growth difference among these reassortants in eggs was found as assayed by hemagglutination test. accordingly, the pb2 of b ⁄ panama ⁄ 45 ⁄ 90 is considered to be the sole factor involved in the high yield phenotype donated to the vaccine strain. we previously found that the b ⁄ lee ⁄ 40 np gene was important in producing high yield b reassortants. it was of interest to examine whether b ⁄ lee ⁄ 40 np and b ⁄ panama pb2 could work together to produce even higher yields. to test this possibility, bx-33b (2:6 reassortant: pb2 and m genes from b ⁄ panama and the rest of the genes from b ⁄ brisbane) was selected and further reassorted with b ⁄ lee ⁄ 40. despite some difficulty in removing the na gene of b ⁄ lee ⁄ 40 (r-4c, 7b, 10b in table 2 ), by monitoring ha and na genes of resultant viruses after each antibody selection passage with anti b ⁄ lee ⁄ 40 hana antibodies, we were able to isolate and clone a triple reassortant, nymc bx-35, which contains the np gene from b ⁄ lee ⁄ 40 and pb2 and m genes from b ⁄ panama; the remaining genes are from b ⁄ brisbane ⁄ 60 ⁄ 2008 (table 2 ). in comparison with bx-33b, no significant growth enhancement (nor reduction) in eggs was found for bx-35 over that seen for bx-33b. nevertheless, bx-35 stably produces high virus yield and has been utilized as a seed virus for influenza b vaccine production for the 2010-2011 season by one or more vaccine manufacturers. there are contradictory reports 6-8 about the usefulness of reassortment for high yield influenza b viruses. however, we have been preparing b reassortants for several years by classical reassortment 1 using b ⁄ lee ⁄ 40 as a donor, and have been able to generate higher virus yield than wt strains. 4 in this study, we found that b ⁄ panama ⁄ 45 ⁄ 90 serves as an efficient donor in providing the high growth capacity to b ⁄ brisbane ⁄ 60 ⁄ 2008 (a recommended vaccine virus of victoria lineage for 2009 ⁄ 10-2010 ⁄ 11 seasons), and that the pb2 of b ⁄ panama ⁄ 45 ⁄ 90 is associated with the high yield phenotype. this particular strain from yamagata lineage might be useful to prepare high yield reassortants for other victoria lineage vaccine viruses. we noticed in this study that there may be segment incompatibilities between b ⁄ panama ⁄ 45 ⁄ 90 and b ⁄ brisbane ⁄ 60 ⁄ 2008. as shown in table 1 , the pa and ns genes of all the high yield reassortants examined are derived from wt, b ⁄ brisbane ⁄ 60 ⁄ 2008, not from the donor, b ⁄ panama ⁄ 45 ⁄ 90. this indicates that in this reassortment, the pa and ns genes are not replaceable with that of the donor to obtain high yield viruses. this degree of incompatibility might be common in b reassortment, resulting in low donor ⁄ wt reassortants, such as 2:6 and even 1:7 reassortants that we obtained in this study. if this is the case, reverse genetics based on 6:2 configuration may not result in generating high yield b reassortants unless a variety of donor ⁄ wt combinations are designed. one can speculate that in influenza b viruses, the surface glycoproteins (ha and na) and some of the internal proteins are functionally more closely related than in influenza a virus, as was seen in that pa and ns genes of b ⁄ brisbane ⁄ 60 ⁄ 2008 reassort together with the ha and na genes of the same parent (table 1 ). in our recent study on a reassortment between b ⁄ lee ⁄ 40 and b ⁄ panama ⁄ 45 ⁄ 90, it appeared that ha shapes overall gene constellations of the resultant reassortants, namely the reassortants tend to have more internal genes from the same parent of ha, no matter which parent's ha is selected by antibodies against the surface antigens of the other parent (data not shown). because of success in influenza a virus reassortment with pr8, it is generally believed that reassortant with 6:2 or 5:3 configuration is optimal for virus yield. this may be the case in most instances of influenza a reassortment, but is not necessarily so in b reassortment. as shown in this study, only a single donor gene is capable of improving the yield of vaccine strain by reassortment. influenza a ⁄ h1n1v has spread rapidly in all parts of the world in 2009 as a true pandemic. 1 epidemic events in russia occurred during the last week of september 2009 starting from far east region (yuzhno-sakhalinsk). kaliningrad (the western most russian city) was the second starting point of the epidemic. during october 2009 the epidemic spread over the whole russian territory. in a short period the new virus started to change genetically as it began to adapt to human populations during this pandemic (http://www.who.int; http://www.euroflu.org). in the period from may to december 2009, 1558 clinical samples (nasopharyngeal swabs and postmortem materials) of patients with influenza-like illness from different regions of russian federation were analyzed to confirm the diagnosis using real-time reverse transcription pcr (rrt-pcr). clinical nasopharyngeal swabs and bronchoalveolar lavage and post mortal fragments of trachea, lungs, bronchi, spleen from saint petersburg hospitals and 49 basic laboratories of federal influenza center were included in this study. all specimens were taken from patients with influenza-like illness or viral pneumonia. specimens were tested by rrt-pcr according to cdc protocols, i.e. using superscript iii platinum one-step qrt-pcr system (invitrogen) with primers and probes for infa, h1 seasonal, and h1sw (biosearch technologies). in addition, the test-systems 'amplisense influenza virus a ⁄ b-fl' and 'influenza virus a ⁄ h1-swine-fl' for pcr-detection, typing and subtyping of influenza viruses were also used. these test-systems are produced by central institute of epidemiology, moscow, russia and recommended by russian ministry of health as tests for influenza diagnosis. sequencing was carried out on an abi prism 3100-avant genetic analyzer (applied biosystems, usa) with bigdye terminator cycle sequencing kit. phylogenetic analysis was performed using programs vector nti 10.0 (invitrogen) and mega 4.1 (psu, usa) by maximum likelihood with the tim+i+g model for ha, and -hky+i+g model for na. evolutionary model was selected by akaike information criterion (aic) in model-test (posada, crandall, 1998). statistical reliability of tree branches was evaluated by bootstrap test (100 replications). immunohistochemical study was performed using novalink antibodies to ha and np with novocastra visualization system. influenza virus a ⁄ h1n1v rna was detected in 409 patients with severe form of influenza-like illness and 163 fatal cases. out of pcr-confirmed flu recovered cases 58% were patients under 14 years of age, 41% were aged 15-64 years, and 1% were older than 65 years. mean age of recovered patients was 15ae7 years (from 1 month to 77 years). viral rna in postmortem materials was detected mostly in lung tissue (86% of specimens) and trachea fragments (64%), and less commonly in spleen (17%). mean age of the deceased with confirmed flu (h1n1v) infection was 39ae3 years with age ranging from 7 months to 75 years. in 84% of fatal cases, influenza was complicated by viral or secondary bacterial pneumonia. median time from the onset of illness until death was 10 days. according to our data, 4% of patients died had diabetes, 4ae5% were obese, and 4% were pregnant women in the 2nd or 3rd trimester. ha and np were detected by immunohistochemical assay in lung tissue of dead patients with confirmed influenza virus a ⁄ h1n1v infection. ha and np was revealed in the endothelium of different sized blood vessels (capillaries and arterioles). these influenza virus proteins were also detected in some tissue macrophages apart from epithelium and endothelium. the localization of the two proteins was different: ha is mostly localized in cell membrane and cytoplasm, and np -mostly in the nucleus. here we present data on molecular genetic characteristics of 51 strains of pandemic virus, 40 strains obtained from clinical specimens, and 11 from post mortal ones isolated in the research institute of influenza. all the strains studied contain the s31n substitution in m2 protein, which indicates resistance to the adamantane antivirals, and have no h275y substitution in the neuraminidase, which indicates resistance to oseltamivir. the phylogenetic analysis showed that russian viruses were similar to influenza viruses a ⁄ texas ⁄ 05 ⁄ 2009 and a ⁄ california ⁄ 07 ⁄ 2009 (ha similarity 98ae9%). all russian viruses could be divided in two clusters: the first one includes viruses similar to the reference strain a ⁄ california ⁄ 07 ⁄ 2009, and the second one, which is the majority of viruses analyzed includes strains with substitutions ha s203t, na n248d, v106i, and ns i23v (figure 1 ). bootstrap support was 59. the isolates with ha s203t substitution can be classified in one of the five minor genome variants of a ⁄ h1n1v viruses found in the united states and mexico in 2009. 2 several viruses had strain-specific substitutions in antigenic sites sb and ca and the mutation d222g in ha receptor-binding site. the substitution of amino acid residue asp to gly at position 222 of ha was found in eight of eleven isolates (73%) from postmortem lung and trachea samples and two of forty isolates (5%) from nasopharyngeal swabs of patients with severe course of the disease. appearance of amino acid substitutions in the ha receptor-binding site (d187e and d222g ⁄ e) could be associated with influenza virus passaging on eggs. 3 five strains that contained g at position 222 of ha were isolated from post mortal specimens on mdck cells in this study, thereby excluding the possibility of substitution appearance hence to virus adaptation on eggs. in order to reveal genome changes in a(h1n1)v, strains isolated on the territory of russian federation during the 2009 pandemic, 67 full genome sequences from genbank, and research institute of influenza database were analyzed comparing two groups of viruses (isolated before and after 22 sept 2009). nine amino acid changes observed predominantly in late pandemic strains were found. five of them (s128p, s162n, d222g, v234i, v321i) reside in ha, two in na (i263v, n386k), two in pb2 (k340n, t588i), and one in pa (f35l). towards the end of the epidemic the viral population had demonstrated statistically certain rise in number of strains containing mutations in four genes. difference between groups was statistically significant (chisquare test, p = 0ae05). if v 2 > 3ae84, than difference between early and late strains is statistically significant. additionally fisher's test determined whether 'early strains' and 'late strains' differ significantly in the proportion of 'no mutation event' and 'mutation event' attributed to them in each particular position. all calculations were performed in fisher_tk freeware by vladimir belyaev similar to calcfisher (haseeb, 2003) fully described here (http://www.jstatsoft.org/v08/i21/paper). we have selected positions with statistically significant amino acid changes in late strains (p-value 0ae05). according to full genome analysis of influenza virus a ⁄ h1n1v 2009 strains, seven clades were distinguished, but the divergence between representatives of different clades remained small. 4 (figure 2 ). besides the strain a ⁄ perth ⁄ 12 ⁄ 2010 also contains substitution s181f in the same ha antigenic site. according to data obtained, the 2009 epidemic in russia was caused only by influenza virus a ⁄ h1n1v. unlike the previous epidemic periods when most severe influenza cases were registered among the children under 5 years and among elderly people aged over 65 years, the first wave of pandemic due to influenza virus a (h1n1)v resulted in increased level of mortality mainly among the people aged 18-53 years. though all pandemic viruses showed comparative genetic homogeneity, some evolutionary trends could be outlined. for clarification of the exact pathogenic role of mutation d222g in ha receptor binding site, further studies are necessary. full-genome analysis of influenza virus a ⁄ h1n1v strains circulating in the southern hemisphere in the new epidemic season 2010 revealed the phylogenetic subgroup distinguished by seven substitutions in inner proteins (pb2, pb1, np, ns1) and sa antigenic site of ha (n142d). the changes revealed could be caused by adaptation of the virus to an immunized human population. nasal and throat swabs (placed in 3 ml mem and frozen at )80°c until use for viral rna extraction and tissue culture inoculation) were collected from patients with febrile illness, i.e., >38ae0°c. samples were received from clinics in us embassies and us military laboratories located throughout the world since the initial who declaration of 2009 novel h1n1 outbreaks as a global pandemic on june 11, 2009. viral isolates were obtained from inoculating cultures of mdck cells with 0ae1-0ae2 ml viral suspensions collected in mem originated from patients after 2-7 days incubation. [10] [11] [12] [13] due to low viral titers in normal clinical samples, most of full viral genome sequences were derived from viral stocks obtained by tissue culturing passages (mdck, 1-2 times). viral rna was extracted from clarified supernatant fluid of nasal ⁄ throat swabs or mdck cultures using the 'charge-switch' rna extraction system based on the user manual protocol from the manufacturer (invitrogen inc., ca, usa). total rna was eluted into volume equal to original sample volume, i.e., 100 ll starting viral supernatant used to yield final 100 ll rna in molecular grade water (invotrogen inc.) and stored at )80°c until tested. generating ⁄ preparing overlapped cdnas for full genome coverage of 2009 novel h1n1 viruses by multiple rt-pcr amplifications the first step in the high-throughput sequencing pipeline for full influenza genome sequences was to establish a robust rt-pcr amplification scheme consisting 67 different rt-pcr primer pairs covering all 8 rna segments to ensure 100% amplification coverage of full viral genomes of all the incoming targeted viruses (houng, hs. 2011, submitted for publication). extracted viral rna (5 ll), derived from mostly mdck culturing stock or clinical sample containing sufficient viral load (>100 infectious units per ml) was added to primer-free rt-pcr total master mixture (10ae5 ll) for each virus followed by adding primer pair (2 ll, 3 pmole ⁄ ll per primer). rt-pcr was then performed: rt reaction through two hold-steps (50°c, 15 minutes and 94°c, 2 minutes); 35 cycling amplifications (94°c for 15 seconds, 53°c for 15 seconds, 72°c for 2ae5 minutes). specific cdna amplicons corresponding to each individual primer pair were routinely monitored and visualized by agarose gel electrophoresis. pooled 67 cdna products (2-5 lg) from each viral rt-pcr amplification run were used as sequencing substrates according to the roche 454 flx user manual and bulletins by incorporating adaptors containing individually multiplex identifier [mid]-key assigned to each individually pooled viral cdna. up to 12 different mid-keyed viral cdna were further pooled together to be clonally amplified on capture beads in water-in-oil emulsion micro-reactors (em amplifications), and pyrosequenced using one of two regions of a 40 · 75 mm picotiterplate. for each individual viral genome containing multiple assemblies (8 rna segments), we obtained sff file(s) containing raw sequencing reads from which nucleotide sequence data and phredlike quality scores were extracted. on average, 80ae0-90ae0% of 3-9 million mid-key specific nucleotides were extracted and mapped for consensus genome sequences. roche 454 gsmapper (v. 2.0 and 2.3) software was used to assemble all sequencing raw data and sff files into consensus sequences. new reference mapping projects were created to assemble each individually mid-keyed viral cdna into consensus viral sequences. one of the earliest 2009 h1n1 genomes of california origin, a ⁄ california ⁄ 04 ⁄ 2009(h1n1), deposited in genbank, was routinely employed as a reference genome sequence for most of gsmapper projects. the resultant consensus sequences obtained were further verified and validated through the ncbi annotation utility check and ultimately deposited to the ncbi influenza database, genbank. nucleotide sequences specific to each individual rna gene were aligned by the geneious pro 4.7.5 software (http://www.geneious.com). 14 trees were built based on the tamura-nei genetic distance model using the neighbor-joining method with no outgroup used via geneious pro 4.7.5. phylogenetic trees of the 2009 h1n1 genomes were constructed by importing fasta files containing specific concatenated target sequences of pb1, pb2, pa, ha, np, na, mp, and ns from each individual virus into the geneous pro software and going through the sequence assembling and tree building steps. high-throughput pyrosequencing of pooled 2009 novel h1n1 cdnas by roche 454 flx system up to 24 viral cdnas could be routinely sequenced to completion for 24 different full viral genomes from a single roche 454 flx picotiter plate by utilizing the combination of pico-titer plate's two distinct regions as well as 12 different mid-keyed adaptors. the 'shotgun' sequencing approach employed in this study is a feasible method to viral isolates (n) sequence multiple pooled 2009 h1n1 viral genomes. for each pyrosequencing experiment, approximately 800 000-1 000 000 passed key reads (single fragment per bead) were obtained that yielded readable nucleic acid sequences. among those close to a million passed key reads, only 500 000-600 000 passed key reads had an average sequencing read length of >220 bps, defined as 'long reads' (220 bps · 500 000 reads = total of 110 million bases of nucleic acid sequences) that were used to assemble into influenza genome sequences. mathematically, 110-130 million bases of raw sequencing data from each single roche 454 flx experiment would provide sufficient sequencing bases to cover 24 full genome sequences with approximate 3-400 · of sequencing depth coverage of influenza a with average genome size of 13 500 bps for the total of eight segmented rnas. so far, more than 100 full 2009 h1n1 genomes sampled worldwide have been successfully sequenced and deposited in the ncbi database by division of viral diseases, walter reed army institute of research (wrair). the bioinformatics derived from unique viral genome sequences generated from this study based on constant rt-pcr amplification scheme and identical roche 454 pyrosequencing protocols provide a reliable data set in predicting the evolutionary patterns of pandemic viruses. wrair received clinical samples from us embassies and military personnel throughout the world since the initial who announcement of 2009 novel h1n1 outbreaks. nearly equal distributions of sequenced viruses derived from three broadly categorized geographic regions, north america, central ⁄ south america, and asia ⁄ europe ⁄ africa (data not shown). besides the geographic distribution pattern of viral isolates, figure 1 displays the viral isolation time lines of all the sequenced viruses reflecting two peaks that coincided with two waves of 2009 pandemic infections, early-mid summer and fall of 2009. 15 phylogenetic trees of the eight influenza a segments of all sequenced viruses were tentatively generated. it was found that the substitution frequencies per site for the ha, na, and ns genes are at much higher rate than the other five genes, pb2, pb1, pa, np, and mp genes (data not shown). the observed higher genetic variations for ha and na genes of 2009 h1n1 are consistent with the historical genomic and epidemiological dynamics data of human influenza a revealing higher temporal fluctuations in ha and na genes. [16] [17] [18] [19] analysis of full influenza genomes containing concatenated eight complete rna segments revealed the existence of two distinctive genetic clades in circulation since the beginning of 2009 pandemic, as shown in fig-ure 2. it is noteworthy that all viruses of mexico and california origins (clade 1 shown at the top of figure 2 ) were isolated at the beginning of 2009 pandemic prior to the isolation of all other viruses belonging to the second genetic clade 2. 18, 20 discussion during the past decade, the advance of dna sequencing technology, such as development of ngs, in making full viral genome sequences readily available have enabled study of far broader and more detailed aspects of evolutionary change for any new emergent infectious pathogen. the massive sequencing capacity of roche 454 flx system allows simultaneously process and sequence millions of individual cdna molecules, in contrast to processing and sequencing individual cdna fragments by conventional sanger sequencing method. within a short period of few months since the beginning of the 2009 pandemics, wrair accomplished large number of representative 2009 h1n1 full genomes of worldwide origins via roche 454 flx system. sequencing data derived from this study illustrates a much higher genetic variation rate for ha and na genes of 2009 h1n1 that is compatible to the higher temporal fluctuation rate for ha and na genes of seasonal influenza a derived from decades of intensive monitoring and comparison studies and analyses. [16] [17] [18] [19] following the mexican and us reported cases, confirmed outbreaks of 2009 swine h1n1 rapidly proliferated and spread throughout europe, asia, africa, and south america, most probably via global airline travel. 5, 6 it seemed that new cases in the us and most cases throughout the world had been clinically mild relative to the initial reported cases in mexico. [21] [22] [23] [24] here we demonstrate through the phylogenetic relationship of sequenced 2009 h1n1 full genomes that the clinical isolates could be divided into two different clades of viruses, i.e., the clade 1 genetic group contains only viruses isolated at the beginning (march ⁄ april 2009, mexico and california) of 2009 pandemics and the rest of other viruses all belong to the 2nd genetic group, clade 2. thus, it's likely that the currently circulating 2009 h1n1 of clade 2 causing worldwide infections is genetically different from the initial 2009 h1n1 isolates that caused the early infections in mexico and california. 18, 20 introduction a pandemic influenza virus (2009 h1n1) was recently introduced into the human population. the hemagglutinin (ha) gene of 2009 h1n1 is derived from 'classical swine h1n1' virus, which likely shares a common progenitor strain with the human h1n1 virus that caused the pandemic in 1918. 1 since antigenic changes of influenza virus ha occur more slowly in swine than in humans, 2 we hypothesized that 2009 h1n1 might still retain an antigenic structure similar to that of 1918 h1n1 or the early isolates of its descendants. in this study, we compared ha antigenic structures of 2009 h1n1 and human h1n1 viruses by a molecular modeling approach to demonstrate the existence of shared epi-topes for neutralizing antibodies. we found that has of 2009 h1n1 and the 1918 h1n1 virus shared a significant number of amino acid residues in known antigenic regions. from this observation, we hypothesize that the 2009 h1n1 ha antigenic sites will be targeted by antibody-mediated selection pressure in humans in the near future. we further discuss possible directions of antigenic changes in the evolutionary process of 2009 h1n1. sequence data of ha genes modeller 9v6 was used for homology modeling of ha structures. after 100 models of the ha trimer were generated, the model was chosen by a combination of the mod-eller objective function value and the discrete optimized protein energy statistical potential score. after addition of hydrogen atoms, the model was refined by energy minimization with the minimization protocols in the accelrys discovery studio 2.1 software package using a charmm force field. steepest descent followed by conjugate gradient minimizations was carried out until the root mean square gradient was less than or equal to 0ae01 kcal ⁄ mol ⁄ a. the generalized born implicit solvent model was used to model the effects of solvation. the ha model was finally evaluated by using procheck, whatcheck, and verify-3d. custom-made programs were developed with the ruby language and used for investigating the numbers of potential n-glycosylation sites and candidate codons (cand1) in ha sequences. it is known that the h1 ha molecules have four distinct antigenic sites: sa, sb, ca, and cb. 3, 4 as a result, these sites consist of the most variable amino acids in the ha molecule of the seasonal human h1n1 viruses that have been subjected to antibody-mediated immune pressure since its emergence in 1918, although it was absent in humans from 1957 to 1976. to investigate the structures of these antigenic sites of 2009 h1n1, 3d structures of the ha molecules of sc1918, the recent seasonal human h1n1 virus (br2007), and 2009 h1n1 (ca2009) were constructed by a homology modeling approach, and compared by mapping all the amino acid residues that were distinct from those of sc1918 ha (data not shown). we found that most of these antigenic sites of br2007 ha predominantly contained altered amino acid residues if compared with sc1918. by contrast, amino acid residues at these positions were relatively conserved in ca2009 ha when compared with sc1918 ha. notably, the sa and sb sites, which contain many amino acids involved in neutralizing epitopes near the receptor binding pockets, remain almost intact ( table 1 ), suggesting that antibodies raised by natural infection with sc1918 or its antigenically related descendant viruses play a role in specific immunity against ca2009. these observations lead us to hypothesize that such antigenic sites involving the conserved amino acids will soon be targeted by antibody-mediated selection pressure in the human population. based on this hypothesis, we speculated that 2009 h1n1 would undergo patterns of amino acid substitutions in ha similar to those seen in seasonal human h1n1 viruses during its epidemic period (i.e. those that have been substituted since 1934) (figure 1) . we then predicted possible amino acid substitutions of 2009 h1n1 from the sequence similarity of the antigenic sites. for example, both sc1918 and ca2009 had an asn residue at position 142 in the sa site. for sc1918, the residue at this position has altered from asn to lys since 1942. combining these two facts, it seems reasonable to hypothesize that ca2009 will also undergo an amino acid substitution from asn to lys at position 142 in the future. interestingly, we found that some of the recent variants of the 2009 h1n1 virus have indeed undergone substitutions identical to those predicted in figure 1 . it is important to monitor whether such variants will be selected and survive in sustained circulation in humans. next, we analyzed the acquisition of potential n-glycosylation sites associated with antigenic changes. previously, we reported that cand1 sites, a set of three codons that require single nucleotide substitution to produce n-glycosylation sequons, were important motifs to rapidly acquire n-glycosylation sequons. 5 therefore, we investigated the number and location of potential n-glycosylation sites and cand1 sites in 2009 h1n1 ha. we found that ca2009 also had a single n-glycosylation sequon at the same position in the globular head region of ha, and lacked the multiple n-glycosylations that have been observed in the antigenic changes of the human h1n1 virus during the early epidemic of this virus. we also found that ca2009 ha possessed three cand1 sites that were present at the same position in sc1918 ha (positions of the first asn residue, 177, 179, and 184). of these, the cand1 sites with positions at 177 and 179 had actually become potential n-glycosylation sites in human h1n1 viruses. this result suggests the likelihood of additional n-glycosylation at these sites during future antigenic changes of 2009 h1n1 ha. notably, some of the recent 2009 h1n1 variants (as of march 24, 2010) have an additional n-glycosylation sequon at position 179, where the 1918 h1n1 virus readily acquired an n-glycosylation site during its circulation. the present study suggests that the antigenic structure of 2009 h1n1 ha is similar, at least in part, to that of the 1918 h1n1 ha. the 2009 and 1918 h1n1 has share unique three-codon motifs that are important to readily acquire n-glycosylation sequons in their globular head region. based on these similarities, we predicted possible amino acid substitutions that might be associated with future antigenic changes of 2009 h1n1, and confirmed that such substitutions occurred in some of the recent variants of this virus. the present study provides an insight into likely future antigenic changes in the evolutionary process of 2009 h1n1 in the human population. influenza viruses are classified into three types, a, b, and c, based upon the antigenic properties of nucleoproteins and matrix proteins. influenza a virus infects a wide range of hosts, including human, bird, swine, equine, and marine mammal species, while influenza b and c are less pathogenic than influenza a and are mainly found in humans, although there is evidence that they can also infect other species. influenza a has evolved in association with its various hosts on different continents for extended periods of time. 1 to survive as a successful pathogen, the influenza viruses have developed a number of mechanisms, including antigenic mutation and genome reassortment, to continuously evolve and evade the surveillance of the host immune systems. antigenic and genetic analyses have provided important insights into the molecular dynamics of influenza virus evolution. 2 however, a comprehensive understanding of influenza viral genetic divergence and diversity remains lacking. neuraminidase (na) is a major surface glycoprotein of influenza a and b, but is absent in influenza c. it plays a key role in virus replication through removing sialic acids from the surface of the host cell and releasing newly formed virions. influenza a viral na genes are classified into nine subtypes (na1-na9) based upon their antigenic properties, while na genes of influenza b are not classified into subtypes. furthermore, most na subtypes of influenza a have evolved into distinct lineages and sub-lineages, which correspond to specific hosts or geographical locations. in this study, we conducted large-scale analyses of influenza na sequences in order to infer their evolution and to identify lineages (or sub-lineages) of influenza a viruses. a total of 11 454 na sequences that excluded laboratory recombinant sequences were downloaded from genbank. sequences were aligned with muscle and mafft. the alignments were adjusted manually using translatorx, 3 based upon corresponding protein sequences. phylogenetic analyses were conducted using the maximum-likelihood (ml) method in raxml. 4 a set of perl scripts were written by us to facilitate this computational analysis. lineages and sub-lineages were determined based on the topology of the ml trees. additional information such as hosts, geographical regions, and circulation years were also considered in the classification. we used the same lineage nomenclature as described in, 5 but with the following modifications: a single digit is used to represent one of the nine subtypes and a letter is used to represent a lineage; a sub-lineage is also represented using a digit; a dot is used to separate a lineage and a sublineage. for example, 2a.2 means na2 subtype, lineage a, and sub-lineage 2. the time of most recent common ancestor (tmrca) was estimated using the bayesian mcmc method in beast. 6 in all cases, we employed the gtr + u4 nucleotide substitution model, in which the first and second codon positions are allowed different rates relative to the third codon position. all data sets were analyzed under a relaxed molecular clock and the bayesian skyline population coalescent prior. the maximum clade credibility (mcc) tree across all plausible trees was computed from the beast trees using the treeannotator program, with the first 10% trees removed as burn-in. phylogenetic analysis based upon na sequences revealed two large groups corresponding to influenza a and b, respectively ( figure 1a ). within influenza a, two subgroups were found, one consisting of na1, na4, na5, and na8 and the other consisting of the remaining five subtypes. subtype na9 was found to be a sister subtype of na6, na1 being a sister subtype of na4, and na8 a sister subtype of na5. finally, each na subtype forms a distinct cluster, indicating its genetic uniqueness. influenza a and b viral na were estimated to have diverged around 2641 years ago ( figure 1b ). however, it had large 95% hpd values which ranged from 1259 years to 4299 years ago. the na subtypes of influenza a diverged from more than 1000 to several hundred years ago. the time of most recent common ancestor (tmrca) of each subtype of influenza a virus was generally recent and ranged from the calendar years 1767 to 1928 (figure 1b ). in addition, the tmrca for influenza b viral na was dated back to 1936. a total of 23 lineages were identified in influenza a (table 1) . three lineages, 1a, 1b, and 1c, were identified for na1 based upon the tree topology. linage 1a originated from avian viruses and was further divided into 5 sub-lineages: 1a.0, 1a.1, 1a.2, 1a.3, and 1a.4. linage 1b consists of north american swine influenza viruses whereas 1c is a human lineage. two large lineages, 2a and 2b, were identified in na2. lineage 2a is a human-specific lineage. interestingly, five major swine clades were observed within this lineage. lineage 2b is an avian-specific lineage, and consists of 3 sub-lineages, 2b.0, 2b.1, and 2b.2. three lineages were found in na3. lineage 3a was found in north american avian, 3b in eurasian ⁄ oceanian avian, and 3c also in avian, but it does not show any geographical pattern. for na4, na5, and na6, each was classified into 2 lineages, one found in north american avian (4a, 5a, 6a) and the other in eurasian ⁄ oceanian avian (4b, 5b, 6b). three lineages identified respectively in na7 and na8 are north american avian (7a, 8a), equine (7b, 8b), and eurasian avian (7c, 8c). na9 was also found to have 3 lineages: north american avian (9a), eurasian ⁄ oceanian avian i (9b), and eurassian ⁄ oceanian avian ii (9c), respectively. in this study, we conducted large-scale phylogeny and evolutionary analyses using influenza viral na sequences. the results showed that divergence between influenza a and b viruses occurred earlier than between any influenza a subtypes. this observation was consistent with previous findings based upon phylogenetic analysis of the ha gene, one of the most important genes related to host infection. 7 within influenza a, two sub-groups were found, one consisting of na1, 4, 5, and 8 and the other consisting of the rest of five subtypes (na2, 3, 6, 7, 9) . this observation does not agree with the result described by liu et al., 8 where na subtypes 1, 3, 4, 5, and 8 formed one group and the remaining four subtypes (na2, 6, 7, and 9) formed the other group. this difference is apparently caused by the fact that an outgroup was not used in their phylogenetic analyses. in the present study, both influenza a and b viral na sequences were included in the analysis. high bootstrap values were obtained for major groups, indicating that the inferred evolutionary relationship should be highly reliable. classification and designation of the lineages and sublineages within the influenza a virus are essential for studies of viral evolution, ecology, and epidemiology. a total of 23 lineages were identified within nine influenza a viral na subtypes and with the majority of the identified lineages found to be host or geographic specific or both. our results demonstrated a comprehensive view for the evolution of na genes and provided a framework for the inference of evolutionary history of pandemic viruses and for further exploring of viral circulations in multiple hosts. for example, the global pandemics of human h1n1 in 1918, h2n2 in 1957, the pandemic of human h3n2 virus in 1968, 9 the crisis of h5n1 hpai in hong kong in 1997, 10 and swine-origin h1n1 influenza in 2009, 11 all can be mapped onto the lineages and sub-lineages identified in this study. such information will facilitate not only identification of known genetic origins but also early detection of novel influenza a viruses. influenza viruses constantly evolve to avoid the human immune pressure in the process of antigenic drift. 1 through sequencing of viral genomes, the rates and direction of virus evolution can be observed. moreover, comparison of protein sequences allows us to determine amino acid substitutions that are related to immune pressure and antigenic drift. the creation of global influenza genetic databases, along with concurrent development of analytical tools, allows the comparison of multiple influenza virus strains. 2 the main aim of this study was to perform antigenic and genetic comparison of pandemic influenza viruses (h1n1) isolated during the 2009-2010 pandemic in ukraine and in other countries. nasopharyngeal swabs and autopsy materials collected from infected patients were received from the areas of ukraine. in addition, field isolates of influenza viruses from the 2009 ⁄ 10 season and strain specific serum were used for identification by hemagglutinin inhibition assay. influenza viruses were identified and subtyped using real-time rt-pcr analyses using cdc primers and adopted protocols. 3 sequencing was performed in two world health organization (who) influenza collaboration centers (centers for disease control and prevention, atlanta and national institute for medical research, london). hemagglutinin inhibition assay was conducted using chicken and guinea pig red blood cells following standard who protocols. 4 the all 6 ukrainian isolates of influenza viruses, which were isolated in ukraine during august-november 2009, were identified as a ⁄ california the phylogenetic analyses confirmed the evolutionary relationship between ukrainian isolates and viruses from other countries, which were isolated during the first wave of the pandemic. high genetic and antigenic conservation of pandemic influenza viruses from ukraine and other countries also were demonstrated. considering that the emergence of the novel pandemic influenza strain occurred in countries of northern hemisphere during summer, it was very interesting and significant tracking the dynamics of genetic changes in influenza viruses, which were isolated at the beginning of epidemic and those isolated during the rise of the epidemic in ukraine. influenza a virus causes moderate to severe epidemics annually and catastrophic pandemics sporadically. due to the evasiveness of the influenza virus and the nature of its genome (eight single-stranded and negative-sense rna segments), it is essential to understand the evolution of this important pathogen. 1 influenza virus evolves by two major mechanisms: mutation and reassortment. antigenic and genetic analyses have revealed partially the molecular dynamics of influenza virus evolution. 2,3 however, important questions, such as how many genotypes in the influenza a virus, remain unanswered. one of the major issues pertaining to this genotyping problem is how many lineages or sub-lineages can be determined for a subtype and according to what criteria. because of the unique structure of the influenza a viral genome, the computational genotyping methods developed for other viruses cannot be applied to the influenza virus. constructing phylogenetic trees is a powerful technique for the identification of evolutionary groupings (i.e., lineages ⁄ clades). however, for large trees, it is hard to determine how many lineages and the boundaries for each lineage. in this regard, multivariate analysis methods, such as multidimensional scaling (mds) and model-based hierarchical clustering, both taking advantage of dimension reduction and visualization, can complement conventional phylogenetic methods. 4 hemagglutinin (ha), the fastest evolving segment, is recognized as the most important gene in the influenza virus that plays a key role in viral pathogenesis. however, we have only limited knowledge of lineages and sub-lineages occurring in the hemagglutinin (ha) gene of influenza a virus, 5 although much effort has been made in assigning clades or sub-clades in highly pathogenic avian influenza (hpai) virus ha. 6 in this study, both model-based hierarchical clustering and phylogenetic methods were used for sequence analysis. one objective for this study is to explore and develop a more accurate lineage approach for further comprehensive influenza lineage and genotype analyses. a total of 11 821 hemagglutinin (ha) sequences (approximately 1778 nucleotides long), excluding laboratory recombinant sequences, were downloaded from genbank as of march, 2010. 7 sequences were aligned with muscle 8 and mafft. 9 the genetic distance matrix of all pairwise sequences was computed using the k2p model under mega 5.0. we then used the distance matrix as input to the cmdscale module in r 2.5.1 for the mds analysis. the principle coordinates resulting from mds were used for the model-based hierarchical clustering analysis, again in r 2.5.1 (the r foundation. available at: http://www.r-project.org/). the bayesian information criterion (bic) values were computed based upon ten different statistical data models -eii, vii, eei, evi, vei, vvi, eee, eev, vev, and vvv. the highest bic value was used to determine the number of clusters in the given sequence data. phylogenetic analysis was conducted using maximumlikelihood (ml) in raxml. 10 raxml uses rapid algorithms for bootstrap and maximum likelihood searches and is considered one of the fastest and most accurate phylogeny programs for large-scale sequence analysis. all the analyses were conducted on the supercomputer cluster (holland computing center, http://hcc.unl.edu/main/index.php). the trees were visualized in figtree (version 1.3.1) . lineages and sub-lineages were determined based on both the topology of the ml trees and model-based clustering results. additional information such as hosts, geographical regions, and circulation years were also considered in the classification. we used the same lineage nomenclature as described in, 11 with the following modifications: lineage analysis was conducted for each ha subtype, which agrees with the convention of influenza virologists that 16 ha subtypes were identified in influenza a virus; ha lineages are represented with digits and letters, where the digit(s) represent one of the 16 subtypes and a letter represents a lineage; here, we present sub-lineages or sub-sub-lineages also in digits, with smaller numbers representing earlier lineages or sub-lineages within the same subtype (e.g., lineage 1 occurs earlier than lineage 2); the digit 0 is used to indicate inclusion of ancestral viruses in a lineage (or sub-lineage); a dot is used to separate lineages, sub-lineages, and sub-sub-lineages. for example, 1a.1ae2 means ha1 subtype, lineage a, and sub-lineage 1, and sub-sub-lineage 2. the sub-lineage level can be extended as necessary. the model-based clustering method corroborates commonly used phylogenetic methods in lineage and sub-lineage assignment. here we use the h12 subtype as an example to show the lineage and sub-lineage assignment. the bayesian information criterion (bic) reaches its maximum when the number of clusters for h12 equals 3, regardless of which mode we choose ( figure 1a ). therefore, based on bic, the optimal number of clusters for the h12 subtype is 3. as a result, a total of 3 clusters based upon the vvv model were identified ( figure 1b) . a significant correlation was found in lineage assignments by the phylogenetic method and the model-based hierarchical clustering method ( figure 1b,c) . lineages 12a and 12b were identified for h12, which correspond to north american avian and eurasian avian, respectively. lineage 12a was further divided into 2 sub-lineages, 12a.0, 12a.1, where 12a.0 is the ancestral sub-lineage in 12a. based on both model-based hierarchical clustering and phylogenetic analyses, a total of 41 distinct lineages were identified among 16 subtypes, averaging out to be 2ae56 lineages per subtype ( table 1 ). the majority of the identified lineages were found to be host or geographic specific or both. for example, three lineages, 1a, 1b, and 1c, were identified for ha1. lineage 1a was further divided into two sub-lineages, 1a.0 and 1a.1. the 1a.0 is swine-specific, whereas 1a.1 is a human pandemic 2009 h1n1 sub-line how to accurately identify an evolutionary lineage of influenza a viruses is challenging. one commonly used approach is molecular phylogeny, where phylogenetic trees are constructed, and the tree topology is used for lineage determination. here, we used a bayesian model-based clustering method, along with phylogenetic methods, to decide lineages and sub-lineages of influenza a viruses based upon sequence data. the results demonstrated that the modelbased clustering method corroborates phylogenetic methods and increases the accuracy of lineage assignment. one salient feature of this study is its large-scale analysis of all available influenza a hemagglutinin sequences. a total of 41 distinct lineages and 81 sub-lineages were classified; the majority of them were found to be host or geographic specific. this observation agrees largely with previous findings. 5 we are conducting further analyses of other influenza a segments 12 and expect to identify their lineages and create a comprehensive genotypes database for all influenza a viruses. such information will allow us to detect the genetic origin of newly found viruses, track their genetic changes, and identify potential genome reassortments. a hierarchical nomenclature system has been proposed and adopted for hpai ha clades and sub-clades by who influenza surveillance centers. 6 wan et al. also proposed a hierarchical approach for influenza a viral genotypes system. 13 the work presented here is one of the first steps towards the development of a nomenclature system for influenza a virus lineages (at the segment level) and genotypes (at the genome level). whether the naming system will be accepted and used by the influenza research community is more challenging than the lineage analysis itself. identification of the genetic origins of influenza a viruses will enhance our understanding the evolution and adaptation mechanisms of influenza viruses. the phylogenetic analysis is the traditional approach to identify the influenza progenitor. first, the nucleotide sequences are aligned using multiple sequence alignment methods, such as clustalw, 3 muscle, 4 and t-coffee. 5 second, phylogenetic analysis is performed on these aligned sequences to infer their evolutionary relationship using neighbor-joining (nj), 6 likelihood, or bayesian inference. 8 bootstrap analyses or computation of posterior probability are usually applied to estimate the phylogenetic uncertainty. however, this phylogenetic analysis is time consuming due to intensive computations in multiple sequence alignments and phylogenetic inferences. it is difficult to perform an analysis using this method on a large dataset, for instance, with more than 1000 taxa, as is the common case for influenza studies. alternatively, blast 9 is applied to identify the prototype genes in the database. blast determines a similarity by identifying initial short matches and starting local alignments. since influenza viral sequences have very high similarities, especially for most conserved regions, blast usually generates a large number of outputs, which will not be helpful for progenitor identification. since blast is a local sequence alignment, the results from blast may not reflect the global evolutionary information between the sequences. the blast scores cannot be used to define the evolutionary relations between viruses, especially in the context of the entire genetic pool. recently, we have developed a distance measurement method, complete composition vector (ccv), that can calculate genetic distance between influenza a viruses without performing multiple sequence alignments. 10, 11 we also adapted the minimum spanning tree (mst) clustering algorithm for influenza reassortment identification. 12 the application of this approach in the analyses of pb2 genes of influenza a virus showed that the integration of ccv and mst allows us to identify the potential progenitor genes rapidly and effectively. based on these results, here we develop a webserver called ipminer for influenza progenitor identification. ipminer can identify potential progenitors for a query sequence against all public influenza datasets within a few minutes. in order to improve the computing efficiency, 31 distance matrices were pre-computed by ccv, and they include 16 for ha (h1 to 16), 9 for na (n1 to n9), and one for each of the internal gene segments (pb2, pb1, pa, np, ns, and mp). these 31 pre-computed matrices will be updated weekly. ipminer just needs to compute the query matrices for a query sequence and sequences in the database. the standalone ccv program is also available at http://sysbio.cvm.msstate.edu/ipminer. in order to identify the influenza progenitor genes, ipminer first integrates the query matrix and a corresponding pre-computed matrix into a full distance matrix, which is then clustered by mst clustering algorithm. we adapted the threshold we measured previously in mst, u + nr, where u is the average distance and r is the standard deviation of a cluster. 12 as a result, mst will generate a hierarchical structure for the clusters. in each cluster, we will randomly select 20 viruses or 10% of the cluster size if this cluster has more than 200 viruses. ipminer will return the viruses with the smallest distances when the search reaches to the lowest level (the largest n) in this hierarchical structure. our analyses have shown that the level 5 has generally yielded good results for influenza a viruses. to visualize the overall mst structure, ipminer applies multi-dimensional scaling (mds) method to project all the viruses in the genetic pool onto a two dimensional graph, and the precursor viruses are marked in different shapes ( figure 1 ). the users can select other prototype viruses from the graph for further phylogenetic analyses. a single job with one query sequence takes <2 min. the genbank identifiers and associated genetic distances and sequence identities are displayed. the users can download the sequences for the identified precursor viruses as well as those from the prototypes viruses. in addition, for the users' convenience, ipminer generates a phylogenetic tree using nj method implemented in phylip 13 to illustrate the phylogenetic relationship among the query sequence(s), the identified progenitors, and the selected prototypes viruses. the programs in this solution package are written in java. the shell scripts are written in korn shell script in order to achieve high performance. cascading style sheets (css) are used for a consistent look across the pages. this also enables to change the overall design just by replacing the css definition file. php has been used as server side scripting and is written in java. in order to achieve high performance for computing in a genomic scale, we apply hash function or a binary tree, which enables that the precursor identification has a time complexity of o(n). for single queries, the users can visualize the results online. for batch queries of multiple sequences, the results will be sent to the users by e-mail. ipminer has been tested on microsoft internet explorer, mozilla firefox, and safari. the users need javascript to obtain full function of ipminer server. the webserver is available at http://sysbio.cvm.msstate.edu/ipminer. in summary, ipminer webserver has three major computational features for influenza progenitor identification: (i) it calculates the genetic distances through ccv and identifies the viruses with the shortest ccv distances against the query virus to be the progenitor genes; (ii) it projects influenza viruses onto a two dimensional map, which illustrates the global relationship between the progenitor genes and other viruses in the genetic pool; and (iii) it performs phylogenetic analyses between the query virus, the identified progenitor genes, and other selected prototype viruses. ipminer provides a user friendly web service for influenza progenitor identification in real time. the gisaid initiative offers an alternative to current public-domain database models in response to growing needs of the global influenza community for the sharing of genetic sequence and associated epidemiological and clinical data of all influenza strains. gisaid's publicly accessible epifluô database is governed by a unique sharing mechanism that protects the rights of the submitter, while permitting ongoing research as well as the development of medical interventions, such as drugs and vaccines. for the gisaid initiative, the max planck institute for informatics (mpii) saarbrücken, germany, has developed a web portal that is accessible at http://www.gisaid.org featuring the gisaid epifluô database that offers a unique collection of nucleotide sequence and other relevant data on influenza viruses. the database is based on software by oracle and the dante ò system by a3systems gmbh, germany. extensive metadata are also collected for most isolates. the database provides features for searching, filtering specific datasets for download, and user friendly upload functionality. to uphold gisaid's unique sharing mechanism, all users must positively identify themselves. while access is free of charge, all users agree that they will not attach any restrictions on the data, but will acknowledge both the originator of the specimen and the submitter of the data, and seek to undertake to collaborate with the submitter. all uploaded sequence data are submitted to rigorous curation by the friedrich-loeffler-institute for animal health (fli), germany. the database has been live since september 14, 2009. among its contributors are all five who collaborating centers for influenza who routinely contribute data in addition to using the epifluô database for their semiannual vaccine strain selection. to provide a complete picture of data, all data available in the public domain is routinely imported. as of october 8, 2010, the rapidly growing gisaid dataset comprises 166 989 nucleotide sequences (from 48 779 isolates) with 28 949 (from 11 857 isolates) uniquely submitted to this database. software development is underway to continually extend the spectrum of available data analysis tools. the intergovernmental process of the 62nd world health assembly specifically mentions gisaid as a publicly available database for depositing virus sequence data. starting in 2011, germany's federal ministry of food, agriculture and consumer protection will be the long-term host of the gisaid platform. the mpii will continue to develop the portal and database software and enable gisaid to act as a catalyst for the development of advanced bioinformatics software connected directly to the database. gisaid has become an indispensible resource for the international scientific community on influenza. the consortium will expand its activities and offers to catalyze research and development on a wide variety of issues pertaining to risk analysis, drug development, and therapy of influenza. options for the control of influenza vii ª 2011 blackwell publishing ltd, influenza and other respiratory viruses, 5 (suppl. 1), 416-424 the pandemic h1n1 virus emerged in 2009 and spread rapidly throughout the world, principally affecting children and young adults. as this virus is new to the human population, it is important to determine if these influenza infections are more commonly associated with other respiratory pathogens compared to previously circulating influenza strains. co-infecting respiratory viruses may cause increased morbidity in individuals with pandemic h1n1, and may also be unwanted contaminants in influenza vaccines if original clinical samples containing these adventitious viruses are used to directly inoculate certified cell lines for vaccine production. to examine this issue, stored rna from 300 original clinical samples (nasal swabs, nasal aspirates, throat swabs) from australian and new zealand subjects that were collected in 2009 that were positive for pandemic h1n1 and 300 samples collected in 2008 that were positive for seasonal influenza by real time pcr assay (using the cdc, usa kits), were subjected to a resplex ii -panel version 2.0 (qiagen) pathogen screen. the resplex ii assay detects 18 common respiratory viruses, such as respiratory syncytial viruses (rsv a, b), influenza a and b viruses, parainfluenza viruses (piv1-4), human metapneumo-viruses (hmpv), coxsackieviruses ⁄ echovirus (cvev), rhinoviruses (rhv), adenoviruses (adv b, e), coronaviruses (nl63, hku1, 229e, oc43), and bocaviruses. resplex ii uses a combination of multiplex rt-pcr, hybridization of pcr onto target specific beads followed by detection using luminex-xmap technology. original clinical samples were received at the center from who national influenza centers, who influenza collaborating centers, and other regional laboratories and hospitals from australia, new zealand, and the asia ⁄ pacific region. most samples were from australia and new zealand. these samples consisted of nasal swabs, nasopharyngeal swabs, nasal washes, throat washes, and throat swabs. all samples were stored at )80°c until rna was extracted. rna was extracted from 200 ll of clinical sample using either the magnapure extraction system (roche, australia) or the qiaxtractor system (qiagen, australia) according to the manufacturer's recommendations with an elution volume of 90 ll and stored at )80°c until used. a 5 ll aliquot of rna was used to amplify the selected influenza virus gene using specific primers and probes as supplied by cdc (atlanta, usa) 1 along with super-script iii platinum one-step rt-pcr reagents (invitrogen, australia). real time pcr detection was performed on a 7500 fast system with sds software (applied biosystems, ca, usa). a cut off of a cycle threshold (c t ) of 35 or below was considered positive. resplex ii panel ver2.0 detection the qiagen molecular differential detection (mdd) system was used, which combines qiaplex amplification (multiplex rt-pcr) with detection on the liquichip 200 workstation (luminex's xmap microsphere based multiplexing system) and qiaplex mdd software according to the manufacturer's instructions. a low level cutoff was used (150) to obtain maximum sensitivity. from the 300 clinical specimens that were positive for influenza from 2008 by real time pcr, there were 18 (6%) a(h1n1) seasonal influenza viruses, 106 (35ae3%) a(h3n2) viruses, 174 (58%) b viruses, and 2 (0ae7%) viruses which were influenza a positive, but could not be typed. clinical samples from 2009 selected to study were all influenza a(h1n1) pandemic 2009 positive by real time pcr. detection of influenza virus in respiratory samples was much lower with the resplex ii assay (using a low cut off of 150 units) for pandemic influenza a virus (160 ⁄ 320; sensitivity 53ae3%) and to a lesser extent for seasonal influenza a (116 ⁄ 126; sensitivity of 92ae1%) and b viruses (161 ⁄ 174; sensitivity of 92ae5%) when compared to real time pcr. there were relatively few co-infecting respiratory viruses with either pandemic h1 infections in 2009 (5ae7%) or seasonal influenza infections in 2008 (6ae3%) ( table 1 ). the most common dual infection seen with pandemic h1n1 2009 viruses and 2008 seasonal b viruses was with cvev (7 ⁄ 18; 2009 and 5 ⁄ 14; 2008, respectively) while for 2008 a(h3) viruses there were no dominant co-infecting viruses ( table 2 ). in 2009 one case was detected with three respiratory pathogens in the same sample, a 25 year old female who had pandemic h1n1, cvev, and rhv, and in a 2008 seasonal influenza sample, one case with a triple infection was detected (bocavirus, piv2 and influenza b). the median age of subjects with co-infections was younger for both pandemic h1n1 with a median age of 10 years (range: 1 months to 27 years), compared to the full 2009 sample set which had a median age of 29 years (range: 1 months to 85 years), while for the patients from 2008 with seasonal influenza viruses with co-infections they had a median age of 6ae5 years (range: 1 months to 26 years) compared to all 2008 samples which had a median age of 16 years (range: 1 months to 84 years). there was good concordance in detecting influenza a and b in respiratory samples collected in 2008 between real time rt-pcr and the resplex ii system (100% versus >92ae1% for seasonal influenza a and b respectively). this data compares well with other studies such as li et al. 2 who found that resplex ii had 82ae8% sensitivity and 100% specificity for seasonal influenza a viruses and 90ae0% sensitivity and 100% specificity for influenza b viruses. in contrast, the present study found only 53ae3% sensitivity for the resplex ii detection of influenza a with the 2009 samples that were positive for pandemic h1n1 by real time rt-pcr. a recent study by rebbapragada et al. 3 also showed lower sensitivity for pandemic h1n1 viruses in nasopharyngeal samples with the resplex ii system (55% sensitivity and 100% specificity) compared to other commercial platforms seeplex rvp (95% sensitivity and 100% specificity) and luminex rvp (100% sensitivity and 97% specificity). interestingly the latest version of the resplex system offered by qiagen the resplex ii plus panel ruo now has a separate target for the pandemic h1n1 virus (mexico 2009). 4 in terms of detection of other respiratory viruses such as piv-1, piv-3, rsv and hmpv, high sensitivities (87ae5%, 72ae2%, 73ae2%, and 80%, respectively) and specificities (99ae7-100%) compared to taqman rt-pcr have been reported from testing of nasal wash and nasopharyngeal clinical samples. 5 in both the seasonal influenza positive 2008 and the pandemic h1n1 positive (by real time rt-pcr) clinical specimens, few other respiratory viruses were detected. only 18 of the 2008 samples had another virus detectable and one had two other viruses, while in 2009 16 out had another virus and one had two other viruses detected from a total of 300 influenza virus positive samples collected in each year. enteroviruses, coronaviruses, and parainfluenza viruses were most often found with both seasonal and pandemic infections. younger age appeared to be associated with co-infections with those subjects in 2008 with dual infections having a median age of only 10 years compared to the study groups 29 years; and similarly for 2009, the median age for subjects with dual infections was only 6ae5 years compared to the study groups' median age of 16 years. a study by chong et al. 6 on 298 nasopharyngeal swabs collected during 2007-2009 using resplex ii and luminex xtag rvp fast, they found dual respiratory virus infections in 27 ⁄ 179 (15ae1%) of samples and only 2 (0ae7%) with triple respiratory viral infections; however, these were from cases with any combination of multiple respiratory viruses not necessarily influenza, although influenza positive cases were the most common respiratory virus detected (37ae4% of all positive samples). given the low level and variety of viral co-infections along with both seasonal and pandemic influenza seen in this study, it is unlikely that influenza infections predispose subjects to particular respiratory viruses, but may still allow bacterial colonization, such as has been seen with severe and fatal cases with pandemic h1n1 with various bacteria including streptococcus pneumoniae, streptococcus pyogenes, staphylococcus aureus, or haemophilus influenzae. 7, 8 low levels of other respiratory viruses along with the finding that certain cell lines (like the mdck 33016-cells used in this study) do not propagate a number of these viruses (e.g. rsv a and b, rhinoviruses, coronaviruses), but do propagate others (e.g. parainfluenza 3) 9 should make testing for unwanted viruses that might be co-isolated with influenza viruses more focused and hence easier to detect and eliminate this isolate for future vaccine production. global influenza surveillance is one of the most important approaches to combat spread of disease. current laboratory methods for characterizing influenza are time-consuming and labor-intensive, and few viral strains undergo full characterization. 1 even fewer strains from domestic poultry and swine or from wild aquatic birds are wellcharacterized. these strains are important for global surveillance since they are thought to be the precursors to pandemic influenza strains. 2 we have designed a highthroughput global bio laboratory to address these surveillance needs. the goal of this project was to develop highspeed and high-volume laboratory capabilities for extensive surveillance and rapid and accurate detection and analysis of influenza. the workflow consists of surveillance, sample transportation, laboratory testing, data management and analysis. five robotic systems have been designed for this laboratory: sample accessioning, biobanking, screening, viral culture, and sequencing. sample accessioning logs barcodes, centrifuges, and aliquots samples are then sent to biobanking. the robotic biobank stores samples at )80°c and reformats tubes for screening. the screening system extracts rna and confirms the presence and subtype of influenza. aliquots of positive samples are sent to the viral culturing system for scale-up. finally, cultured samples are extracted and sent to the sequencing system for full genome sequencing. the sample accessioning, sequencing, and biobanking systems have been built, delivered, and validation processes are currently being completed. robotic screening and culturing systems have been fully designed and are ready to be built. a biosafety level 3-enhanced containment laboratory was built to enable the flow of samples containing highly pathogenic avian influenza viruses. in full operation, this approach to surveillance is designed to enable the sequencing of up to10 000 full virus genomes per year, more than the total of all full influenza genomes sequenced to date. the design of a robotic laboratory for influenza surveillance presents unique challenges and opportunities. before a robotic system is built, each assay is worked out on the bench top, each movement of the plates and reagents is defined, and the laboratory information management system (lims) must be able to address each step of the process. alternate assays are conceived for processes that are not automation-friendly. waste streams, worker safety, and space constraints are considered. each possibility is taken to reduce processes that have the potential to aerosolize or cross-contaminate influenza samples. instruments must be found that fit the capabilities needed. detailed specifications for each of the robotic systems were written including all the parameters listed above. once the systems are built, a long validation process takes place where the processes and instruments in each system are adjusted to function together properly. finally, a validation study is performed to ensure that the system is able to produce useful data for influenza research. the entire process takes months from start to finish for each robotic system and requires complete cooperation from a diverse team of researchers. the accessioning system logs initial sample information with the lims system. samples arrive in barcoded cryotubes. the liquid handler brings all samples up to a common volume and clarifies samples by centrifugation. samples are then transferred from screw-cap sample vials into storage plates containing 96 individually punchable storage tubes. each tube (0ae3 ml) is individually identifiable with a 2d barcode on the bottom. six archive aliquots are made, and tubes are individually weld-sealed for storage. tips for aspiration are fixed and undergo a high-pressure plasma process between each use to sterilize tips and destroy nucleic acids. samples are stored at )80°c. each module has a capacity of 600 remp plates or $60 000 samples. the automated freezer system can assemble requested samples as 96-well plates while samples remain frozen. the screening system uses magnetic bead extraction chemistry, real-time pcr, and a liquid handling system to extract samples, confirm and quantify the presence of influenza, and reformat extracted samples for input into the sequencing system. serotype of human influenza samples will be performed by real-time pcr. many samples will not have enough material for further analysis and will need to be scaled up. the culturing system combines incubators, a liquid handling platform, plate reader, and real-time pcr to culture, monitor growth, harvest, and quantify influenza. when the system is not being used for culture and scale-up, it can be used to assay previously cultured influenza samples for drug resistance. a challenge to sequencing large numbers of influenza samples is the manpower required for sample preparation. the sequencing system has the capacity to prepare up to 10 000 samples for sequencing per year for sanger sequencing. sanger sequencing was chosen because it is well-established for influenza surveillance, and automation-friendly. the system is designed to work with multiple primer sets (48, 96, 192) . robotic systems all report to the lims. each process completion, plate movement, and data point are entered and checked by an online, web-based lims. status updates, notification, reporting, and data analysis can be achieved without entering the bsl3 containment facility. routine data analysis such as determining whether a cultured sample is ready to be harvested will be performed by the lims. complex data analysis, while still requiring significant human input, will be made easier by the data-acquisition functions of the lims. the implementation of a high-throughput influenza surveillance laboratory will provide an influenza research and response capacity that far exceeds what is available today. with the addition of each new system, we add a new capability to the influenza community and new opportunities to foster partnerships and collaborations with government, foundations, businesses, and academic institutions. this laboratory will not only enable cutting edge research, but will also enable a more effective response of near real-time surveillance during a pandemic outbreak. pandemics of 1957 and 1968 were believed to arise from avian influenza viruses. 1 the tropism of avian and human seasonal influenza viruses for the human lower respiratory tract deserves investigation. the target cell types that support replication of avian influenza a viruses in the human respiratory tract in the early stages of clinical infection have not well defined. in a previous autopsy studies of human h5n1 disease, influenza a virus were found to infect alveolar epithelial cells 2 and macrophages. 3 in this study, viral infectivity and replication competence of human and high and low pathogenic avian influenza viruses were systematically investigated in the human conducting and lower respiratory tract using ex vivo organ cultures. we compared the replication kinetics of human seasonal influenza viruses (h1n1 and h3n2), low pathogenic avian influenza viruses (h9n2, h5n8) with that of the highly pathogenic h5n1 viruses isolated from human h5n1 disease. a range of human seasonal influenza a viruses of subtypes h1n1 and h3n2 viruses were included in this study from 1975 to 2009. two isolates of low pathogenic avian influenza a (lpai) (h9n2) viruses from different virus lineages isolated from poultry in hong kong in 1997, a low pathogenic influenza a (h5n8) virus isolate from wild ducks in hong kong in 2005, and two virus isolates of highly pathogenic avian influenza (hpai) a subtype h5n1 were included. fragments of human bronchi and lung were cut into multiple 2-3 mm fragments within 2 hours of collection and infected in parallel with influenza a viruses at a titer of 10 6 tcid 50 ⁄ ml and as control cultures were infected with ultraviolet light inactivated virus. these tissues fragments were infected for 1 hours and washed twice with pbs and incubated for 1, 24, and 48h at 37°c. the bronchial tissue was cultured in an air-liquid interface using sponge. viral yield was assessed by titration in mdck cells. one part of the infected tissue were fixed in formalin and processed for immunohistochemistry for influenza antigen. other part of infected tissue was homogenized and underwent rna extraction, and the expression of influenza virus matrix gene was measured by quantitative rt-pcr. human bronchus ex vivo cultures supported human seasonal influenza virus to replicate efficiently. avian influenza h9n2 virus replicated, although less efficiently than that of seasonal influenza viruses, whereas hpai h5n1 did not productively replicate in ex vivo cultures of human bronchus. this is in agreement with our previous finding in the well-differentiated bronchial epithelial cells in vitro. 4 on the other hand, human lung ex vivo cultures supported prominent productive replication of human seasonal influenza h1n1 ( figure 1a ) and hpai h5n1 ( figure 1f ) viruses. lpai, such as h9n2 ( figure 1c -d) and h5n8 ( figure 1e ), also replicated productively, but with a lower viral yield. surprisingly, the replication of human influenza h3n2 viruses ( figure 1b ) across the last three decades was greatly inhibited. there are clear differences in viral tropism of human seasonal and avian influenza viruses for replication in the human bronchus and lung. hpai h5n1 virus can infect and productively replicate in the lower lung, which may account for the severity of human h5n1 disease, but not in the conducting airways. surprisingly, there are marked differences in the replication competence of seasonal influenza viruses in ex vivo lung tissues, with influenza h1n1 viruses being able to replicate efficiently while h3n2 viruses do not. this may be related to the more strict siaa2-6gal binding preference of h3n2 viruses. on the other hand, the efficient replication of influenza h1n1 viruses in the alveolar spaces indicates factors other than tissues tropism alone play a role in the differences in disease severity between human seasonal h1n1 and avian h5n1 virus infections. pre-mrnas of the influenza a virus m and ns genes are poorly spliced in virus-infected cells. by contrast, in influenza c virus-infected cells, the predominant transcript from the m gene is spliced mrna. the present study was performed to investigate the mechanism by which influenza c virus m gene-specific mrna (m mrna) is readily spliced. ribonuclease protection assays showed that the splicing of m mrna in infected cells was much higher than that in m gene-transfected cells, suggesting that viral protein(s) other than m gene-translational products facilitates the splicing of viral mrnas. the unspliced and spliced mrnas of the influenza c virus ns gene encode two nonstructural (ns) proteins, ns1 (c ⁄ ns1) and ns2 (c ⁄ ns2), respectively. the introduction of translational premature termination into the ns gene, which blocked the synthesis of c ⁄ ns1 and c ⁄ ns2 proteins, drastically reduced the splicing of ns mrna, raising the possibility that c ⁄ ns1 or c ⁄ ns2 enhances the splicing of viral mrnas. the splicing of influenza c virus m mrna was increased by co-expression of c ⁄ ns1, whereas it was reduced by co-expression of influenza a virus ns1 protein (a ⁄ ns1). the splicing of influenza a virus m mrna was also increased by co-expression of c ⁄ ns1, whereas it was inhibited by that of a ⁄ ns1. these results suggest that influenza c virus ns1, but not a ⁄ ns1, can up-regulate the splicing of viral mrnas. pre-mrnas of the influenza a virus m and ns genes are poorly spliced in virus-infected cells. 1, 2 the inefficient splicing of viral pre-mrnas can be understood partly by the fact that influenza a virus ns1 protein is associated with spliceosomes and inhibits pre-mrna splicing. 3, 4 cis-acting sequences in the ns1 transcript also negatively regulate splicing. 5 by contrast, in influenza c virus-infected cells, the predominant transcript from the m gene is spliced mrna. 6 the present study was performed to investigate the mechanism by which influenza c virus m gene-specific mrna (m mrna) is readily spliced. 7 the yamagata ⁄ 1 ⁄ 88 strain of influenza c virus was grown in the amniotic cavity of 9-day-old embryonated hen's eggs. cos-1 and 293t cells were cultured in dulbecco's modified eagle's medium containing 10% fetal calf serum. subconfluent monolayers of cos-1 cells were transfected with pme18s containing influenza c virus m gene cdna using the lipofectamine procedure and then incubated at 37°c. total rna was extracted from both the transfected cells and cells infected with c ⁄ yamagata ⁄ 1 ⁄ 88 virus using the rneasy mini kit (qiagen). ribonuclease protection assay was performed using a ribonuclease protection assay kit rpa iii (ambion). 6 briefly, a [ 33 p]-labeled influenza c virus rna 6-specific rna probe (vrna sense) was synthesized by in vitro transcription and hybridized with the total rna at 42°c overnight. hybrids were digested with rnase a (0ae08 u) and rnase t1 (3 u) at 37°c for 30 minutes and then analyzed on a 4% polyacrylamide gel containing 4 m urea. hmv-ii cells infected with c ⁄ yamagata ⁄ 1 ⁄ 88 and cos-1 cells transfected with pme18s expressing influenza c virus ns1 were fixed with carbon tetrachloride at various times after infection and transfection, respectively. the cells were then stained by an indirect method using anti-gst ⁄ ns1 serum as the primary antibody and fluorescein isothiocyanate-conjugated goat anti-rabbit igg (seikagaku kogyo) as the secondary antibody. the splicing efficiency of influenza c virus m gene-specific mrna (m mrna) in infected cells was higher than that in m gene-transfected cells the ratio of m1 encoded by a spliced m mrna to cm2 encoded by an unspliced m mrna in influenza c virusinfected cells was about 10 times larger than that in m gene-transfected cells. ribonuclease protection assays showed that the splicing of m mrna in infected cells was much higher than that in m gene-transfected cells (figure 1 ). these data suggest that viral protein(s) other than m gene-translational products facilitates viral mrna splicing. the influenza c virus ns gene translational product may up-regulate the splicing of viral mrnas the unspliced and spliced mrnas of the influenza c virus ns gene encode two nonstructural (ns) proteins, ns1 (c ⁄ ns1) and ns2 (c ⁄ ns2), respectively. the introduction of translational premature termination into the ns gene, which blocked the synthesis of c ⁄ ns1 and c ⁄ ns2 proteins, drastically reduced the splicing of ns mrna, suggesting that c ⁄ ns1 or c ⁄ ns2 enhances viral mrna splicing. immunofluorescent staining showed that ns1 localized in the nucleus in the early phase of infection, and was distributed in both the nucleus and cytoplasm in the late phase of infection, raising the possibility that influenza c virus ns1 protein plays a role in viral mrna splicing that occurs in the nucleus. the splicing of influenza c virus m mrna was increased by co-expression of c ⁄ ns1, whereas it was reduced by co-expression of influenza a virus ns1 protein (a ⁄ ns1) (figure 2a ). the splicing of influenza a virus m mrna was also increased by co-expression of c ⁄ ns1, though it was inhibited by that of a ⁄ ns1 ( figure 2b ). these results suggest that influenza c virus ns1, but not a ⁄ ns1, can up-regulate the splicing of viral mrnas. 7 in influenza a virus-infected cells, splicing is controlled so that the steady-state amount of spliced mrnas is only 5-10% of that of unspliced mrnas. 1, 2 the mechanisms by which influenza a virus ns pre-mrnas are poorly spliced have been investigated and the following confirmed. influenza a virus ns1 protein associates with spliceosomes and inhibits pre-mrna splicing. 3, 4 two cis-acting sequences in the ns1 transcript (positions 153-465 in the intron and positions 775-860 in the 3¢ exon region) inhibit splicing. 5 by contrast, influenza c virus m gene-specific mrna (m mrna) is efficiently spliced in influenza c virus-infected cells. 6 in this study, we examined the mechanism by which influenza c virus m mrna is efficiently spliced and the regulatory mechanism of the splicing of ns gene-specific mrna (ns mrna). the introduction of a translational pre-mature termination into the influenza c virus ns gene, thereby blocking the synthesis of influenza c virus ns1 (c ⁄ ns1) and ns2 (c ⁄ ns2) proteins, drastically reduced the splicing rate of ns mrna. we further examined whether c ⁄ ns1 potentially facilitates viral mrna splicing. the splicing rate of m mrna of influenza c virus was increased by co-expression with c ⁄ ns1, whereas it was reduced by co-expression with influenza a virus ns1 protein (a ⁄ ns1) (figure 2a ). the splicing of influenza a virus m gene-specific mrna was also increased by co-expression with c ⁄ ns1, though it was inhibited by co-expression with a ⁄ ns1 ( figure 2b ). these results suggest that influenza c virus ns1 can facilitate viral mrna splicing, but in no way inhibit it, which is in striking contrast to the inhibitory effect of influenza a virus ns1 on pre-mrna splicing. 3, 4 the mechanism for splicing enhancement by c ⁄ ns1 also remains to be determined. we speculate that c ⁄ ns1 may interact with some host proteins involved in splicing, thereby leading to an up-regulation in splicing, or that c ⁄ ns1 may bind to pre-mrna, increasing its accessibility to the spliceosome. the spliced mrna of the influenza c virus m gene encodes the m1 protein, which plays an important role in virus formation and determines virion morphology. 8, 9 therefore, it is speculated that the mechanism for efficient splicing of m mrna, which provides the m1 protein necessary for virus assembly in a redundant amount, has been maintained in the influenza c virus. by contrast, unspliced mrna from the influenza c virus m gene encodes the cm2 ion channel, which is permeable to chloride ions, 10 and also has ph-modulating activity. 11 although the role of the influenza c virus cm2 ion channel in virus replication remains to be determined, it is conceivable that the over-expression of the cm2 protein has a deleterious effect on virus replication since the fact that a high level of influenza a virus m2 protein expression inhibits the rate of intracellular transport of the influenza a virus ha protein and other integral membrane glycoproteins has been demonstrated. 12 if this is the case, efficient splicing of m mrna may control the amount of cm2 synthesized to optimize virus replication. therefore, we speculate that efficient splicing of m mrna leads to a high level of m1 expression and the reduced expression of cm2, thereby creating conditions that are optimal for virus replication. in this study, we provided evidence that c ⁄ ns1 facilitates the splicing of m mrna. furthermore, c ⁄ ns1 may regulate the splicing efficiency of its own ns mrna during infection, controlling the amount of c ⁄ ns1 and c ⁄ ns2 proteins in infected cells. c ⁄ ns2 plays an important role in the nuclear export of vrnp, and is also associated with vrnp in the later stages of infection in virus-infected cells and is incorporated into virions, 13 suggesting that c ⁄ ns2 is involved, not only in the sorting of vrnp into the assembly site, but also in virus assembly. therefore, it is likely that there is a mechanism by which an appropriate amount of c ⁄ ns2 is provided during infection to accomplish these functions. in conclusion, c ⁄ ns1, which enhances the splicing of viral mrna, may regulate both the expression level of m gene-derived m1 and cm2 proteins, and that of ns gene-derived ns1 and ns2 proteins, thereby leading to optimal virus replication. propagation of the human influenza viruses in embryonated hen's eggs always results in a selection of variants with amino acid substitutions in the hemagglutinin (ha) that affect viral receptor-binding characteristics (reviewed 1 ). brookes et al. 2 recently studied infection in pigs using the egg-grown virus that contained a mixture of the original a ⁄ california ⁄ 07 ⁄ 09 (h1n1pdm) and its two egg-adaptation mutants with single amino acid substitutions d222g and q223r (225 and 226 in h3 numbering system). only the original virus and the variant with 222g were detected in the directly inoculated animals, indicating that the variant with 223r failed to infect. only the original virus was detected in nasal secretions of contact infected pigs, suggesting that the d222g mutant failed to transmit. in contrast, there was an apparent selection of the d222g mutant in the lower respiratory tract samples from directly inoculated pigs. the d222g substitution is of a special interest as it can emerge during virus replication in humans and was associated with severe and fatal cases of pandemic influenza in 2009-2010 3-7 and 1918. 8 here we compared phenotypic properties of the original clinical isolate of h1n1pdm virus a ⁄ hamburg ⁄ 5 ⁄ 09 and its d222g and d223r mutants to explain observed effects of these mutations on virus replication in swine 2 and to predict their potential effects on virus replication in humans. a ⁄ hamburg ⁄ 5 ⁄ 09 (ham) was isolated from clinical material by two passages in mdck cells. the virus was passaged twice in 11-day-old embryonated hen's eggs and plaqued in mdck cells. the plaques were amplified in mdck cells and the sequences of the viral ha were determined. the variants with single mutation d222g and q223r were aliquoted and designated ham-e and ham-e1, respectively. the receptor-binding specificity of the viruses was assessed by assaying their binding to desialylated-resialylated peroxidase-labeled fetuin containing either a2-3-linked sialic acid (2-3-fet) or a2-6-linked sialic acid (2-6-fet). 9 in brief, viruses adsorbed in the wells of 96-well eia micro plates were incubated with serial dilutions of 2-3-fet or 2-6-fet, and the amount of bound fetuin probe was quantified by peroxidase activity. the binding data were converted to scatchard plots (a490 ⁄ c versus a490), and the association constants of the virus-fetuin complexes were determined from the slopes of these plots. viral cell tropism and replication efficiency in human airway epithelium were studied using fully differentiated cultures of human tracheo-bronchial epithelial cells (htbe). 10, 11 to determine cell tropism, cultures were infected at a moi 1, fixed 8 hours after infection, and double immuno-stained for virus antigen and cilia of ciliated cells. infected cells were counted under the microscope (100· objective with oil immersion) in the epithelial segment that included 15-30 consecutive microscopic fields containing between 5% and 20% ciliated cells relative to the total number of superficial cells. percentages of infected ciliated cells and infected non-ciliated cells relative to the total number of infected cells were calculated. ten segments per culture were analyzed and the results were averaged. to compare growth kinetics of ham and ham-e, replicate htbe cultures were infected with 200 plaque-forming units of the viruses followed by incubation at 35°c under airliquid interface conditions. at 24, 48, and 72 hours postinfection, we added dmem to the apical compartments of the cultures and incubated for 30 minutes at 35°c. the apical washes were harvested, stored at )80°c, and analyzed simultaneously for the presence of infectious virus by titration in mdck cells as described previously. 11 the non-egg-adapted h1n1pdm virus ham, similarly to the seasonal human virus a ⁄ memphis ⁄ 14 ⁄ 96 (h1n1), bound to 2-6-fet ( figure 1a ) and did not show any significant binding to 2-3-fet. this result contrasted with the binding of h1n1pdm viruses to several 2-3-specific probes in carbohydrate microarray analysis. 12 reduced avidity of virus interactions with soluble glycoprotein in solution as compared to its binding to the probe clustered on the microarray surface could account for these differences in the assay results. the d222g mutant ham-e differed from the parent virus by its ability to bind to 3-fet and by its reduced binding to 6-fet. the q223r mutant only bound to 2-3-fet, although less strongly than did the avian virus a ⁄ duck ⁄ alberta ⁄ 119 ⁄ 98 (h1n1). the viral cell tropism in htbe cultures ( figure 1b ) correlated with receptor specificity. ham and mem ⁄ 96 showed a typical human-virus-like tropism 10, 11 with preferential infection of non-ciliated cells (<5% of infected cells were ciliated). the mutant with 223r and control duck virus displayed a typical avian-virus-like tropism (preferential infection of ciliated cells). the d222g mutant displayed a cell tropism that was intermediate between those of human and avian viruses; in particular, this mutant infected significantly higher proportion of ciliated cells than ham and mem ⁄ 96. observed alteration of receptor specificity and cell tropism ( figure 1 ) suggested that egg-derived mutations can affect replication of the h1n1pdm virus in human airway epithelium. to test this, we first compared the capacity of the viruses to initiate infection in htbe cultures. replicate cultures were infected with identical doses of the viruses, fixed 8 hours post-infection, and immuno-stained for viral antigen. under these conditions, ham and ham-e infected comparable numbers of cells, whereas ham-e1 infected at least 10 times less cells (data not shown). this result indicated that the mutation q223r markedly impaired the ability of ham-e1 to infect human airway epithelial cultures. we next compared two other viruses ham and ham-e for their multi-cycle replication in htbe cultures and found that the original virus reached threefold higher peak titers 48 hours post infection than did the d222g mutant ( figure 2 ). the d222g mutation in h1n1pdm virus facilitates virus binding to 2-3-linked receptors and alters viral cell tropism in human airway epithelium. these changes could account for increased replication of the d222g mutant in the lower respiratory tract in humans 5-7 and pigs 2 and correlation of this mutation with severe pulmonary disease. [3] [4] [5] [6] [7] the d222g mutant replicates less efficiently in human airway cultures than the original virus. this finding correlates with an apparent lack of transmission of variants with 222g in humans 3 and pigs. 2 egg-derived mutation q223r abolishes virus binding to 2-6-linked receptors and strongly decreases infection in cultures of human airway epithelium. this result agrees with poor infectivity of the q223r mutant in pigs 2 and highlights potential pitfalls of using egg-adapted viruses with this mutation for the preparation of live influenza vaccines. nin-esterase-fusion (hef), nucleoprotein (np), matrix (m1) protein, cm2, and the non-structural proteins ns1 and ns2. 1,2 cm2 is the second membrane protein of the virus and is encoded by rna segment 6 (m gene). [3] [4] [5] [6] [7] [8] it is composed of three distinct domains: a 23-residue n-terminal extracellular domain, a 23-residue transmembrane domain, and a 69-residue cytoplasmic domain. 3, 9, 10 it is abundantly expressed at the plasma membranes of infected cells and is incorporated in a small amount into virions. 10,11 cm2 forms disulphide-linked dimers and tetramers, and is posttranslationally modified by n-glycosylation, palmitoylation, and phosphorylation. [10] [11] [12] analyses of a number of cm2 mutants revealed the positions of the amino acids involved in the posttranslational modifications. 10, 13 evidence was obtained that the n-glycosylation was not required for either the formation of disulfide-linked multimers or transport to the cell surface, 10 and that none of dimer-or tetramer-formation, palmitoylation or phosphorylation was essential to the transport of cm2 to the cell surface. 13 in the present study, in order to investigate the effect of cm2 palmitoylation on influenza c virus replication, we generated a cm2 palmitoylation-deficient influenza c virus, in which a cysteine at residue 65 of cm2 was mutated to alanine, and examined the viral growth and viral protein synthesis in infected cells. 293t and hmv-ii cells were maintained as described previously. 14,15 llc-mk 2 cells were maintained at 37°c in minimal essential medium with 5% foetal bovine serum and 5% calf serum. monoclonal antibodies (mabs) against the hef, np, and m1 proteins of c ⁄ ann arbor ⁄ 1 ⁄ 50 (aa ⁄ 50), and antisera against the aa ⁄ 50 virion and the cm2 protein were prepared as described previously. 3, [16] [17] [18] the seven pol i plasmids for the expression of viral rnas of aa ⁄ 50, and the nine plasmid dnas for the expression of the influenza c viral proteins were reported previously. 6, 14 plasmid dna, ppoli ⁄ cm2-acy(-), in which 995-tgt-997 of the m gene was replaced with 995-gct-997, was constructed based on ppoli ⁄ m. to generate a recombinant wild-type (rwt) virus, the above-mentioned 16 plasmids were transfected into 293t cells as described previously. 6 to rescue a mutant virus, rcm2-c65a, a recombinant influenza c virus lacking a cm2 palmitoylation site, the plasmid ppoli ⁄ cm2-acy(-), instead of ppoli ⁄ m, was transfected together with the other 15 plas-mids. at 90 hours posttransfection (p.t.), the respective culture medium of the transfected-293t cells was inoculated into the amniotic cavity of 9-day-old embryonated chicken eggs, and a stock of the recombinant virus was prepared. the infectious titres of the stocked recombinant viruses and the supernatants of recombinant-infected hmv-ii cells were determined according to the procedure reported previously. 19 radioimmunoprecipitation hmv-ii cells infected with recombinants were labeled with [ 35 s]methionine or [ 3 h]palmitic acid. cells were then disrupted and subjected to immunoprecipitation with the indicated antibodies. the immunoprecipitates obtained were then analysed by sds-page on 17ae5% gels containing 4 m urea, and processed for fluorography. 18 flotation analysis was performed according to the procedure described previously. 6 to examine whether the cm2 protein without palmitoylation is synthesized in rcm2-c65a-infected cells, hmv-ii cells infected with the recombinants were subjected to , and the lysates of the cells were immunoprecipitated with anti-cm2 serum and analysed by sds-page. as shown in figure 1 , the cm2 protein was synthesized both in the rwt-and rcm2-c65a-infected cells, but no incorporation of [ 3 h]palmitic acid into the cm2 proteins synthesized in the rcm2-c65ainfected cells was observed, indicating that cm2 in the rcm2-c65a-infected cells was not palmitoylated. the rwt or rcm2-c65a viruses were infected to hmv-ii cells at an m.o.i. of 5 and incubated at 33°c for up to 96 hours. the infectious titres (p.f.u. ⁄ ml) of rwt were approximately 5-to 10-fold higher than those of rcm2-c65a at 24-96 hours p.i. (data not shown), indicating that rwt grew more efficiently than did rcm2-c65a. thus palmitoylation of cm2 appears to have some effect on the generation of infectious virions in cultured cells. to investigate the reason(s) for the difference in growth kinetics between the two recombinants, we analysed viral proteins synthesized in the infected hmv-ii cells. pulsechase experiments of hmv-ii cells revealed no significant differences in the synthesis and maturation of the hef, np, m1, and cm2 proteins between the rwt-and rcm2-c65a-infected cells (data not shown). the infected cells pulse-labeled and chased were respectively immunoprecipitated with anti-cm2 serum in the presence of 50 mm iodoacetamide and analysed by sds-page in non-reducing condition. in both populations of infected cells, several bands corresponding to cm2a-monomer, -dimer, and -tetramer, as well as cm2b-dimer and -tetramer were detected (data not shown). these results demonstrate an absence of any significant differences between palmitoylation-deficient cm2 and authentic cm2 in terms of conformational maturation and transport in infected cells. membrane flotation analysis revealed that no significant differences in the kinetics of the hef, m1, and cm2 proteins were observed between rwt-and rcm2-c65ainfected cells (data not shown). in contrast, a slight difference in np kinetics was observed. the pulse-labeled np proteins were recovered in the bottom fractions in both rwt-and rcm2-c65a-infected cells. in the chase experiment, the amount of membrane-associated np proteins in fractions 3 and 4 was 30 % of the total np in the rwt-infected cells, which was higher than that (19%) in the rcm2-c65a-infected cells (data not shown). this finding may suggest that the affinity of the np protein, presumably representing the viral ribonucleoprotein (vrnp) complex, to the plasma membrane in the rcm2-c65ainfected cells is lower than that in rwt-infected cells, leading to the less efficient generation of infectious virions. since cm2 is structurally similar to m2, an influenza a virus membrane protein known to be involved in infectious virus production, [20] [21] [22] [23] [24] it is possible that the cytoplasmic tail of cm2 participates in the genome packaging through interaction with vrnp. in the present study, we showed that the affinity of np to the plasma membrane of rcm2-c65a-infected cells was slightly lower than that to the plasma membrane of rwt-infected cells. this observation may suggest that palmitoylation of cm2 is involved in the viral ribonucleoprotein (vrnp) incorporation, leading to efficient infectious virion generation. we hypothesize that palmitoylation contributes to proper regional structure formation in the cm2 cytoplasmic tail, which is competent to recruit vrnp efficiently into virions. alternatively, the cm2 cytoplasmic tail without palmitoylation is not likely to reach the proper conformation, resulting in reduced interaction with vrnp and less efficient generation of infectious progeny virions. the questions of if and how the m1 protein is involved in the interaction between the cm2 cytoplasmic tail and np remains to be clarified. we showed that cm2 synthesized in rcm2-c65ainfected cells was oligomerized and transported to the cell surface. this finding is consistent with the previous observation that palmitoylation is not required for the transport of cm2 to the cell surface in cm2-expressing cos-1 cells. 13 however, the use of reverse-genetics system has enabled us to conclude that the palmitoylation of cm2 is required for efficient infectious virus production. this suggests that the significance of the other posttranslational modifications of cm2 during virus replication can be clarified using recombinant viruses lacking the respective modification sites. sialic acid (sia) linked glycoproteins are the classical influenza receptors for influenza virus haemagglutinin to bind. the distribution of sia on cell surfaces is one of the determinants of host tropism, and understanding its expression on human cells and tissues is important for understanding influenza pathogenesis. previous research has shown the differences in apical versus basolateral infection and release of different influenza virus from polarized epithelial cells 1 and correlated this with sialic acid distribution in the human respiratory tract. moreover, mass spectrometric analysis was recently employed to elucidate the glycans present in the tissue in a higher resolution in human lung. 2 the objective of this study was to examine in detail the distribution of these sia-linked glycans at the cellular level by the use of confocal microscopy. human primary type i-like and type ii pneumocytes were isolated from human non-tumor lung tissue by tissue fragmentation, percoll density gradient centrifugation, and magnetic cell sorting. 1 the cells were seeded on coverslips and maintained in small airway growth medium. when confluence was reached, cell monolayers were fixed with 4% paraformaldehyde. we used the plant lectins, sambucus nigra glutinin (sna) from roche which binds to siaa2-6gal, maackia amurensis agglutinin (maa)i and maaii from vector lab, which bind the siaa2-3gal linked glycans using vector red as fluorescent chromogen. the cells were counter-stained with dapi or with fitc-conjugated antibody against endoplasmic recticulum (protein disulfideisomerase, pdi). the cells were imaged with multi-photon excitation laser scanning microscopy using zeiss 510 lsm. the optical cross-section pictures were reconstructed by zeiss lsm510 meta. we found that there was more binding of maai and ma-aii to type ii pneumocytes than type i-like pneumocytes and more overall binding of these lectins than binding of sna ( figure 1 ). in keeping with results from other polarized cells there was more binding to the apical than basolateral aspect, thus, explaining the previously published data on apical versus basolateral infection. 1 as sialic acid has been implicated in the targeting of proteins to the surface, the relative lack of sialic acid on the basolateral aspect can explain why there is little seasonal influenza virus dissemination to the systemic circulation in human infections. furthermore, though there was little binding of sna to the figure 1 . primary human type i-like and type ii pneumocytes stained with lecins (red), pdi (green), and dapi (blue) and imaged captured with confocal microscope. apical or basolateral aspects of the pneumocytes, the experimental findings of infection by influenza h3n2 virus that has a strict siaa2-6gal tropism 3 suggests that there are siaa2-6gal glycans present, which are not readily bound by the lectin sna. the in vitro model of primary human type i-like and type ii pneumocytes system formed a polarized epithelium that has a similar lectin distribution to human alveoli in vivo which demonstrated that it is a physiologically relevant model to study the tropism and pathogenesis of influenza a virus. human disease caused by highly pathogenic avian influenza (hpai) h5n1 virus is associated with fulminant viral pneumonia and mortality rates in excess of 60%. 1 cytokine dysregulation is thought to contribute to its pathogenesis. 2, 3 we previously found delayed onset of apoptosis in h5n1 infected human macrophages and, therefore, a longer survival time of the target cells for prolonged virus replication and cytokine and chemokine secretion, which may contribute to the pathogenesis of h5n1 disease in humans. 4 as bronchial and alveolar epithelial cells are target cells of influenza virus because of their proximal physiological location and interaction with macrophages, we further investigated if the differential onset of apoptosis could be found in influenza h5n1 and seasonal influenza h1n1 infected human respiratory epithelia. we dissected the apoptotic pathways triggered by influenza virus infection. seasonal influenza h1n1 virus (a ⁄ hk ⁄ 54 ⁄ 98), a low pathogenic avian influenza h9n2 lineage isolated from poultry (a ⁄ quail ⁄ hk ⁄ g1 ⁄ 97), and two virus isolates of hpai a subtype (a ⁄ hk ⁄ 483 ⁄ 97 and a ⁄ vn ⁄ 1203 ⁄ 04) were included. primary human bronchial and alveolar epithelial cells were infected with influenza viruses at moi of 2 and the cell monolayer was collected at 24, 30, and 48 hours post infection for tunel assay, and supernatant were collected for ldh assay. fragments of human lung tissues were cut into multiple 2-3 mm fragments within 2 hours of collection and infected with influenza a viruses at a titer of 10 6 tcid 50 ⁄ ml. these tissues fragments were infected for 1 hours and incubated for 48 hours at 37°c. one part of the infected tissue was fixed in formalin and processed for immunohistochemistry for influenza antigen, and the other part was homogenized and underwent rna extraction. apoptosis cdna superarray platform (sabioscience) was employed to conduct apoptosis pathway analysis. in bronchial epithelial cells, seasonal influenza h1n1 virus induced a high percentage of apoptotic cells by tunel assay at 24, 30, and 48 hours post infection with a peak of (figure 1) . a similar observation of delayed onset of apoptosis was found in influenza h5n1 and h9n2 infected alveolar epithelial cells. besides, cdna array data of ex vivo infected human lung showed that both influenza h5n1 and h1n1 virus induced trail expression compared with mock-infected tissue (approximately 15 folds) at 48 hours post infection, but influenza h3n2 virus infected lung induced significantly more trail (27 folds compared to mock infected cells), albeit with a limited viral replication ( figure 2 ). influenza h3n2 virus infected lung also elicited more tnf-alpha and fasr transcription than either h5n1 or h1n1. these observations can account for the greater apoptotic response in influenza h3n2 virus infected lung. as little impact on the expression of intrinsic pathway components was observed, it seems that the apoptotic response to influenza virus infection in lung was mainly through the extrinsic pathways. no significant changes in the expression of anti-apoptotic protein gene was found, except for a moderate induction of birc3 by influenza h3n2 virus, which may act to modulate the apoptotic response. the delayed onset of apoptosis by hpai h5n1 and low pathogenic avian influenza h9n2 virus infected respiratory epithelial cells may be a mechanism for the influenza viruses to have more prolonged replication within the human respiratory tract, and this may contribute to the pathogenesis of human disease. hemagglutination (ha) assay 10% crbc suspension was treated by 625 mu a2, 3-specific sialidase at 37°c for 15 minutes. complete elimination of a2, 3-receptor on sialidase-treated crbcs was confirmed by receptor staining and flow cytometry. ha assay of live viruses with 1% crbc or 1% sialidase-treated crbc were performed in bsl-3 facility. synthetic 3¢sln-paa-biotin(pa191), 6¢sln-paa-biotin(pa190), 6¢sln-ln-paa-biotin(pa343) was provided by the scripps research institute (tsri). as described elsewhere with some modifications, 3 generally, serial dilutions of sialyglycopolymers were coated in 96-well-flat-bottom polystryrene plates, and 32 hau live virus ⁄ well were added. alternatively, the plates were precoated with 5 lg ⁄ ml sialyglycopolymers, and then 8, 16, 32, 64, 128 hau live virus ⁄ well influenza viruses were added. rabbit antisera against a ⁄ ah ⁄ 1 ⁄ 05 diluted in pbs containing 1% bsa was added into the wells. bound antibody was detected by use of hrp-conjugated anti-rabbit igg antibody and tetramethylbenzidine substrate solution. each sample was determined in duplicates and the absorbance read at 450 nm. a total of 31 h5n1 virus strains were obtained from 2003 to 2009. the name and passage history of influenza viruses used in the study are listed in table 1 . as the same sequences of eight rna segments were detected in a ⁄ js ⁄ 1 ⁄ 07 and a ⁄ js ⁄ 2 ⁄ 07, only a ⁄ js ⁄ 2 ⁄ 07 was tested here. three amantadine-resistant variants with m2 mutation of screening of receptor-binding preference by ha assay representative results from three sets of independent experiments are shown in table 1 . complete ha with sialidasetreated crbcs, which were only with a2,6-receptors, was detected in human influenza virus (a ⁄ brisbane ⁄ 59 ⁄ 2007, h1n1) and two human h5n1 virus strains, a ⁄ gd ⁄ 1 ⁄ 06 and a ⁄ gx ⁄ 1 ⁄ 08. high binding of a2,3 oligosaccharides to h5n1 viruses was detected ( figure 1a -c). and enhanced a2, 6-binding preference was also detected in a ⁄ gd ⁄ 1 ⁄ 06 and a ⁄ gx ⁄ 1 ⁄ 08. the a2, 6-binding was dose dependent for sialyglycopolymers and virus titer. notably, as compared with a ⁄ gd ⁄ 1 ⁄ 06 of both short-and long-a2, 6 recognition, a ⁄ gx ⁄ 1 ⁄ 08 prefers to bind to long-a2, six oligosaccharides at low viral titer ( figure 1b,c) . however, both of them showed strong affinity to short-and long-a2, 6 oligosaccharides at high viral loads ( figure 1d ). sialoside-, galactoside-, mannoside-and sulfo-os-binding are the four types of carbohydrate-binding properties of influenza virus. 12 binding of influenza virus to the a2, 3-or a2, 6-linked sialylated glycans on cell surface is important for host range restriction, and the preference to a2, 3 of h5n1 virus limited its efficient infection in human. here, dual receptor-binding preferences were detected in a ⁄ gd ⁄ 1 ⁄ 06 and a ⁄ gx ⁄ 1 ⁄ 08, which are of clade 2ae3ae4. although there is no direct evidence supporting the occurrence of human-to-human transmission in these infection events or the association between viral virulence and receptor-binding switching, viral systemic disseminations are found in the both fatal cases (data not shown). furthermore, with the introduction of clade 2ae3ae4 into the adjacent countries of china, 4 the finding of h5n1 virus with 2-6binding in human should be of concern. though h5n1 virus with human-type receptor-binding was isolated from one patient treated by oseltamivir and those viruses were with ha and ⁄ or na substitutions, 13 whether the substitutions responsible for receptor specificity switching is pre-existed or selected in human host remains unknown. our finding that three mutant viruses bearing m2 mutations of a30s, a30t, and s31n cloned from one isolate a ⁄ hb ⁄ 1 ⁄ 06 suggested it is likely that the resistant viruses emerged in the host environment. no variation was found in their ha and na sequence, and all of them show high affinity to a2-3-binding. our data suggest that the binding-specificity was not affected by the mutations on viral envelope protein m2. with the adaptation from wild aquatic birds to domestic poultry or even in human host environment, influenza virus may possess broader carbohydrate-binding spectrum or topology conformation. 11, 14 we demonstrated differential a2, 6-binding property of two human h5n1 viruses, a ⁄ gd ⁄ 1 ⁄ 06 and a ⁄ gx ⁄ 1 ⁄ 08. though minor effect of short-a2, 6-binding was detected in viruses a ⁄ gx ⁄ 1 ⁄ 08 at low virus titer, both were of high affinity to long-a2, 6 glycans, even at the low titer which are rich on apical side of human upper respiratory epithelia. 11 notably, no evident binding preference switching was detected in the viruses isolated from the sporadic human infection cases at the early of 2009 in china (table 1) . however, higher affinity to the long-a2, 6 glycans was observed in bj ⁄ 1 ⁄ 09, gz ⁄ 1 ⁄ 09, and xj ⁄ 1 ⁄ 09 (data not shown). the discrepancy from the findings obtained by sialidase-treated crbc maybe associated with a limited abundance of n-linked a2-6 with long branches on crbc, as demonstrated in a recent study. 11 therefore, glycan dose-dependent binding assay is valuable and should be applied in flu surveillance. the underlying cause of the tendency is unknown, and further research on receptor-binding specificity of h5n1 viruses is required. influenza a viruses of migrating wild aquatic birds in north america towards improved influenza a virus surveillance in migrating birds european union council directive 92 ⁄ 40 ⁄ eec the neighbor-joining method: a new method for reconstructing phylogenetic trees confidence limits on phylogenies: an approach using the bootstrap prospects for inferring very large phylogenies by using the neighbor-joining method mega4: molecular evolutionary genetics analysis (mega) software version 4.0 the influenza virus resource at the national center for biotechnology information characterization of low-pathogenic h5 subtype influenza viruses from eurasia: implications for the origin of highly pathogenic h5n1 viruses h5n1 virus outbreak in migratory waterfowl a ⁄ h13 and a ⁄ h16 influenza viruses: different lines of one precursor evolution and ecology of influenza a viruses evolutionary processes in influenza viruses: divergence, rapid evolution, and stasis antigenic and genetic conservation of h3 influenza virus in wild ducks biologic characterization of chicken-derived h6n2 low pathogenic avian influenza viruses in chickens and ducks genetic and pathogenic characterization of h6n1 avian influenza viruses isolated in taiwan between 1972 and experimental selection of virus derivatives with variations in virulence from a single low-pathogenicity h6n1 avian influenza virus field isolate evolution and ecology of influenza a viruses is china an influenza epicenter genesis of a highly pathogenic and potentially pandemic h5n1 influenza virus in eastern asia evolution and molecular epidemiology of h9n2 influenza a viruses from quail in southern china establishment of influenza a virus (h6n1) in minor poultry species in southern china h5n1 influenza viruses: outbreaks and biological properties seroprevalance and identification of influenza a virus infection from migratory wild waterfowl in china avian flu: h5n1 virus outbreak in migratory waterfowl migration of waterfowl in the east asian flyway and spatial relationship to hpai h5n1 outbreaks avian influenza monitoring in migrating birds in taiwan during establishment of an h6n2 influenza virus lineage in domestic ducks in southern china evolution and molecular epidemiology of h9n2 influenza a viruses from quail in southern china the genesis and evolution of h9n2 influenza viruses in poultry from southern china human infection with influenza h9n2 human infection with an avian h9n2 influenza a virus in hong kong in 2003 antigenic and genetic characterization of h9n2 swine influenza in china cocirculation of avian h9n2 and contemporary ''human'' h3n2 influenza a viruses in pigs in southeastern china: potential for genetic reassortment? h9n2 influenza a viruses from poultry in asia have human virus-like receptor specificity characterization of h9 subtype influenza viruses from the ducks of southern china: a candidate for the next influenza pandemic in humans? bioedit: a user-friendly biological sequence alignment editor and analysis program for window 95 ⁄ 98 ⁄ nt genetic algorithm approaches for the phylogenetic analysis of large biological sequence datasets under the maximum likelihood criterion phylogenetic analysis using parsimony (and other methods) 4.0 beta a novel genotype h9n2 influenza virus possessing human h5n1 internal genomes has been circulating in poultry in eastern china since 1998 characterization of h9n2 influenza viruses isolated from vaccinated flocks in an integrated broiler chicken operation in eastern china during a 5 year period characterization of avian h9n2 influenza viruses from united arab emirates phylogenetic analysis of influenza a viruses of h9 haemagglutinin subtype h9n2 subtype influenza a viruses in poultry in pakistan are closely related to the h9n2 viruses responsible for human infection in hong kong diversified reassortants h9n2 avian influenza viruses in chicken flocks in northern and eastern china genotypic evolution and antigenic drift of h9n2 influenza viruses in china from the nucleoprotein as a possible major factor in determining host specificity of influenza h3n2 viruses pigs as the ''mixing vessel'' for the creation of new pandemic influenza a viruses origins and evolutionary genomics of the 2009 swine-origin h1n1 influenza a epidemic pandemic (h1n1) 2009 outbreak on pig farm reassortment of pandemic h1n1 ⁄ 2009 influenza a virus in swine from where did the 2009 'swine-origin' influenza a virus (h1n1) emerge? substitution of lysine at 627 position in pb2 protein does not change virulence of the 2009 pandemic h1n1 virus in mice evolution and ecology of influenza a viruses the origins of pandemic influenza -lessons from the 1918 virus characterization of the 1918 influenza virus polymerase genes origins and evolutionary genomics of the 2009 swine-origin h1n1 influenza a epidemic emergence of a novel swine-origin influenza a (h1n1) virus in humans antigenic and genetic characteristics of swine-origin 2009 a(h1n1) influenza viruses circulating in humans genetic reassortment of avian, swine, and human influenza a viruses in american pigs detection of two antigenic subpopulations of a(h1n1) influenza viruses from pigs: antigenic drift or interspecies transmission genetic relatedness of hemagglutinins of the h1 subtype of influenza a viruses isolated from swine and birds cases of swine influenza in humans: a review of the literature triple-reassortant swine influenza a (h1) in humans in the united states reassortment of pandemic h1n1 ⁄ 2009 influenza a virus in swine ferrets as a transmission model for influenza: sequence changes in ha1 of type a (h3n2) virus the ferret: an animal model to study influenza virus human and avian influenza viruses target different cells in the lower respiratory tract of humans and other mammals medical management of influenza infection incidence of adamantane resistance among influenza a (h3n2) viruses isolated worldwide from 1994 to 2005: a cause for concern surveillance of resistance to adamantanes among influenza a(h3n2) and a(h1n1) viruses isolated worldwide peramivir emergency use authorization peramivir and its use in h1n1 influenza cs-8958, a prodrug of the new neuraminidase inhibitor r-125489, shows long-acting antiinfluenza virus activity detection of influenza viruses resistant to neuraminidase inhibitors in global surveillance during the first 3 years of their use surveillance of influenza isolates for susceptibility to neuraminidase inhibitors during the 2000-2002 influenza seasons surveillance for neuraminidase inhibitor resistance among human influenza a and b viruses circulating worldwide from oseltamivir-resistant influenza viruses a (h1n1), norway, 2007-08 influenza activity -united states and worldwide, 2007-08 season emergence of resistance to oseltamivir among influenza a(h1n1) viruses in europe oseltamivir-resistant influenza virus a (h1n1), europe, 2007-08 season widespread oseltamivir resistance in influenza a viruses (h1n1), south africa and composition of the 2009-10 influenza vaccine emergence of h274y oseltamivir-resistant a(h1n1) influenza viruses in japan during the 2008-2009 season pyrosequencing as a tool to detect molecular markers of resistance to neuraminidase inhibitors in seasonal influenza a viruses neuraminidase sequence analysis and susceptibilities of influenza virus clinical isolates to zanamivir and oseltamivir host cell selection of influenza neuraminidase variants: implications for drug resistance monitoring in a(h1n1) viruses neuraminidase receptor binding variants of human influenza a(h3n2) viruses due to substitution of aspartic acid 151 in the catalytic site -role in virus attachment? neuraminidase inhibitor susceptibility testing in human influenza viruses: a laboratory surveillance perspective update: drug susceptibility of swine-origin influenza a (h1n1) viruses comprehensive assessment of 2009 pandemic influenza a (h1n1) virus drug susceptibility in vitro detection of molecular markers of drug resistance in 2009 pandemic influenza a (h1n1) viruses by pyrosequencing pandemic (h1n1) 2009 and oseltamivir resistance in hematology/oncology patients fluview: a weekly influenza surveillance report prepared by the influenza division development of a sensitive chemiluminescent neuraminidase assay for the determination of influenza virus susceptibility to zanamivir evaluation of neuraminidase enzyme assays using different substrates to measure susceptibility of influenza virus clinical isolates to neuraminidase inhibitors: report of the neuraminidase inhibitor susceptibility network surveillance for neuraminidase inhibitor resistance among human influenza a and b viruses circulating worldwide from different neuraminidase inhibitor susceptibilities of human h1n1, h1n2, and h3n2 influenza viruses isolated in germany from oseltamivir-and amantadine-resistant influenza viruses a (h1n1) infections with oseltamivirresistant influenza a (h1n1) virus in the united states isolation of oseltamivirresistant influenza a ⁄ h1n1 virus of different origins in yokohama city, japan, during the 2007-2008 influenza season a two-year survey of the oseltamivir-resistant influenza a(h1n1) virus in yamagata, japan and clinical effectiveness of oseltamivir and zanamivir oseltamivir-resistant influenza a pandemic (h1n1) 2009 virus detection of an oseltamivir-resistant pandemic influenza a ⁄ h1n1 virus in hong kong a novel small-molecule inhibitor of the avian influenza h5n1 virus determined through computational screening against the neuraminidase aurintricarboxylic acid is a potent inhibitor of influenza a and b virus neuraminidases oseltamivir-resistant influenza a viruses are transmitted efficiently among guinea pigs by direct contact but not by aerosol oseltamivir is adequately absorbed following nasogastric administration to adult patients with severe h5n1 influenza prevention and control of influenza influenza-associated hospitalizations in the united states mortality associated with influenza and respiratory syncytial virus in the united states severe respiratory disease concurrent with the circulation of h1n1 influenza staphlococcus aureus pneumonia studies on influenza in the pandemic of 1957-1958. ii. pulmonary complications of influenza bacterial pneumonia during the hong kong influenza epidemic of 1968-1969 streptococcus pneumonia diagnostic value of microscopic examination of gram-stained sputum and sputum cultures in patients with bacteremic pneumococcal pneumonia pneumonitis-associated hyperprocalcitoninemia clinical review 167: procalcitonin and the calcitonin gene family of peptides in inflammation, infection, and sepsis: a journey from calcitonin back to its precursors procalcitonin in sepsis and systemic inflammation: a harmful biomarker and a therapeutic target procalcitonin in children with suspected novel influenza a (h1n1) infection in vitro and in vivo calcitonin i gene expression in parenchymal cells: a novel product of human adipose tissue serum procalcitonin and c-reactive protein levels as markers of bacterial infection: a systematic review and meta-analysis procalcitonin as a diagnostic test for sepsis in critically ill adults and after surgery or trauma: a systematic review and meta-analysis respiratory syncytial virus infection in elderly and high-risk adults procalcitonin assay in systemic inflammation, infection, and sepsis: clinical utility and limitations simultaneous detection of chlamydophila pneumoniae and mycoplasma pneumoniae by use of molecular beacons in a duplex real-time pcr 10 years influenza therapy with antivirals (zanamivir and oseltamivir), 2nd world summit of antivirals the art of viewing and influenza -its importance for diagnosis, 3rd european influenza conference 9 emergence of a novel swine origin influenza a (h1n1) virus in humans origins and evolutionary genomics of the 2009 swine-origin h1n1 influenza a epidemic molecular detection of a novel human influenza (h1n1) of pandemic potential by conventional and real-time quantitative rt-pcr assays pandemic influenza a(h1n1)v: human to pig transmission in norway? an investigation into human pandemic influenza virus (h1n1) 2009 on an alberta swine farm pandemic (h1n1) 2009 infection in swine herds implications of the emergence of a novel h1 influenza virus emergence of a novel swine-origin influenza a virus (s-oiv) h1n1 virus in humans rapid detection of reassortment of pandemic h1n1 ⁄ 2009 influenza virus rapid diagnosis of a coronavirus associated with severe acute respiratory syndrome (sars) universal primer set for the fulllength amplification of all influenza a viruses cocirculation of avian h9n2 and contemporary ''human'' h3n2 influenza a viruses in pigs in southeastern china: potential for genetic reassortment? hog in the limelight: swine flu's got new genes on reassortment of pandemic h1n1 ⁄ 2009 influenza a virus in swine pediatric hospitalizations associated with 2009 pandemic influenza a (h1n1) in argentina factors associated with death or hospitalization due to pandemic 2009 influenza a(h1n1) infection in california mortality from pandemic a ⁄ h1n1 2009 influenza in england: public health surveillance study antigenic and genetic characteristics of swine-origin 2009 a(h1n1) influenza viruses circulating in humans characterization of a newly emerged genetic cluster of h1n1 and h1n2 swine influenza virus in the united states experimental inoculation of pigs with pandemic h1n1 2009 virus and hi cross-reactivity with contemporary swine influenza virus antisera triple-reassortant swine influenza a (h1) in humans in the united states identification of human h1n2 and human-swine reassortant h1n2 and h1n1 influenza a viruses among pigs in ontario genetic and pathobiologic characterization of pandemic h1n1 2009 influenza viruses from a naturally infected swine herd archive number pandemic (h1n1) 2009 outbreak on pig farm reassortment of pandemic h1n1 ⁄ 2009 influenza a virus in swine a simple method for estimating 50 percent endpoints occurrence of haemagglutinin mutation d222g in pandemic influenza a(h1n1) infected patients in the west of scotland rrt-pcr) protocol for detection and characterization of swine influenza (version diagnostic hybrids, inc. readycells respiratory training r-mixô&d3ô dfa respiratory virus screening & id kit handbook centers for disease and control and prevention. influenza laboratory course comparison study of a real-time reverse transcription polymerase chain reaction assay with an enzyme immunoassay and shell vial culture for influenza a and b virus detection in adult patients comparison of a commercial qualitative real-time rt-pcr kit with direct immunofluorescence assay (dfa) and cell culture for detection of influenza a and b in children prospective comparison of r-mix shell vial system with direct antigen tests and conventional cell culture for respiratory virus detection optimized detection of respiratory viruses in nasopharyngeal secretions evaluation of r-mix freshcells in shell vials for detection of respiratory viruses time lines of infection and disease in human influenza: a review of volunteer challenge studies comparative epidemiology of pandemic and seasonal influenza a in households viral shedding and clinical illness in naturally acquired influenza virus infections population modeling of influenza a/h1n1 virus kinetics and symptom dynamics sensitivity of influenza rapid diagnostic tests to h5n1 and 2009 pandemic h1n1 viruses clinical evaluation of an immunochromatography test kit, capilia flua,b, for rapid diagnosis of influenza time to positive reaction of influenza rapid diagnosis kit and its clinical significance the speed and reliability of the antigen detection kit capilia flu a, b when used in a clinical setting options for the control of influenza vi role of rapid immunochromatographic antigen testing in diagnosis of influenza a virus 2009 h1n1 infection poor clinical sensitivity of rapid antigen test for influenza a pandemic (h1n1) 2009 virus evaluation of new rapid antigen test for the detection of pandemic influenza a ⁄ h1n1 2009 virus application of real-time cell electronic sensing (rt-ces) technology to cell-based assays world health organization. pandemic (h1n1) role of the laboratory in diagnosis of influenza during seasonal epidemics and potential pandemics rapid detection and identification of 12 respiratory viruses using a dual priming oligonucleotide systembased multiplex pcr assay world health organization. influenza world health organization. cdc protocol of realtime rtpcr for influenza a (h1n1) development of a reverse transcription-loop-mediated isothermal amplification assay for detection of pandemic (h1n1) 2009 virus as a novel molecular method for diagnosis of pandemic influenza in resource-limited settings one-step real-time rt-pcr for pandemic influenza a virus (h1n1) 2009 matrix gene detection in swine samples detection of pandemic influenza a ⁄ h1n1 ⁄ 2009 virus by real-time reverse transcription polymerase chain reaction detection of influenza a(h1n1)v virus by real-time rt-pcr development of a real-time rt-pcr for the detection of swine-lineage influenza a (h1n1) virus infections evaluation of multiple test methods for the detection of the novel 2009 influenza a (h1n1) during the new york city outbreak design and clinical application of a molecular method for detection and typing of the influenza a ⁄ h1n1pdm virus confirmation of the first hong kong case of human infection by novel swine origin influenza a (h1n1) virus diagnosed using ultrarapid, real-time reverse transcriptase pcr evaluation of four real-time pcr assays for detection of influenza a(h1n1)v viruses diagnostic assay recommended by the world health organization for swine origin influenza a (h1n1) virus cross-reacts with h5n1 influenza virus increasing immunization coverage the cost-effectiveness of influenza vaccination for people aged 50 to 64 years: an international model national seasonal influenza vaccination survey in europe seasonal influenza and vaccination coverage influenza vaccination coverage rates in five european countries during season 2006 ⁄ 07 and trends over six consecutive seasons influenza in the elderly -a mini-review cost of vaccine administration among pediatric practices cost of universal influenza vaccination of children in pediatric practices surveillance report: respiratory infections strategies for mitigating an influenza pandemic methods for monitoring influenza surveillance data approaches to syndromic surveillance when data consist of small regional counts rrt-pcr) protocol for detection and characterization of swine influenza (version influenza virus strains circulating in mongolia in 2003-2007 swine influenza who (2009) influenza a(h1n1) world now at the start of 2009 influenza pandemic general statement following the ninth meeting of the emergency committee who (2010) influenza updates us dept of defense global emerging infections surveillance è me recensement general de la population et l'habitat (rgph 2009), ministè re de l'economie des finances et de la planification ministry of health djibouti a survey of human cases of h5n1 avian influenza report by who before international health regulation (ihr) electronic public health surveillance in developing settings statistical analyses in disease surveillance systems potential for a global dynamic of influenza a (h1n1) infectious disease informatics: syndromic surveillance for public health and biodefense visualization techniques and graphical user interfaces in syndromic surveillance systems statistical inference for infectious diseases. risk-specific household and community transmission parameters evaluation of seasonal influenza vaccination effectiveness based on antibody efficacy among the institutionalized elderly in japan australian influenza surveillance summary report. canberra: australian government department of health and ageing australia's winter with the 2009 pandemic influenza a (h1n1) virus seroprevalence of 2009 pandemic influenza a(h1n1) virus in australian blood donors pandemic vaccination survey: summary results. cat. no. phe 128 concepts and procedures for laboratory-based influenza surveillance. who collaborating centres for reference and research on influenza influenza a(h1n1) seroconversion rates and risk factors among distinct adult cohorts in singapore ecdc. forward look risk assessment: likely scenarios for influenza in 2010 and the 2010 ⁄ 2011 influenza season in europe and the consequent work priorities. stockholm: european centre for disease prevention and control epidemiology of influenza -summary of influenza workshop iv sampling of populations: methods and applications comparison of influenza serological techniques by international collaborative study eliminating bias in the estimation of the geometric mean of hi titres reply to ''nauta jjp, eliminating bias in the estimation of the geometric mean of hi titers novel influenza a virus infections. 2010 case definition institut national de la statistique et des é tudes é conomiques. taille des mé nages dans l'union europé enne estimation de la population au 1er janvier par ré gion, dé partement, sexe et â ge sero-immunity and serologic response to pandemic influenza a (h1n1) 2009 virus in hong kong pandemic influenza a(h1n1) virus in scotland: geographically variable immunity in spring 2010, following the winter outbreak high prevalence of antibodies to the 2009 pandemic influenza a(h1n1) virus in the norwegian population following a major epidemic and a large vaccination campaign in autumn incidence of 2009 pandemic influenza a h1n1 infection in england: a cross-sectional serological study prevalence of seroprotection against the pandemic (h1n1) virus after the 2009 pandemic antibodies in residents of new south wales, australia, after the first pandemic wave in the 2009 southern hemisphere winter older age and a reduced likelihood of 2009 h1n1 virus infection pre-existing antibody response against 2009 pandemic influenza h1n1 viruses in taiwanese population cross-reactive antibody responses to the 2009 pandemic h1n1 influenza virus seasonal influenza vaccine and increased risk of pandemic a ⁄ h1n1-related illness: first detection of the association in british columbia association between the 2008-09 seasonal influenza vaccine and pandemic h1n1 illness during spring-summer 2009: four observational studies from canada does seasonal influenza vaccination increase the risk of illness with the 2009 a ⁄ h1n1 pandemic virus seasonal h1n1 2007 influenza virus infection is associated with elevated pre-exposure antibody titers to the serum cross-reactive antibody response to a novel influenza a (h1n1) virus after vaccination with seasonal influenza vaccine world now at the start of 2009 influenza pandemic rki working group pandemic. the first wave of pandemic influenza (h1n1) 2009 in germany: from initiation to acceleration department of health and ageing. australian influenza surveillance canadian surveillance studies help to understand h1n1 patterns influenza pandé mica a un añ o de la primera ola. ¿qué podemos decir ahora? revista chilena de infectología pandemic (h1n1) 2009 briefing note 10 options for the control of influenza vii managing and reducing uncertainty in an emerging influenza pandemic the severity of pandemic h1n1 influenza in the united states assessing the severity of the novel influenza a ⁄ h1n1 pandemic tecumseh study of illness. xiii. influenza infection and disease, 1976-1981 the infection attack rate and severity of 2009 pandemic influenza (h1n1) in hong kong school closure and mitigation of pandemic (h1n1) 2009, hong kong comparative epidemiology of pandemic and seasonal influenza a in households incidence of 2009 pandemic influenza a h1n1 infection in england: a cross-sectional serological study mortality from pandemic a ⁄ h1n1 2009 influenza in england: public health surveillance study monthly summary tables of influenza virus isolation different epidemic curves for severe acute respiratory syndrome reveal similar impacts of control measures estimating in real time the efficacy of measures to control emerging communicable diseases the effective reproduction number of pandemic influenza in hong kong: prospective estimation multiple imputation: a primer comparative epidemiology of pandemic and seasonal influenza a in households strategies for containing an emerging influenza pandemic in southeast asia estimation of the serial interval of influenza school closure and mitigation of pandemic (h1n1) 2009, hong kong estimating the reproduction number of the novel influenza a virus (h1n1) in a southern hemisphere setting: preliminary estimate in new zealand early transmission characteristics of influenza a(h1n1)v in australia: victorian state methods for monitoring influenza surveillance data comparative estimation of the reproduction number for pandemic influenza from daily case notification data how to maintain surveillance for novel influenza a h1n1 when there are too many cases to count a computational framework for influenza antigenic cartography mapping the antigenic and genetic evolution of influenza virus proc of kdd cup and workshop matrix factorization techniques for recommender systems exact matrix completion via convex optimization a singular value thresholding algorithm for matrix completion matrix completion from a few entries matrices with prescribed entries and eigenvalues combining predictions for accurate recommender systems the bellkor solution to the netflix grand. prize the pragmatic theory solution to the netix grand prize feature-weighted linear stacking technical report. of health and ageing (australia). the australian health management plan for pandemic influenza influenza pandemic contingency plan, version 8.0 world health organisation. pandemic influenza preparedness and response prophylaxis or treatment? optimal use of an antiviral stockpile during an influenza pandemic the response to novel influenza a (h1n1) epidemic in taiwan and analysis of the initial 61 confirmed cases mitigation strategies for pandemic influenza in the united states strategies for mitigating an influenza pandemic containing pandemic influenza at the source efficient simulation of the spatial transmission dynamics of influenza world health organization. mathematical modelling of the pandemic h1n1 a mathematical model for the global spread of influenza pandemic potential of a strain of influenza a (h1n1): early findings effectiveness of the 2003-2004 influenza vaccine among children 6 months to 8 years of age, with 1 vs 2 doses cdc estimates of 2009 h1n1 influenza cases, hospitalizations and deaths in the united states accessed 1 entry screening for severe acute respiratory syndrome (sars) or influenza: policy evaluation pandemic of influenza 1918-1919 america's forgotten pandemic. the influenza of 1918 influenza: a survey of the last 50 years in the light of modern work on the virus of epidemic influenza understanding mortality in the 1918-19 influenza pandemic in england and wales genomic analysis of increased host immune and cell death responses induced by 1918 influenza virus the age pattern of mortality in the 1918-1919 influenza pandemic: an attempted explanation based on data for england and wales protective immunity and susceptibility to infectious diseases: lessons from the 1918 influenza pandemic deaths from bacterial pneumonia during 1918-1919 influenza pandemic understanding influenza transmission, immunity and pandemic threats a biological model for influenza transmission: pandemic planning implications of asymptomatic infection and immunity prior immunity helps to explain wave-like behaviour of pandemic influenza in 1918-1919 influenza: accounting for prior immunity 1918-1919 influenza pandemic mortality in england and wales. colchester, essex: uk data archive [distributor] cross-reactive antibody responses to the 2009 pandemic h1n1 influenza virus australia's winter with the 2009 pandemic influenza a (h1n1) virus model predictions and evaluation of possible control strategies for the 2009 a ⁄ h1n1v influenza pandemic in italy the role of population heterogeneity and human mobility in the spread of pandemic influenza age-prioritized use of antivirals during an influenza pandemic mitigation measures for pandemic influenza in italy: an individual based model considering different scenarios strategies for containing an emerging influenza pandemic in southeast asia modeling targeted layered containment of an influenza pandemic in the united states reducing the impact of the next influenza pandemic using household-based public health interventions mitigation strategies for pandemic influenza in the united states effective, robust design of community mitigation for pandemic influenza: a sytematic examination of proposed us guidance containing pandemic influenza at the source strategies for mitigating an influenza pandemic new influenza a (h1n1) virus: global epidemiological situation velluci l. enhanced epidemiological surveillance of influenza a(h1n1)v in italy changes in reporting requirements for pandemic (h1n1) 2009 virus infection use of the oral neuraminidase inhibitor oseltamivir in experimental human influenza guidance document: pandemic influenza preparedness and response situation updates -pandemic (h1n1) pandemic potential of a strain of influenza a (h1n1): early findings estimating the impact of school closure on influenza transmission from sentinel data preexisting immunity to pandemic (h1n1) understanding the dynamics of seasonal influenza in italy: an analysis of disease incidence and population susceptibility what is the optimal therapy for patients with h5n1 influenza meta-analysis: convalescent blood products for spanish influenza pneumonia: a future h5n1 treatment? treatment with convalescent plasma for influenza a (h5n1) infection prophylactic and therapeutic efficacy of human monoclonal antibodies against h5n1 influenza logistical feasibility and potential benefits of a population-wide passive-immunotherapy program during an influenza pandemic strategies for pandemic and seasonal influenza vaccination of schoolchildren in the united states mitigation strategies for pandemic influenza in the united states pandemic potential of a strain of influenza a (h1n1): early findings transmission potential of the new influenza a(h1n1) virus and its age-specificity in japan the transmissibility and control of pandemic influenza a (h1n1) virus effect of clinical and virological parameters on the level of neutralizing antibody against pandemic influenza a virus h1n1 world health organization. 10 facts on blood transfusion hyperinduction of cyclooxygenase-2-mediated proinflammatory cascade: a mechanism for the pathogenesis of avian influenza h5n1 infection the role of influenza virus gene constellation and viral morphology on cytokine induction, pathogenesis, and viral virulence induction of proinflammatory cytokines in human macrophages by influenza a (h5n1) viruses: a mechanism for the unusual severity of human disease? analysis of relative gene expression data using real-time quantitative pcr and the 2 acetylsalicylic acid (asa) blocks influenza virus propagation via its nf-kappab-inhibiting activity cyclooxygenase 2 plays a pivotal role in the resolution of acute lung injury role of 15-deoxy delta(12,14) prostaglandin j2 and nrf2 pathways in protection against acute lung injury resolution of inflammation in murine autoimmune arthritis is disrupted by cyclooxygenase-2 inhibition and restored by prostaglandin e2-mediated lipoxin a4 production inhibition of cox-2 aggravates neutrophil migration and pneumocyte apoptosis in surfactantdepleted rat lungs re-emergence of fatal human influenza a subtype h5n1 disease apoptosis and pathogenesis of avian influenza a (h5n1) virus in humans h5n1 infection of the respiratory tract and beyond: a molecular pathology study molecular biology, and pathogenesis of avian influenza a (h5n1) infection in humans passive immunity in prevention and treatment of infectious diseases prophylactic and therapeutic efficacy of human monoclonal antibodies against h5n1 influenza neutralizing human monoclonal antibody against h5n1 influenza ha selected from a fab-phage display library passive immunoprophylaxis and therapy with humanized monoclonal antibody specific for influenza a h5 hemagglutinin in mice generation, characterization and epitope mapping of two neutralizing and protective human recombinant antibodies against influenza a h5n1 viruses igy: clues to the origins of modern antibodies chicken antibodies: a clinical chemistry perspective oral immunotherapy with yolk antibodies to prevent infections in humans and animals good effect of igy against pseudomonas aeruginosa infections in cystic fibrosis patients avian influenza vaccines: a practical review in relation to their application in the field with a focus on the asian experience vaccination: a tool for the control of avian influenza immunoglobulins from egg yolk: isolation and purification a simple method of estimating fifty per cent endpoints characterization of the reconstructed 1918 spanish influenza pandemic virus detection of antibody to avian influenza a (h5n1) virus in human serum by using a combination of serologic assays avian influenza (h5n1) viruses isolated from humans in asia in 2004 exhibit increased virulence in mammals heterosubtypic immunity to influenza a virus infection requires a properly diversified antibody repertoire early control of h5n1 influenza virus replication by the type i interferon response in mice successful treatment of avian influenza with convalescent plasma chicken antibodies: taking advantage of evolution -a review oral administration of specific yolk antibodies (igy) may prevent pseudomonas aeruginosa infections in patients with cystic fibrosis: a phase i feasibility study avian flu: isolation of drug-resistant h5n1 virus reduced sensitivity of influenza a (h5n1) to oseltamivir enzymatic properties of the neuraminidase of seasonal h1n1 influenza viruses provide insights for the emergence of natural resistance to oseltamivir surveillance for neuraminidase inhibitor resistance among human influenza a and b viruses circulating worldwide from a polyphenol rich plant extract, cystus052, exerts anti influenza virus activity in cell culture without toxic side effects or the tendency to induce viral resistance cystus052, a polyphenol-rich plant extract, exerts anti-influenza virus activity in mice cistus incanus (cystus052) for treating patients with infection of the upper respiratory tract. a prospective, randomised, placebocontrolled clinical study in vitro and in vivo assay systems for study of influenza virus inhibitors new low-viscosity overlay medium for viral plaque assays antibacterial and antifungal activities of cistus incanus and c. monspeliensis leaf extracts flavan-3-ols and proanthocyanidins from cistus incanus inhibition of the infectivity of influenza virus by tea polyphenols attenuation of inducible respiratory immune responses by oseltamivir treatment in mice infected with influenza a virus anti-inflammatory activity of macrolides: a new therapeutic potential? macrolide antibiotics as immunomodulatory medications: proposed mechanisms of action proteases essential for human influenza virus entry into cells and their inhibitors as potential therapeutic agents modified pulmonary surfactant is a potent adjuvant that stimulates the mucosal iga production in response to the influenza virus antigen boost of mucosal sectretory immunoglobulin a response by clarithromycin in paediatric influenza intranasal interleukin-12 is a powerful adjuvant for protective mucosal immunity immunological effects of the orally administered neuraminidase inhibitor oseltamivir in influenza virus-infected and uninfected mice clarithromycin in the treatment of rsv bronchitis: a double-blind, randomized placebo-controlled trial bird flu''), inflammation and anti-inflammatory ⁄ analgesic drugs clinical effects of low-dose long-term erythromycin chemotherapy on diffuse panbronchiolitis proteolytic activation of influenza viruses by serine proteases tmprss2 and hat from human airway epithelium proteolytic activation of the 1918 influenza virus hemagglutinin protease-activated receptors: new concepts in regulation of g protein-coupled receptor signaling and trafficking protective role for protease-activated receptor-2 against influenza virus pathogenesis via an ifn-gamma-dependent pathway agonists of proteinaseactivated receptor-2 enhance ifn-gamma-inducible effects on human monocytes: role in influenza a infection plasmacytoid dendritic cells migrate in afferent skin lymph annexin ii incorporated into influenza virus particles supports virus replication by converting plasminogen into plasmin plasminogen promotes influenza a virus replication through an annexin ii-dependent pathway in absence of neuraminidase costimulatory receptors in a teleost fish: typical cd28, elusive ctla4 trypsin increases pseudorabies virus production through activation of the erk signalling pathway characterization of hla-g1, -g2, -g3, and -g4 isoforms transfected in a human melanoma cell line immunosuppressive hla-g molecule is upregulated in alveolar epithelial cells after influenza a virus infection influenza virus propagation is impaired by inhibition of the raf ⁄ mek ⁄ erk signalling cascade influenza viruses and the nf-kappab signaling pathway -towards a novel concept of antiviral therapy the immunotolerance role of hla-g hla-g1 co-expression boosts the hla class i-mediated nk lysis inhibition hla-g inhibits the allogeneic proliferative response hla-g-mediated inhibition of antigen-specific cytotoxic t lymphocytes the world health organization. h1n1 influenza vaccine task force pandemic influenza: a potential role for statins in treatment and prophylaxis commentary: from scarcity to abundance: pandemic vaccines and other agents for ''have not'' countries new technologies for meeting the global demand for pandemic vaccines confronting the next influenza pandemic with inexpensive generic agents: can it be done? was bacterial pneumonia the predominant cause of death in the 1918-1919 influenza pandemic? meeting the challenge of influenza pandemic preparedness in developing countries confronting the next influenza pandemic with antiinflammatory and immunomodulatory agents: why they are needed and how they might work molecular and cellular aspects of sepsis-induced immunosuppression immune homeostasis in the respiratory tract and its impact on heterologous infection identification of oxidative stress and toll-like receptor 4 signaling as a key pathway of acute lung injury effector t cells control lung inflammation during acute influenza virus infection by producing il-10 a detrimental effect of interleukin-10 on protective pulmonary humoral immunity during primary influenza a virus infection inhibition of influenza a virus replication by resveratrol increased survival after gemfibrozil treatment of severe mouse influenza tnf ⁄ inos-producing dendritic cells are the necessary evil of lethal influenza virus infection peroxisome proliferator-activated receptor and amp-activated protein kinase agonists protect against lethal influenza virus challenge in mice acetylsalicylic acid (asa) blocks influenza virus propagation via its nf-kappab-inhibiting activity new low-viscosity overlay medium for viral plaque assays cyproquant-pcr: a real time rt-pcr technique for profiling human cytokines, based on external rna standards, readily automatable for clinical use role of hypercytokinemia in nf-kap-pab p50 deficient mice after h5n1 influenza a virus infection developing new antiviral agents for influenza treatment: what does the future hold? influenza viruses and the nf-kappab signaling pathway -towards a novel concept of antiviral therapy rna viruses and the mitogenic raf ⁄ mek ⁄ erk signal transduction cascade world health organization writing group. non-pharmaceutical interventions for pandemic influenza, international measures effects of school closures, 2008 winter influenza season, hong kong entry screening to delay local transmission of 2009 pandemic influenza a (h1n1) simple physical interventions such as hand washing and wearing masks can reduce spread of respiratory viruses hand hygiene and virus transmission demographic and attitudinal determinants of protective behaviours during a pandemic: a review perceptions related to human avian influenza and their associations with anticipated psychological and behavioral responses at the onset of outbreak in the hong kong chinese general population perceptions about status and modes of h5n1 transmission and associations with immediate behavioral responses in the hong kong general population longitudinal assessment of community psychobehavioral responses during and after the 2003 outbreak of severe acute respiratory syndrome in hong kong public perceptions, anxiety, and behaviour change in relation to the swine flu outbreak: cross sectional telephone survey community psychological and behavioral responses through the first wave of the 2009 influenza a(h1n1) pandemic in hong kong main tables of the 2006 population census. hong kong: government of the statistical analysis with missing data the impact of community psychological responses on outbreak control for severe acute respiratory syndrome in hong kong economic impact of sars: the case of hong kong sars-related perceptions in hong kong group members: stream 1 -ilaria oie -world organisation for animal health, france; malik peiris us department of health and human services, usa; masato tashiro, national institute of infectious diseases, japan; marie-paule kieny 1 , david wood 1 . stream 4 -tawee chotpitayasunondh world influenza centre at national institute for medical research philippe veltsos 1 , cathy roth 1 , gregory härtl 1 options for the control of influenza vii who public health research agenda for influenza outbreak communication best practices for communicating with the public during an outbreak, doc who ⁄ cds ⁄ 2005.32. geneva: world health organization who outbreak communications guidelines, geneva: world health organization characterization of h9 subtype influenza viruses from the ducks of southern china: a candidate for the next influenza pandemic in humans? genetic conservation of hemagglutinin gene of h9 influenza virus in chicken population in mainland china continuing evolution of h9 influenza viruses in korean poultry evolution and molecular epidemiology of h9n2 influenza a viruses from quail in southern china characterization of the pathogenicity of members of the newly established h9n2 influenza virus lineages in asia sequence analysis of the hemagglutinin gene of h9n2 korean avian influenza viruses and assessment of the pathogenic potential of isolate ms96 molecular characterization of h9n2 influenza viruses: were they the donors of the ''internal'' genes of h5n1 viruses in hong kong? h9n2 influenza viruses possessing h5n1-like internal genomes continue to circulate in poultry in southeastern china role of infectious bronchitis live vaccine on pathogenicity of h9n2 avian influenza virus manual of diagnostic tests and vaccines for terrestrial animals (mammals, birds and bees) an avian influenza virus of h10 subtype that is highly pathogenic for chickens, but lacks multiple basic amino acids at the haemagglutinin cleavage site human influenza virus hemagglutinin with high sensitivity to proteolytic activation insertion of a multibasic cleavage motif into the hemagglutinin of a low-pathogenic avian influenza h6n1 virus induces a highly pathogenic phenotype acquisition of a polybasic hemagglutinin cleavage site by a low-pathogenic avian influenza virus is not sufficient for immediate transformation into a highly pathogenic strain pathology and virus distribution in chickens naturally infected with highly pathogenic avian influenza a virus (h7n7) during the 2003 outbreak in the netherlands implication of the proprotein convertases furin, pc5 and pc7 in the cleavage of surface glycoproteins of hong kong, ebola and respiratory syncytial viruses: a comparative analysis with fluorogenic peptides substrate cleavage analysis of furin and related proprotein convertases. a comparative study cloning and expression of novel mosaic serine proteases with and without a transmembrane domain from human lung host envelope glycoprotein processing proteases are indispensable for entry into human cells by seasonal and highly pathogenic avian influenza viruses world health organization ⁄ oie ⁄ fao h5n1 evolution working group. toward a unified nomenclature system for highly pathogenic avian influenza virus (h5n1) novel type ii transmembrane serine proteases, mspl and tmprss13, proteolytically activate membrane fusion activity of hemagglutinin of highly pathogenic avian influenza viruses and induce their multicycle replication is virulence of h5n2 influenza viruses in chickens associated with loss of carbohydrate from the hemagglutinin? targeted infection of endothelial cells by avian influenza virus a ⁄ fpv ⁄ rostock ⁄ 34 (h7n1) in chicken embryos detection of circulating asian h5n1 viruses by a newly established monoclonal antibody a novel influenza a virus mitochondrial protein that induces cell death influenza a virus pb1-f2 protein contributes to viral pathogenesis in mice expression of the 1918 influenza a virus pb1-f2 enhances the pathogenesis of viral and secondary bacterial pneumonia the influenza a virus pb1-f2 protein targets the inner mitochondrial membrane via a predicted basic amphipathic helix that disrupts mitochondrial function influenza virus pb1-f2 protein induces cell death through mitochondrial ant3 and vdac1 a single mutation in the pb1-f2 of h5n1 (hk ⁄ 97) and 1918 influenza a viruses contributes to increased virulence the effects of influenza a virus pb1-f2 protein on polymerase activity are strain specific and do not impact pathogenesis eight-plasmid system for rapid generation of influenza virus vaccines pb1-f2 proteins from h5n1 and 20 century pandemic influenza viruses cause immunopathology pneumonia and respiratory failure from swine-origin influenza a (h1n1) in mexico emergence of a novel swine-origin influenza a (h1n1) virus in humans dating the emergence of pandemic influenza viruses in vitro and in vivo characterization of new swine-origin h1n1 influenza viruses emergence of a novel swine-origin influenza a virus (s-oiv) h1n1 virus in humans influenza: emergence and control systemic cytokine responses in patients with influenza-associated encephalopathy inflammatory cytokines in vascular dysfunction and vascular disease identification of trypsin i as a candidate for influenza a virus and sendai virus envelope glycoprotein processing protease in rat brain mechanisms of matrix metalloprotease-9 upregulation and tissue destruction in various organs in influenza a virus infection activation of influenza a viruses by trypsin treatment host envelope glycoprotein processing proteases are indispensable for entry into human cells by seasonal and highly pathogenic avian influenza viruses dystroglycan is selectively cleaved at the parenchymal basement membrane at sites of leukocyte extravasation influenza virus-cytokine-protease cycle in the pathogenesis of vascular hyperpermeability in severe influenza molecular pathogenesis of influenza a virus infection and virus-induced regulation of cytokine gene expression active nf-jb signaling is a prerequisite for influenza virus infection nf-jb and virus infection: who controls whom tnf-induced mitochondrial damage: a link between mitochondrial complex i activity and left ventricular dysfunction acute encephalopathy associated with influenza and other viral infections molecular structure and assembly of the tight junction identification of oxidative stress and toll-like receptor 4 signaling as a key pathway of acute lung injury human influenza virus infection and apoptosis induction in human vascular endothelial cells fatal outcome of human influenza a (h5n1) is associated with high viral load and hypercytokinemia h5n1 infection of the respiratory tract and beyond: a molecular pathology study fatal avian influenza a (h5n1) in a child presenting with diarrhea followed by coma influenza virus m2 protein mediates escrt-independent membrane scission the protein-protein interaction map of helicobacter pylori development of a real-time rt-pcr for the detection of swine-lineage influenza a (h1n1) virus infections long-term storage and recovery of buccal cell dna from treated cards pandemic a(h1n1)2009 influenza virus detection by real time rt-pcr: is viral quantification useful? the effect of storage at different temperatures on the stability of hepatitis c virus rna in plasma samples impact of various handling and storage conditions on quantitative detection of hepatitis c virus rna blood storage at 4 degrees c-factors involved in dna yield and quality effects of prolonged storage of whole plasma or isolated plasma dna on the results of circulating dna quantification assays pathogenesis and transmission of swine-origin 2009 a(h1n1) influenza virus in ferrets viral tropism and the pathogenesis of influenza in the mammalian host influenza: pathogenesis and host defense cutting edge: stealth influenza virus replication precedes the initiation of adaptive immunity genetic and antiviral drug susceptibility profiles of pandemic a(h1n1)v influenza virus circulating in portugal observed association between the ha1 mutation d222g in the 2009 pandemic influenza a(h1n1) virus and severe clinical outcome world health organization. sequencing primers and protocol antiviral drug profile of seasonal influenza viruses circulating in portugal from predicting the antigenic structure of the pandemic (h1n1) 2009 influenza virus hemagglutinin genomic signature and mutation trend analysis of pandemic (h1n1) 2009 influenza a virus weekly update on oseltamivir resistance to influenza a (h1n1) 2009 viruses pandemic a(h1n1)2009 influenza virus detection by real-time rt-pcr: is viral quantification useful? virulence associated substitution d222g in hemagglutinin of 2009 pandemic influenza a(h1n1) virus affects receptor binding the first wave of pandemic influenza (h1n1) 2009 in germany: from initiation to acceleration new york city swine flu investigation team, et al. the severity of pandemic h1n1 influenza in the united states from rhinovirus delayed the circulation of the pandemic influenza a(h1n1)2009 virus in france dynamics and impact of the a(h1n1)2009 epidemic in metropolitan france clinical aspects of pandemic 2009 influenza a(h1n1) virus infection preliminary estimates of mortality and years of life lost associated with the 2009 a ⁄ h1n1 pandemic in the us and comparison with past influenza seasons oseltamivir resistance in adult oncology and hematology patients infected with pandemic (h1n1)2009 virus household transmission of the 2009 pandemic a ⁄ h1n1 influenza virus: elevated laboratoryconfirmed secondary attack nrates and evidence of assymptomatic infections time lines of infection and disease in human influenza: a review of volunteer challenge studies facemasks and hand hygiene to prevent influenza transmission in households comparative epidemiology of pandemic and seasonal influenza a in households viral shedding and clinical illness in naturally acquired influenza virus infections r: a language and environment for statistical computing the guinea pig as a transmission model for human influenza viruses in the control of epidemic and pandemic influenza 12-13 blocking interhost transmission of influenza virus by vaccination in the guinea pig model phenotypic properties resulting from directed gene segment reassortment between wild-type a ⁄ sydney ⁄ 5 ⁄ 97 influenza virus and the live attenuated vaccine strain emergence of a novel swine-origin influenza a (h1n1) virus in humans pneumonia and respiratory failure from swine-origin influenza a (h1n1) in mexico cumulative number of confirmed human cases of avian influenza a ⁄ (h5n1) reported to who factors involved in the aerosol transmission of infection and control of ventilation in healthcare premises airborne transmission of disease in hospitals noninvasive positive-pressure ventilation: an experimental model to assess air and particle dispersion airflow and droplet spreading around oxygen masks: a simulation model for infection control research coughing and aerosols. images in clinical medicine antimicrobial products registered for use against influenza a virus on hard surfaces. office of pesticide programs u.s. environmental protection agency antimicrobials division the genesis of a pandemic influenza virus influenza in the elderly -a mini review control of influenza in the long term care facility: a review of established approaches and newer options a study of the impact of influenza on the functional status of frail older people influenza in long-term care facilities: preventable, detectable, treatable mortality associated with influenza and respiratory syncytial virus in the united states causes of death australia 2007: leading causes of death australian government department of heath and ageing. pandemic (h1n1) 2009 update bulletins for pandemic (h1n1) 2009 update bulletins for 26 world health organization. pandemic (h1n1) 2009-update 112 factors associated with death or hospitalization due to pandemic 2009 influenza a (h1n1) infection in california influenza a (h1n1) 2009 antibodies in residents of new south wales, australia, after the first pandemic wave in the 2009 southern hemisphere winter qualification of the hemagglutination inhibition assay in support of pandemic influenza vaccine licensure managing outbreaks of viral respiratory infection in aged care facilities -challenges and difficulties during the first pandemic wave laboratory diagnosis of influenza virus infection antigenic and genetic characteristics of swine-origin 2009 a(h1n1) influenza viruses circulating in humans incidence of 2009 pandemic influenza a h1n1 infection in england: a cross-sectional serological study cross-reactive antibody responses to the 2009 pandemic h1n1 influenza virus in vitro and in vivo characterization of new swine-origin h1n1 influenza viruses cross-reacting antibodies against pandemic influenza a (h1n1) 2009 virus in elderly australians rhinovirus outbreak in a long term care facility for elderly persons associated with unusually high mortality a rhinovirus outbreak among residents of a long term care facility lessons from a respiratory illness outbreak in an aged-care facility detection and isolation of pandemic (h1n1) influenza virus infection among pediatric patients reporting diarrhea and influenza-like illness significance of seasonal influenza viruses in the stool of pediatric patients viral load in patients infected with pandemic h1n1 2009 influenza a virus haemagglutinin mutations responsible for the binding of h5n1 influenza a viruses to human-type receptors biedermann e references 1 world health organization writing group. nonpharmaceutical interventions for pandemic influenza, international measures avian influenza, including influenza a (h5n1), in humans: who interim infection control guideline for health care facilities inactivation of influenza a viruses in the environment and modes of transmission: a critical review quantitative method for evaluating virucidal activity of biocides used on hard surfaces nonpharmaceutical interventions for pandemic influenza, international measures avian influenza, including influenza a (h5n1), in humans: who interim infection control guideline for health care facilities inactivation of influenza a viruses in the environment and modes of transmission: a critical review distribution of airborne influenza virus and respiratory syncytial virus in an urgent care medical clinic characterisation of stability, cytotoxicity, and skin sensitisation and irritation for respirators with antiviral coating nonpharmaceutical interventions for pandemic influenza, international measures avian influenza, including influenza a (h5n1), in humans: who interim infection control guideline for health care facilities c600cf/0/infectioncontrol.pdf international organization for standardization. 10993: biological evaluation of medical devices, part 5: tests for cytotoxicity: in vitro methods effects of antiviral drugs on viral detection in influenza patients and on the sequential infection to their family members. serial examination by quick diagnosis (capiria) and virus culture intrafamilial transmission of influenza a and b analysis of infectious diseases data stochastic models for analysis of household transmission data: examining human-made transmission experiments factors influencing the effectiveness of oseltamivir and amantadine for the treatment of influenza. a japanese study of the multi-center study of the 2002 ⁄ 2003 influenza season clinical evaluation of an immunochromatography test kit, capilia flua, b for rapid diagnosis of influenza usefulness of a self-blown nasal discharge specimen for use with immunochromatography based influenza rapid antigen test. options for control of influenza vii 2010; 184 abstract book. options for the control of influenza vii strategies for containing an emerging influenza pandemic in southeast asia flute, a publicly available stochastic influenza epidemic simulation model the japanese experience with vaccinating schoolchildren against influenza mass vaccination of schoolchildren against influenza and its impact on the influenza-associated mortality rate among children in japan effect of vaccination of a school-age population upon the course of an a2-hong kong influenza epidemic herd immunity in adults against influenza-related illnesses with use of the trivalent-live attenuated influenza vaccine (caiv-t) in children the effect of mass influenza immunization in children on the morbidity of the unvaccinated elderly effect of influenza vaccination of children on infection rates in hutterite communities: a randomized trial efficacy of live attenuated and inactivated influenza vaccines in schoolchildren and their unvaccinated contacts in novgorod, russia effectiveness of influenza vaccination of day care children in reducing influenza-related morbidity among household contacts effectiveness of schoolbased influenza vaccination strategies for pandemic and seasonal influenza vaccination of schoolchildren in the united states strategy for distribution of influenza vaccine to high-risk groups and children estimating the impact of childhood influenza vaccination programmes in england and wales population-wide benefits of routine vaccination of children against influenza prevention and control of seasonal influenza with vaccines: recommendations of the advisory committee on immunization practices (acip) protective efficacy of seasonal influenza vaccination against seasonal and pandemic influenza virus infection during 2009 in hong kong partial protection of seasonal trivalent inactivated vaccine against novel pandemic influenza a ⁄ h1n1 2009: case-control study in mexico city association between the 2008-09 seasonal influenza vaccine and pandemic h1n1 illness during spring-summer 2009: four observational studies from canada cross-reactive antibody responses to the 2009 pandemic h1n1 influenza virus effectiveness of 2008-09 trivalent influenza vaccine against 2009 pandemic influenza a (h1n1) -united states seasonal influenza vaccine and protection against pandemic (h1n1) 2009-associated illness among us military personnel interim analysis of pandemic influenza (h1n1) 2009 in australia: surveillance trends, age of infection and effectiveness of seasonal vaccination pandemic (h1n1) 2009 risk for nurses after trivalent vaccination infection and death from influenza a h1n1 virus in mexico: a retrospective analysis birth weight, infant growth, and childhood body mass index: hong kong's children of 1997 birth cohort comparative epidemiology of pandemic and seasonal influenza a in households facemasks and hand hygiene to prevent influenza transmission in households: a cluster randomized trial partial protection against challenge with the highly pathogenic h5n1 influenza virus isolated in japan in chickens infected with the h9n2 influenza virus prior h1n1 influenza infection and susceptibility of cleveland family study participants during the h2n2 pandemic of 1957: an experiment of nature transmission of pandemic h1n1 influenza virus and impact of prior exposure to seasonal strains or interferon treatment prior infection with an h1n1 swine influenza virus partially protects pigs against a low pathogenic h5n1 avian influenza virus protection against a european h1n2 swine influenza virus in pigs previously infected with h1n1 and ⁄ or h3n2 subtypes infection of mice with a human influenza a ⁄ h3n2 virus induces protective immunity against lethal infection with influenza a ⁄ h5n1 virus evidence of a cross-protective immune response to influenza a in the cotton rat model heterosubtypic immunity to influenza a virus: where do we stand? safety and immunogenicity of as03b adjuvanted split virion versus non-adjuvanted whole virion h1n1 influenza vaccine in uk children aged 6 months-12 years: open label, randomised, parallel group, multicentre study options for live attenuated influenza vaccines (laiv) in the control of epidemic and pandemic influenza 12-13 pilot studies on recombinant cold-adapted live type a and b influenza virus vaccines development of cold-adapted recombinant live, attenuated influenza vaccines in usa and ussr clinical and epidemiological evaluation of a live, cold-adapted influenza vaccine for 3-14-year-olds revised requirements for influenza vaccine (live) global pandemic influenza action plan to increase vaccine supply characteristics of human influenza a ⁄ h1n1, a ⁄ h3n2 and b viruses isolated characteristics of human influenza a ⁄ h1n1, a ⁄ h3n2 and b viruses isolated characteristics of human influenza a ⁄ h1n1, a ⁄ h3n2 and b viruses isolated characteristics of human influenza a ⁄ h1n1, a ⁄ h3n2 and b viruses isolated dynamics of antigenic and genetic changes in the hemagglutinins of influenza a ⁄ h3n2 viruses of three consecutive seasons molecular epidemiology of a ⁄ h3n2 and a ⁄ h1n1 influenza virus during a single epidemic season in the united states an mf59-adjuvanted inactivated influenza vaccine containing a ⁄ panama ⁄ 1999 (h3n2) induced broader serological protection against heterovariant influenza virus strain a ⁄ fujian ⁄ 2002 than a subunit and a split influenza vaccine response of influenza vaccines against heterovariant influenza virus strains in adults with chronic diseases cross-protection by mf59 tmadjuvanted influenza vaccine: neutralizing and hemagglutinationinhibiting antibody activity against a (h3n2) drifted influenza viruses quantifying influenza vaccine efficacy and antigenic distance detection of antibody to avian influenza a (h5n1) virus in human serum by using a combination of serologic assays world health organization global influenza programme. who manual on animal influenza diagnosis and surveillance antigenic characterization of influenza b virus with a new microneutralisation assay: comparison to haemagglutination and sequence analysis seroprotection rate, meanfold increase, seroconversion rate: which parameter adequately expresses seroresponse to influenza vaccination? the adjuvanted influenza vaccines with novel adjuvants: experience with the mf59-adjuvanted vaccine experimental inoculation of pigs with pandemic h1n1 2009 virus and hi cross-reactivity with contemporary swine influenza virus antisera current status of live attenuated influenza vaccine in the united states for seasonal and pandemic influenza failure of protection and enhanced pneumonia with a us h1n2 swine influenza virus in pigs vaccinated with an inactivated classical swine h1n1 vaccine isolation and characterization of h3n2 influenza a virus from turkeys a new generation of modified live-attenuated avian influenza viruses using a two-strategy combination as potential vaccine candidates a dna transfection system for generation of influenza a virus from eight plasmids efficacy of inactivated swine influenza virus vaccines against the 2009 a ⁄ h1n1 influenza virus in pigs mortality associated with influenza and respiratory syncytial virus on the united states realities and enigma of human viral influenza: pathogenesis, epidemiology and control immune response after adjuvant mucosal immunization of mice with inactivated influenza virus stimulation of protective and cross-protective imunity against influenza b virus after adjuvant mucosal immunization of mice protective and cross-protective mucosal immunization of mice by influenza virus type a with bacterial adjuvant synergistic action of cholera toxin b subunit (and escherichia coli heat-labile toxin b subunit) and a trace amount of cholera whole toxin as an adjuvant for nasal influenza vaccine effects of intanasal administration of chlera toxine (or escherichia coli heat-labila enterotoxin b subunits)supplemented with a trace amount of the holotoxin on the brain the adjuvants mf59 and lt-k63 enhance the mucosal and systemic immunogenicity of subunit influenza vaccine administrated intranasaly in mice generation of influenza vaccine strains by genetic reassortment a description of the process of seasonal and h5n1 influenza vaccine virus selection and development pcr restriction analysis if genome composition and stability of cold-adapted reassortants live influenza vaccines options for live attenuated influenza vaccines (laiv) in the control of epidemic and pandemic influenza 12-13 characterization of an influenza a h5n2 reassortant as a candidate for live-attenuated and inactivated vaccines against highly pathogenic h5n1 viruses with pandemic potential are there alternative avian influenza viruses for generation of stable attenuated avian-human influenza a reassortant viruses? organization of production, quality control and provision of the republic of kazakhstan with the a ⁄ h5n1 vaccine development of diagnostics and vaccines against influenza a ⁄ h5n1 and a ⁄ h1n1 is the basis for biological safety and protection of population in kazakhstan maior results of developing a technology for production of a pandemic influenza a ⁄ h5n1 vaccine» abstract book ''options for the control of influenza vii abstract book international scientific conference of students, graduates and young scientists ''lomonosov isolation of influenza viruses in cell cultures and chicken embryos and their identification development of quality and safety control methods of the inactivated vaccine based on the strain a ⁄ astanarg ⁄ 6:2 ⁄ 2009 (h5n1 control methods of medical immunobiological preparations injected to humans. guidelines, muk 4.1/4.2.588-96 (adopted by goskomsanepidnadzor rf ob 31 scientific center for expertise of medical products, moscow. russian national pharmacopeia. ist edn the guidance for pre-clinical testing of pharmaceutical substances for safety and effectiveness (regulations of good laboratory practice in the russian federation -glp). manual on experimental (pre-clinical) assay of new pharmaceutical substances. moscow: ministry of public health ⁄ department of quality, effectiveness and safety control of medicines ⁄ scientific center of expert examination and national control of medicines ⁄ pharmacological national committee manual on experimental (pre-clinical) assay of new pharmaceutical substances. moscow: medline status of the development of influenza a ⁄ h5n1 vaccines in the world and in russia approval of instruction for preclinical testing of pharmacological and medical substances in the repubilic of kazakhstan preclinical testing of new medical immunobiological substances. basic principles. moscow: ussr ministry of health order and methods of control of vaccine immunobiological safety. general methodic principles. moscow: ussr ministry of health planning and performance of clinical study of medicines federal pharmaceutical substances law. #122-fz russian federation national standard. good clinical practice. # 232-st order of the minister of health of russian federation # 266 signed 19.06.2003. rules for clinical practice in russian federation a single-dose influenza a (h5n1) vaccine safe and immunogenic in adult and elderly patients: an approach to pandemic vaccine development cross-reactive immunity to clade 2 strains of influenza virus a subtype h5n1 induced in adults and elderly patients by fluval, a prototype pandemic influenza virus vaccine derived by reverse genetics, formulated with a phosphate adjuvant, and directed to clade 1 strains develop of domestic vaccinal and diagnostic preparations for influenza a ⁄ h5n1 and a ⁄ h1n1 as a basic principle of biological safety and security for the population of kazakhstan guidelines: ''isolation of influenza viruses in cell cultures and chicken embryos and their identification''. moscow: a-print validation of control methods for chemical and physicochemical characteristics of mibp quality: organization the guidance for pre-clinical testing of pharmaceutical substances for safety and effectiveness (regulations of good laboratory practice in the russian federation -glp). manual on experimental (pre-clinical) assay of new pharmaceutical substances. moscow: ministry of public health ⁄ department of quality, effectiveness and safety control of guidance on assessment of allergenic characteristics of pharmaceutical substances on adoption of the instructions for performance of pre-clinical assays and tests of pharmaceutical substances and medicines in the republic of kazakhstan national pharmacopoeia of the republic of kazakhstan planning and performance of clinical study of medicines federal law on medicines no.122-aii dated national standard of the russian federation ''proper clinical practice'' no. 232-st guidance on clinical practice in the russian federation'' (in russian) aluminum hydroxide adjuvants activate caspase-1 and induce il-1beta and il-18 release safety and immunogenicity of a 2009 pandemic influenza a h1n1 vaccine when administered alone or simultaneously with the seasonal influenza vaccine for the 2009-10 influenza season: a multicentre, randomised controlled trial current status of live attenuated influenza vaccine in the united states for seasonal and pandemic influenza a simple method of estimation fifty per cent endpoints use of animal models to understand the pandemic potential of highly pathogenic avian influenza viruses evaluation of replication and pathogenicity of avian influenza a h7 subtype viruses in a mouse model results of preclinical studies of live monovalent influenza vaccine influvir swine influenza a (h1n1) infection in two children -southern california comparison of rna hybridization, hemagglutination assay, titration of infectious virus and immunofluorescence as methods for monitoring influenza virus replication in vitro concepts and procedures for laboratory-based influenza surveillance. us department of health and human services and pan-american health organization detection of antibody to avian influenza a (h5n1) virus in human serum by using a combination of serologic assays fda ⁄ nih ⁄ who public workshop on immune correlates of protection against influenza a viruses in support of pandemic vaccine development coinfection of avian influenza virus (h9n2 subtype) with infectious bronchitis live vaccine co-infection of staphylococcus aureus or haemophilus paragallinarum exacerbates h9n2 influenza a virus infection in chickens human infection with an avian h9n2 influenza a virus in hong kong in 2003 avian-to-human transmission of h9n2 subtype influenza a viruses: relationship between h9n2 and h5n1 human isolates human infection with influenza h9n2 universal primer set for the full-length amplification of all influenza a viruses phylogenetic analysis of hemagglutinin and neuraminidase genes of h9n2 viruses isolated from migratory ducks the neighbor-joining method: a new method for reconstructing phylogenetic trees application of a microtechnique to viral serological investigations biological activity of monoclonal antibodies to operationally defined antigenic regions on the hemagglutinin molecule of a ⁄ seal ⁄ massachusetts ⁄ 1 ⁄ 80 (h7n7) influenza virus development of vaccine strains of h5 and h7 influenza viruses isolation of ortho-and paramyxoviruses from feral birds in hokkaido head-to-head comparison of four nonadjuvanted inactivated cell culture-derived influenza vaccines: effect of comparison, spatial organization and immunization route on the immunogenicity in a murine challenge model evaluation of growth, genetic, and antigenic characteristics of pandemic h1n1 viruses isolated and passaged in qualified mdck suspension cells a (h1n1) influenza virus pandemic: a review influenza vaccine: from surveillance through production to protection trivalent mdck cell culture-derived influenza vaccine optaflu (novartis vaccines) validation of the safety of mdck cells as a substrate for the production of a cell-derived influenza vaccine genetic composition of a high-yielding influenza a virus recombinant: a vaccine strain against ''swine'' influenza rescue of influenza a virus from recombinant dna of health, public health service. concepts and procedures for laboratory based influenza surveillance a rapid method for immunotitration of influenza viruses using flow cytometry generation of live attenuated novel influenza a ⁄ california ⁄ 7 ⁄ 09 (h1n1) vaccines with high yield in embryonated chicken eggs haemagglutinin quantification and identification of influenza a&b strains propagated in per.c6 cells: a novel rp-hplc method comparison of egg and high yielding mdck cell-derived live attenuated influenza virus for commercial production of trivalent influenza vaccine: in vitro cell susceptibility and influenza virus replication kinetics in permissive and semi-permissive cells effect of neuraminidase antibody on hong kong influenza infection-permissive immunization with influenza virus neuraminidase prevents weight loss in infected mice protection of mice with recombinant influenza virus neuraminidase cross-reactive neuraminidase antibodies afford partial protection against h5n1 in mice and are present in unexposed humans note for guidance on harmonization of requirements for influenza vaccines who meeting on the role of neuraminidase in inducing protective immunity against influenza infection who manual on animal influenza diagnosis and surveillance [electronic resource] ⁄ world health organization, department of communicable disease surveillance and response measurement of anti-influenza neuraminidase antibody using a peroxidase-linked lectin and microtitre plates coated with natural substrates detection of antibody to avian influenza a(h5n1) virus in human serum by using a combination of serologic assays large trials confirm immunogenicity of h1n1 vaccines pre-existing immunity against swine-origin h1n1 influenza viruses in the general human population yuelong shu references 1 novel swine-origin influenza a (h1n1) virus investigation team. emergence of a novel swine-origin influenza a (h1n1) virus in humans serum crossreactive antibody response to a novel influenza a (h1n1) virus after vaccination with seasonal influenza vaccine detection of antibody to avian influenza a (h5n1) virus in human serum by using a combination of serologic assays concepts and procedures for laboratory-based influenza surveillance manual on animal influenza diagnosis note for guidance on harmonization of requirements for influenza vaccines (cpmp ⁄ bwp ⁄ 214 ⁄ 96). european agency for the evaluation of medicinal products cross-reactive antibody responses to the 2009 pandemic h1n1 influenza virus serologic survey of pandemic (h1n1) 2009 virus influenza a(h1n1) seroconversion rates and risk factors among distinct adult cohorts in singapore swine influenza a outbreak reflections on the 1976 swine flu vaccination program the 1918 flu and other influenza pandemics: ''over there'' and back again evidence of an absence: the genetic origins of the 1918 pandemic influenza virus the 1918 influenza virus: a killer comes into view antigenic and genetic characteristics of swine-origin 2009 a(h1n1) influenza viruses circulating in humans pathogenesis and immunogenicity of influenza viruses with genes from the 1918 pandemic virus lifetime of plasma cells in the bone marrow humoral immunity due to long-lived plasma cells no one is naive: the significance of heterologous t-cell immunity infection outcomes in patients with common variable immunodeficiency disorders: relationship to immunoglobulin therapy over 22 years signaling by toll-like receptors 8 and 9 requires bruton's tyrosine kinase antiviral cytotoxic t-cell memory by vaccination with recombinant listeria monocytogenes cell-mediated immune response of human lymphocytes to influenza a ⁄ ussr (h1n1) virus infection severe deficiency of switched memory b cells (cd27(+)igm(-)igd(-)) in subgroups of patients with common variable immunodeficiency: a new approach to classify a heterogeneous disease an early humoral immune response in peripheral blood following parenteral inactivated influenza vaccination multifunctional th1 cells define a correlate of vaccine-mediated protection against leishmania major note for guidance on harmonization of requirements for influenza vaccines (cpmp ⁄ bwp ⁄ 214 ⁄ 96) active vaccination in patients with common variable immunodeficiency (cvid) serum bactericidal antibody response to serogroup c polysaccharide meningococcal vaccination in children with primary antibody deficiencies specific anti-influenza virus antibody production in vitro by lymphocytes from a subset of patients with hypogammaglobulinemia shu a a state key laboratory for infectious disease control and prevention, national institute for viral disease control and prevention, chinese center for disease control and prevention, beijing, china. b state key laboratory for molecular virology and genetic engineering, national institute for viral disease control and prevention roles of adjuvant igv combined with influenza m2e based virus like particle in the immune response and cross protection usa) 1ae5 h at 37°c. tetramethylbenzidine (tmb) was applied, and the optical den epidemiology and pathogenesis of influenza epidemiology and virology of influenza illness influenza vaccine: the challenge of antigenic drift evolving strategies for the prevention of influenza infection: potential for multistrains targeting a ''universal human influenza'' a vaccine universal influenza a vaccine: optimization of m2-based constructs the epitope recognized by a monoclonal antibody in influenza a virus m2 protein is immunogenic and confers immune protection high-epitope density in a single recombinant protein molecule of the extracellular domain of influenza a virus m2 protein significantly enhances protective immunity pre-clinical study of influenza virus a m2 peptide conjugate vaccines in mice, ferrets, and rhesus monkeys induction of influenza type a virus-specific resistance by immunization of mice with a synthetic multiple antigenic peptide vaccine that contains ectodomains of matrix protein 2 a universal influenza a vaccine based on the extracellular domain of the m2 protein protective immunity against influenza a virus induced by immunization with dna plasmid containing influenza m gene protection against influenza virus challenged by topical application of influenza dna vaccine influenza a vaccine based on the extracellular domain of m2: weak protection mediated via antibody-dependent nk cell activity t cell immunoglobulin mucin-3 crystal structure reveals a novel ligand binding surface structures of t cell immunoglobulin mucin protein 4 show a metal-ion-dependent ligand binding site where phosphatidylserine binds structures of t cell immunoglobulin mucin receptors 1 and 2 reveal mechanisms for regulation of immune responses by the tim receptor family the tim gene family: emerging roles in immunity and disease th1-specific cell surface protein tim-3 regulates macrophage activation and severity of and autoimmune disease identification of a surface glycoprotein on african green monkey kidney cells as a receptor for hepatitis a virus the igv domain of human b7-2 (cd86) is sufficient to co-stimulate t lymphocytes and induce cytokine secretion vaccination with cell immunoglobulin mucin-1 antibodies and inactivated influenza enhances vaccine-specific lymphocyte proliferation, interferon-r production and cross-strain reactivity incorporation of membrane-anchored flagellin into influenza virus-like particles enhances the breadth of immune response virus-like particle induces protective immunity against homologous and heterologous strains of influenza virus site accessed on 1st study designs for timely estimation of influenza vaccine effectiveness using european sentinel practitioner networks a new influenza surveillance system in france: the ile-de-france ''grog''. 1. principles and methodology site accessed on 1st i-move'' towards monitoring seasonal and pandemic influenza vaccine effectiveness: lessons learnt from a pilot multi-centric case-control study in europe population and risk group uptake of h1n1 influenza vaccine in mainland france 2009-2010: results of a national vaccination campaign of global health and emerging pathogens ns deletions convert the 2009-h1n1 pandemic virus into a live attenuated vaccine. influenza and other respiratory viruses the influenza virus enigma vaccine production capacity for seasonal and pandemic (h1n1) 2009 influenza interferon-induced isg15 conjugation inhibits influenza a virus gene expression and replication in human cells the multifunctional ns1 protein of influenza a viruses influenza a and b viruses expressing altered ns1 proteins: a vaccine approach multiple anti-interferon actions of the influenza a virus ns1 protein the ns1 protein of a human influenza virus inhibits type i interferon production and the induction of antiviral responses in primary human dendritic and respiratory epithelial cells immunization with live attenuated influenza viruses that express altered ns1 proteins results in potent and protective memory cd8+ t-cell responses plasmacytoid dendritic cells delineate immunogenicity of influenza vaccine subtypes regulation of adaptive immunity by the innate immune system the influenza virus ns1 protein: inhibitor of innate and adaptive immunity single-reaction genomic amplification accelerates sequencing and vaccine production for classical and swine origin human influenza a viruses unidirectional rna polymerase i-polymerase ii transcription system for the generation of influenza a virus from eight plasmids avian influenza (h5n1) viruses isolated from humans in asia in 2004 exhibit increased virulence in mammals transmission and pathogenesis of swine-origin 2009 a(h1n1) influenza viruses in ferrets and mice a naturally occurring deletion in its ns gene contributes to the attenuation of an h5n1 swine influenza virus in chickens future influenza vaccines and use of genetic recombinants eight-plasmid system for rapid generation of influenza virus vaccines rescue of influenza b virus from eight plasmid high yield reassortants of influenza type b viruses np gene obtained from high yield donor. options for control of influenza vi. abstract #330 summary of status of development and availability of b ⁄ brisbane ⁄ 60 ⁄ 2008-like candidate vaccine viruses and potency testing reagent reassortants of influenza b viruses for use in vaccines: an evaluation preparation of influenza b virus recombinant strains genetic and phenotypic analysis of reassortants of high growth and low growth strains of influenza b virues pandemic influenza in russia: detection and molecular references evolutionary pattern of pandemic influenza (h1n1) 2009 virus in the late phases of the 2009 pandemic antigenic and genetic characteristics of swine-origin 2009 a(h1n1) influenza viruses circulating in humans effects of egg adaptation on the receptor-binding properties of human influenza a and b viruses the early diversification of influenza a ⁄ h1n1pdm update on influenza a (h1n1) 2009 monovalent vaccines clinical and epidemiologic characteristics of 3 early cases of influenza a pandemic options for the control of influenza vii virus infection, people's republic of china genome evolution of novel influenza a (h1n1) viruses in humans introduction and transmission of 2009 pandemic influenza a (h1n1) virus -kenya emergence of a novel swine-origin influenza a (h1n1) virus in humans first season of 2009 h1n1 influenza origins and evolutionary genomics of the 2009 swine-origin h1n1 influenza a epidemic modeling gene sequences over time in 2009 h1n1 influenza a virus populations differentiation of two distinct clusters among currently circulating influenza a(h1n1)v viruses conversion of mdck cell line to suspension culture by transfecting with human siat7e gene and its application for influenza virus production trivalent mdck cell culture-derived influenza vaccine optaflu (novartis vaccines) use of mdck cells for production of live attenuated influenza vaccine flow cytometric monitoring of influenza a virus infection in mdck cells during vaccine production cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences seroprevalence following the second wave of pandemic molecular evolution of human influenza a viruses in a local area during eight influenza epidemics from evaluation of hemagglutinin subtype 1 swine influenza viruses from the united states the early molecular epidemiology of the swine-origin a ⁄ h1n1 human influenza pandemic the genomic and epidemiological dynamics of human influenza a virus reconstructing the initial global spread of a human influenza pandemic: a bayesian spatial-temporal model for the global spread of h1n1pdm mild form of 2009 h1n1 influenza infection detected by active surveillance: implications for infection control the pandemic 2009 (h1n1) swine influenza virus is mild compared to the pandemic 1918 (h1n1) virus because of a prolineto-serine substitution in the receptor-binding site of its hemagglutinin -a hypothesis pathogenesis and transmission of swine-origin 2009 a(h1n1) influenza virus in ferrets investigation of the first cases of human-to-human infection with the new swine-origin influenza a (h1n1) virus in canada antigenic and genetic characteristics of swine-origin 2009 a(h1n1) influenza viruses circulating in humans molecular evolution of hemagglutinin genes of h1n1 swine and human influenza a viruses antigenic structure of influenza virus haemagglutinin defined by hybridoma antibodies the predicted antigenicity of the haemagglutinin of the 1918 spanish influenza pandemic suggests an avian origin genetically destined potentials for nlinked glycosylation of influenza virus hemagglutinin global patterns of influenza a virus in wild birds antigenic and genetic characteristics of swine-origin 2009 a (h1n1) influenza viruses circulating in humans translatorx: multiple alignment of nucleotide sequences guided by amino acid translations raxml-iii: a fast program for maximum likelihood-based inference of large phylogenetic trees flugenome: a web tool for genotyping influenza a virus beast: bayesian evolutionary analysis by sampling trees origin and evolution of influenza virus hemagglutinin genes panorama phylogenetic diversity and distribution of type a influenza virus early alterations of the receptor-binding properties of h1, h2, and h3 avian influenza virus hemagglutinins after their introduction into mammals human influenza a h5n1 virus related to a highly pathogenic avian influenza virus origin and evolutionary genomics of the 2009 swine-origin h1n1 influenza a epidemic keywords novel influenza viruses, oseltamivir resistance phylogenetic analyses of pandemic influenza viruses textbook of influenza mapping the antigenic and genetic evolution of influenza virus cdc protocol of realtime rt-pcr for swine influenza a(h1n1) revision 1 the hemagglutination and hemagglutination ingibitions test. who collaborating center for surveillance molecular evolutionary genetics analysis (mega) software version 4.0 pandemic influenza: an inconvenient mutation large-scale sequencing of human influenza reveals the dynamic nature of viral genome evolution origin and evolutionary genomics of the 2009 swine-origin h1n1 influenza a epidemic antigenic and genetic characteristics of swine-origin 2009 a (h1n1) influenza viruses circulating in humans genetic subtyping using cluster analysis panorama phylogenetic diversity and distribution of type a influenza virus continuing progress towards a unified nomenclature for the highly pathogenic h5n1 avian influenza viruses: divergence of clade 2.2 viruses the influenza virus resource at the national center for biotechnology information muscle: multiple sequence alignment with high accuracy and high throughput multiple alignment of dna sequences with mafft raxml-iii: a fast program for maximum likelihood-based inference of large phylogenetic trees flugenome: a web tool for genotyping influenza a virus evolution of influenza viral neuraminidase (na) genes revealed by large-scale sequence analysis. proceedings of the options for the control of influenza vii a quantitative genotype algorithm reflecting h5n1 avian influenza niches xiu-feng wan a a department of basic sciences complete composition vector, influenza a virus, minimum spanning tree, phylogeny, progenitor ipminer: a progenitor gene identifier for influenza a virus avian-to-human transmission of the pb1 gene of influenza a viruses in the 1957 and 1968 pandemics triple-reassortant swine influenza a (h1) in humans in the united states clustal w: improving the sensitivity of progressive multiple sequence alignment through sequence weighing, positionspecific gap penalties and weight matrix choice muscle: multiple sequence alignment with high accuracy and high throughput coffee: an objective function for multiple sequence alignments the neighbour-joining method: a new method for reconstructing phylogenetic trees mrbayes: bayesian inference of phylogenetic trees basic local alignment search tool a quantitative genotype algorithm reflecting h5n1 avian influenza niches computational identification of reassortments in avian influenza viruses ubiquitous reassortments in influenza a viruses phylip -phylogeny inference package (version 3.2) pandemic influenza 2009 in russia: isolation, antigenic produced in our laboratory. normal equine serum was purchased from sigma (usa) comparative data on the development of influenza pandemic in russia and in the other european countries (in russian). materials of the ii annual all-russian congress on infectious diseases isolation on 24.05.2009 and deposition in the state virus collection of the first strain a ⁄ moscow ⁄ 01 ⁄ 2009 (h1n1)swl, similar to the swine virus a(h1n1) from the first patient diagnosed in moscow isolation of influenza viruses in cell cultures and in chicken embryos and their identification. methods guidance (in russian) molecular genetic characterization of the strains of pandemic influenza a(h1n1)v isolated on the territory of russian federation in 2009 (in russian). materials of the ii annual all-russian congress on infectious diseases genetic and antigenic analyses of influenza a (h1n1) viruses, 1986-1991 sialic acid receptor detection in the human respiratory tract: evidence for widespread distribution of potential binding sites for human and avian influenza viruses pandemic influenza a(h1n1) virus mutations reported to be associated with severe disease antigenic and genetic characteristics of swine-origin 2009 a(h1n1) influenza viruses circulating in humans barr b,c a novartis vaccines and diagnostics gmbh, marburg, germany. b who collaborating centre for reference and research on influenza co-infecting viruses in pandemic h1n1 and seasonal influenza positive human respiratory samples emergence of a novel swine-origin influenza a (h1n1) virus in humans simultaneous detection and high-throughput identification of a panel of rna viruses causing respiratory tract infections comparison of multiple commercial assays for the detection of swine-origin influenza a (h1n1) virus (soiv) evidence from multiplex molecular assays for complex multipathogen interactions in acute respiratory infections performance comparison of res-plex ii and xtag rvp fast for detecting respiratory viruses clinical virology symposium communityacquired respiratory co-infection (carc) in critically ill patients infected with pandemic 2009 influenza a (h1n1) virus infection bacterial co-infections in lung tissue specimens from fatal cases of 2009 pandemic influenza a (h1n1) -united states a quantitative risk assessment of exposure to adventitious agents in a cell culture-derived subunit influenza vaccine john's hopkins bloomberg school of public health design of an automated laboratory for high-throughput influenza surveillance human influenza surveillance: the demand to expand influenza: an emerging disease avian-to-human transmission of the pb1 gene of influenza a viruses in the 1957 and 1968 pandemics proinflammatory cytokine responses induced by influenza a (h5n1) viruses in primary human alveolar and bronchial epithelial cells induction of proinflammatory cytokines in human macrophages by influenza a (h5n1) viruses: a mechanism for the unusual severity of human disease? influenza h5n1 and h1n1 virus replication and innate immune responses in bronchial epithelial cells are influenced by the state of differentiation mapping of the two overlapping genes for polypepetides ns1 and ns1 on rna segment 8 of influenza virus genome sequences of mrnas derived from genome rna segment 7 of influenza virus: collinear and interrupted mrnas code for overlapping proteins influenza virus ns1 protein inhibits pre-mrna splicing and blocks mrna nucleocytoplasmic transport the influenza virus ns1 protein: a novel inhibitor of pre-mrna splicing identification of cis-acting intron and exon regions in influenza virus ns1 mrna that inhibit splicing and cause the formation of aberrantly sedimenting presplicing complexes identification of a second protein encoded by influenza c virus rna segment 6 influenza c virus ns1 protein upregulates the splicing of viral mrnas identification of an amino acid residue on influenza c virus m1 protein responsible for formation of the cord-like structures of the virus a mutation on influenza c virus m1 protein affects virion morphology by altering the membrane affinity of the protein detection of ion channel activity in xenopus laevis oocytes expressing influenza c virus cm2 protein evidence that the cm2 protein of influenza c virus can modify the ph of the exocytic pathway of transfected cells the ion channel activity of the influenza virus m2 protein affects transport through the golgi apparatus intracellular localization of influenza c virus ns2 protein (nep) in infected cells and its incorporation into virions receptor specificity, host range and pathogenicity of influenza viruses replication, pathogenesis and transmission of pandemic (h1n1) 2009 virus in non-immune pigs world health organization. preliminary review of d222g amino acid substitution in the haemagglutinin of pandemic influenza a (h1n1) 2009 viruses observed association between the ha1 mutation d222g in the 2009 pandemic influenza a(h1n1) virus and severe clinical outcome quasispecies of the d225g substitution in the hemagglutinin of pandemic influenza a(h1n1) virus from patients with severe disease in hong kong a possible association of fatal pneumonia with mutations of pandemic influenza a ⁄ h1n1 sw1 virus in the receptor-binding site of the ha1 subunit severe outcome of influenza a ⁄ h1n1 ⁄ 09v infection associated with 222g ⁄ n polymorphisms in the haemagglutinin: a multicenter study influenza pandemic caused by highly conserved viruses with two receptor-binding variants overexpression of the alpha-2,6-sialyltransferase in mdck cells increases influenza virus sensitivity to neuraminidase inhibitors human and avian influenza viruses target different cell types in cultures of human airway epithelium avian-virus-like receptor specificity of the hemagglutinin impedes influenza virus replication in cultures of human airway epithelium receptor-binding specificity of pandemic influenza a (h1n1) 2009 virus determined by carbohydrate microarray fields virology. philadelphia, pa: lippincott williams & wilkins the molecular virology and reverse genetics of influenza c virus identification of a second protein encoded by influenza c virus rna segment 6 identification of a 374 amino acid protein encoded by rna segment 6 of influenza c virus influenza c virus cm2 protein is produced from a 374 amino acid protein (p42) by signal peptidase cleavage a mutation on influenza c virus m1 protein affects virion morphology by altering the membrane affinity of the protein influenza c virus cm2 integral membrane glycoprotein is produced from a polypeptide precursor by cleavage of an internal signal sequence evidence that the matrix protein of influenza c virus is coded for by a spliced mrna functional properties of the virus ion channels the cm2 protein of influenza c virus is an oligomeric integral membrane glycoprotein structurally analogous to influenza a virus m2 and influenza b virus nb proteins characterization of a second protein (cm2) encoded by rna segment 6 of influenza c virus phosphorylation of influenza c virus cm2 protein the sites for fatty acylation, phosphorylation and intermolecular disulphide bond formation of influenza c virus cm2 protein identification of an amino acid residue on influenza c virus m1 protein responsible for formation of the cord-like structures of the virus a human melanoma cell line highly susceptible to influenza c virus antigenic characterization of the nucleoprotein and matrix protein of influenza c virus with monoclonal antibodies construction of an antigenic map of the haemagglutinin-esterase protein of influenza c virus the synthesis of polypeptides in influenza c virus-infected cells new low-viscosity overlay medium for viral plaque assays the influenza virus m2 protein cytoplasmic tail interacts with the m1 protein and influences virus assembly at the site of virus budding the cytoplasmic tail of the influenza a virus m2 protein plays a role in viral assembly the influenza a virus m2 cytoplasmic tail is required for infectious virus production and efficient genome packaging distinct domains of the influenza a virus m2 protein cytoplasmic tail mediate binding to the m1 protein and facilitate infectious virus production influenza virus m2 ion channel protein is necessary for filamentous virion formation influenza h5n1 virus infection of polarized human alveolar epithelial cells and lung microvascular endothelial cells das181 inhibits h5n1 influenza virus infection of human lung tissues receptor binding specificity of recent human h3n2 influenza viruses differential onset of apoptosis in avian influenza h5n1 and seasonal h1n1 virus infected human bronchial and alveolar epithelial cells: an in vitro and ex vivo study human influenza virus a ⁄ hongkong ⁄ 156 ⁄ 97 (h5n1) infection induction of proinflammatory cytokines in human macrophages by influenza a (h5n1) viruses: a mechanism for the unusual severity of human disease? proinflammatory cytokine responses induced by influenza a (h5n1) viruses in primary human alveolar and bronchial epithelial cells differential onset of apoptosis in influenza a virus h5n1-and h1n1-infected human blood macrophages avian flu: influenza virus receptors in the human airway haemagglutinin mutations responsible for the binding of h5n1 influenza a viruses to humantype receptors an avian influenza h5n1 virus that binds to a human-type receptor evolution of highly pathogenic h5n1 avian influenza viruses in vietnam between evolutionary dynamics and emergence of panzootic h5n1 influenza viruses writing committee of the second world health organization consultation on clinical aspects of human infection with avian influenza a (h5n1) virus recent avian h5n1 viruses exhibit increased propensity for acquiring human receptor specificity a simple screening assay for receptor switching of avian influenza viruses glycan topology determines human adaptation of avian h5n1 virus hemagglutinin h5n1 chicken influenza viruses display a high binding affinity for neu5acalpha2-3galbeta1-4(6-hso3)glcnac-containing receptors a strain of human influenza a virus binds to extended but not short gangliosides as assayed by thin-layer chromatography overlay search for additional influenza virus to cell interactions avian flu: isolation of drug-resistant h5n1 virus the surface glycoproteins of h5 influenza viruses isolated from humans, chickens, and wild aquatic birds have distinguishable properties this study was supported by the li ka shing foundation, the national institutes of health (niaid contract hhsn266200700005c), and the area of excellence scheme of the university grants committee (grant aoe ⁄ m-12 ⁄ 06) of the hong kong sar government. this work was supported by the national institute of allergy and infectious diseases (niaid) contract hhsn266200700005c, the li ka shing foundation, and we thank all french and vietnamese field staff involved in the data collection in viet nam for their enthusiasm and support and we are grateful to the pig farmers participating in the study for their cooperation and patience. this study was a part of the gripavi project and was funded by the french ministry of foreign affairs. this research was supported in part by the national institute of allergy and infectious diseases (niaid) contract hhsn26600700005c and the area of excellence scheme of the university grants commission (grant aoe ⁄ m-12 ⁄ 06) of the hong kong sar government. we acknowledge the food and environmental hygiene department of hong kong for facilitating the study. this work was supported by the national institute of allergy and infectious diseases (niaid) contract hhsn266200700005c, the li ka shing foundation, and the area of excellence scheme of the university grants committee (grant aoe ⁄ m-12 ⁄ 06) of the hong kong sar government. we gratefully acknowledge our colleagues from iiii, shantou university and skleid, hku for their excellent technical assistance. the study was supported by the rfcid commissioned study (lab#16) from research fund secretariat, food and health bureau, hong kong sar; area of excellence scheme of the university grants committee (grant aoe ⁄ m-12 ⁄ 06), hong kong sar; and by niaid contract (sjceirs, hhsn26620070005c), nih, usa.ferrets in all groups inoculated with a ⁄ turkey ⁄ 15 ⁄ 06 virus survived the infection and were observed once daily for 14 days. below lower limit of detection (<0ae75 log 10 eid 50 ⁄ ml).statistical cutoff of ic 50 values for nai susceptibility, determined by x 0ae75 + 3iqr. outliers with ic 50 above this cutoff and >10 times the mean ic 50 for each drug were characterized as extreme outliers; those with known drug-resistance mutations such as h275y were classified as resistant and analyzed separately. h275 wildtype, oseltamivir-susceptible isolates. h275y variants, oseltamivir-resistant virus isolates. iqr, interquartile ranges; nai, neuraminidase inhibitors. we wish to thank our collaborators in the who global influenza surveillance network and united states public health laboratories for the submission of virus isolates and clinical specimens. we also thank our colleagues from the virus reference team and the influenza sequence activity, influenza division, cdc, for their valuable technical assis-the findings and conclusions of this report are those of the authors and do not necessarily represent the views of the centers for disease control and prevention (cdc). we are indebted to yonas araya, theresa wolter, and ivan gomez-osorio for their excellent laboratory techniques and animal handling assistance. we would like to thank andrea ferrero for her laboratory managerial skills. this research was possible through funding by the cdc-hhs grant (1u01ci000355), niaid-nih grant, (r01ai052155), csrees-usda grant (1865-05523), and niaid-nih contract (hhsn266186700010c). we thank c bazzoli for advice. this work was supported by a grant from the european union fp7 project flu-modcont (no. 20160). we thank staff at seoul, incheon, daejeon, gwangju, gangwon, and jeonbuk provincial research institute of health and environments for their laboratory testing. additionally, we would like to acknowledge the contributions of participating sentinel doctors for evaluating the new rat kit. this study was supported by a grant from the korea cdc. we thank roche applied science for providing the materials and equipment for this evaluation. this research was supported in part by the national institute of allergy and infectious diseases (niaid) contract hhsn26600700005c and the area of excellence scheme of the university grants committee (grant aoe ⁄ m-12 ⁄ 06) of the hong kong sar government. the authors would like express their sincere thanks to cdc, usa for supporting the routine surveillance of ili in we would like to acknowledge the australian red cross blood service (the blood service) and the australian government, which fully fund the blood service for the provision of blood products and services to the australian community. we also wish to thank the donors and staff of the blood service, who have assisted in provision of specimens for testing in this protocol, as well as the staff at the who we are grateful to liping long for her assistance in map generation. this project was supported by nih niaid 1rc1ai086830. cz is supported partially by canadian nserc postdoc fellowship. the authors thank the national investigation team based at the national institute of health (istituto superiore di sanita'), italy (in particular antonino bella, maria cristina rota, stefania salmaso) for providing their support in data collection, and the european union this study was supported in part by a grant-in-aid (21249061) and the special coordination funds for promoting science and technology of ministry of education, science, sports and culture of japan. this study was supported in part by a grant-in-aid from the ministry of education, science, and culture of japan (21249061) and the special coordination funds for promoting science and technology of mext of japan. the work described here was supported by phs grant ai-66349 (jam) and alsac. we thank all authors for their participation in data gathering and analysis, and in writing this manuscript. the studies were funded by gsk consumer healthcare, and gsk investigators were involved in all stages of the study conduct and analysis. py, po, dw and kb are employees of glaxosmithkline. this study was funded by glaxosmithkline. we thank all authors for their participation in data gathering and analysis, and in writing this manuscript. the studies were funded by gsk consumer healthcare, and gsk investigators were involved in all stages of the study conduct and analysis. this study was funded by glaxosmithkline. we thank all authors for their participation in data gathering and analysis, and in writing this manuscript. the stud authors are thankful to path for the financial support of this research. we would like to acknowledge jessica d'amico and dr. rick bright of path for their editorial review. this study was supported by path. the authors would like to thank rick bright, jessica d'amico, and vadim tsvetnitsky for editing assistance. the we thank dr. m. enami (kanazawa university) for generously providing plasmids containing cdnas to influenza a virus m and ns genes. we also gratefully thank dr. r. sho (department of public health, yamagata university faculty of medicine) for statistical analysis. some data shown in this study have also been presented in the reference paper. 7 this work was supported in part by a grant-in-aid for scientific research from the ministry of education, culture, sports, science, and technology, japan, takeda science foundation, terumo life science foundation, and a grant-in-aid from the global coe program of the japan society for the promotion of science. we thank markus eickmann for his help in isolation and initial characterization of a ⁄ hamburg ⁄ 5 ⁄ 09 and for providing antisera against h1n1pdm. this study was supported by the european union fp6 global a(h1n1)2009 genetic characterization, molecular evolution dynamics, antiviral susceptibility profiles, and inference of public health implications require nation and region wide systematic analysis of circulating virus. in this study we analysed the genetic and antiviral drug susceptibility profiles of pandemic a(h1n1)2009 influenza virus circulating in portugal. genetic profile analysis was performed in 37 isolates to the hemagglutinin (ha), neuraminidase (na) and mp genes, and in six of these isolates the pb1, pb2, pa, np and ns genes were also analysed. antiviral drug susceptibility profile was analysed for 96 isolates, phenotypically and genotypically to neuraminidase inhibitors (nai) and genotypically to amantadine. the point mutations identified in ha, na, and mp genes of different strains do not seem to evidence an evolutionary trend. this is in agreement with the genetic and antigenic homogeneity that has being described for a(h1n1) 2009 virus. all analysed strains were found to be resistant to amantadine, and five of these strains exhibited a reduced susceptibility profile to nai, three only for oseltamivir and two for both inhibitors. introduction: the dynamics of pandemic influenza a ⁄ h1n1 compared to seasonal strains of influenza is not clearly understood. it is important to understand the patterns of viral shedding and symptoms over time in community-based infections.materials and methods: household infections were followed-up in two large community-based studies. patterns of viral shedding, symptoms and signs, and tympanic temperature were plotted over time and grouped according to strain for analysis.results: the patterns of viral shedding, symptoms and signs, and tympanic temperature in three influenza a strains (pandemic a ⁄ h1n1, seasonal a ⁄ h1n1, and seasonal a ⁄ h3n2) were comparable. peak viral shedding occurred close to the onset of symptoms and resolved after 6-7 days. patterns of viral shedding in influenza b virus infections differed.discussion: the patterns of viral shedding and clinical course of pandemic influenza a ⁄ h1n1 infections were broadly similar to seasonal influenza a ⁄ h1n1 and a ⁄ h3n2. only the clinical course of seasonal influenza b infections was similar to pandemic influenza a ⁄ h1n1. the dynamics of pandemic influenza a ⁄ h1n1 were observed to be largely alike to the dynamics of seasonal influenza a ⁄ h1n1 and a ⁄ h3n2. the coated respirators inactivated a broad range of influenza strains within 1 minute, including the 2009 pandemic strain and human, swine, and avian influenza viruses. antiviral effectiveness was not reduced by hot, humid conditions or repeated saturation, which might occur during prolonged use of respirators. in contrast, infectious virions were detected on the surfaces of all uncoated ffp2 respirators, and could be transferred to glove surfaces during handling of contaminated masks. growth of the viruses was monitored by ha titer using turkey red blood cells, by quantitative real time rt-pcr (qrt-pcr) to detect the influenza a matrix gene, and also by flow cytometry to detect virus positive cells using monoclonal antibodies (imagen influenza virus a and b). 8, 9 matrix gene copy number was determined using qrt-pcr and analysed using the sequence detection software on a 7500 fast system sds (applied biosystems, california, usa). further characterisation was performed through sequence analysis and the ha inhibition (hai) assay. 8 sequence analysis was performed using dnastar 8 and all sequences obtained were compared with the sequence of either the original clinical specimen if available or the conventional atcc derived mdck cell isolate. the hai assay was used to characterize the viruses against a panel of known standard reference viruses and their homologous ferret antiserum.options for the control of influenza vii abstract background: we measured the cross-reactive antibody response to pandemic h1n1 2009 in children and adults before and after vaccination with [2007] [2008] [2008] [2009] influenza season vaccines as part of the rapid public health response to the emergence of ph1n1 and to provide evidence for ph1n1 vaccination policy development in mainland china. materials and methods: archived serum specimens from previous vaccine studies were detected by hemagglutination inhibition assay. results: limited crossreactive antibody response to ph1n1 had been detected among participants of all age groups before and after they had been vaccinated with 2007-2008, 2008-2009 influenza seasonal vaccines. vaccination with seasonal influenza viruses resulted in limited seroconversion to ph1n1 in all age groups, compared with 59-94% of seroconversion to seasonal influenza viruses. but similar to recent studies, a peak of cross-reactive antibody response to ph1n1 was observed in 23% and 39% of participants born from 1915 to 1925 before and after vaccination. conclusions: in order to protect our populations in china, our study strongly suggests vaccination with ph1n1 is required in all age groups and that older populations born before 1925 may be associated with a lower infection rate of ph1n1. on april 15 and april 17, 2009, cases of ph1n1 were identified in specimens obtained from two epidemiologically unlinked patients in the united states and soon thereafter in texas and mexico. 1 since that time, the virus has spread across the globe. assessment of cross-reactive antibody response to the ph1n1 after vaccination with sea-sonal influenza vaccine was first reported from us centers of disease control and prevention (us cdc). according to their results, the seasonal influenza vaccines provided little or no protection against the ph1n1, but some degree of preexisting immunity to the virus existed, especially among adults aged ‡60 years. 2 in this study, using archived serum samples from previous vaccine studies, we measure the level of cross-reactive antibody response to ph1n1 in children and adults vaccinated intramuscularly with trivalent inactivated vaccine developed for the northern hemi serum specimens were collected and provided by provincial centers for disease control and prevention of china as a public health response to the emergence of ph1n1 exempt from human-subjects review. a total of 1308 serum samples were collected from xinjiang uygur autonomous region, yunnan, and shandong provinces. all the serum specimens were grouped by the age of subjects (0-7, 8-17, 18-59, ‡60 years) and by different influenza seasons.hemagglutination inhibition assay was performed according to standard procedures in this study. [3] [4] [5] as with h1n1 components of the vaccine, the seasonal influenza viruses used in this study were a ⁄ solomon islands ⁄ 3 ⁄ 2006 and a ⁄ brisbane ⁄ 59 ⁄ 2007. the ph1n1 influenza virus used in this study was a ⁄ california ⁄ 07 ⁄ 2009 provided by us cdc. all the viruses were propagated in specific pathogenfree embryonated chicken eggs and inactivated by 1& paraformaldehyde. the criteria 6 recommended by the european agency for the evaluation of medical product was applied for the assessment of seasonal influenza vaccine gmt, geometric mean titer; hi, hemagglutination inhibition. three weeks after boosting immunization, spleens were harvested from immunized and control mice. splenocytes were prepared by lymphocyte separation media (ez-sepô, shen zhen, china). the cells were washed and resuspended in complete rpmi-1640 containing fetal bovine serum (hyclone, logan, ut, usa), glutamax, 5 um b-me. splenocytes were cultured in vitro in the presence of inactivated h1n1, h3n2, h5n1, and h9n2 influenza virus antigen for 96 h. quick cell proliferation assay kit (biovision, san francisco, ca, usa) was used to detect the cell proliferation. the 420-480 nm absorbance was read on a plate reader. all experiments have been repeated at least three times.results are presented as mean standard error of the mean (sem). comparison of the data was performed using the student's t-test. significance was defined as a p value of <0ae05. to evaluate the adjuvant effect of recombinant igv, the anti m2e antibody subclasses was measured. igg1 and igg2a were detected after the first and second immunization ( table 1 ). the ratio of igg2a ⁄ igg1 was calculated. immunization with only m2e-hbc showed a lower igg2a ⁄ igg1 ratio <0ae5. igv combined with m2e-hbc led to a high igg2a ⁄ igg1 ratio of up to 5-15 after first and second immunization. these igg subclass distributions indicated that igv can induce a th1 immune response. to determine whether the splenocytes were stimulated in vitro with different subtypes of inactivated influenza antigen after the igv plus m2e-hbc antigen immunization, h1n1, h3n2, h5n1, h9n2 inactivated antigen was used table 1 . the serum igg, igg1, igg2a, and igg2a ⁄ igg1 ratio were measured by elisa after first and second immunization. m2e were coated on the 96 wells plate overnight, and serial dilution sera of day 7, 14, 21 after first and second immunization were added 0, 2ae5, 5, 10, 20, 40 ug ⁄ ml of igg, igg1, igg2a purified antibody were also added for obtaining the standard curve. hrp-labeled goat anti-mouse igg, igg1, or igg2a was then added, washed, and the optical density was read at 492 nm. the results were showed at mean ± sem. day after first immunization days after second immunizationoptions for the control of influenza vii the french grog (groupes régionaux d'observation de la grippe) early warning network collects more than 5000 specimens yearly from cases of acute respiratory illness (ari), using two sampling methods: systematic randomized and non systematic ''ad hoc'' sampling. although vaccines against influenza a virus are the most effective method by which to combat infection, it is clear that their production needs to be accelerated and their efficacy improved. a panel of recombinant live attenuated human influenza a vaccines (laivs), including ns1-73, ns1-126, nsd5, were generated by rationally engineering mutations directly into the genome of a pandemic-h1n1 virus. the vaccine potential of each laiv was determined through analysis of attenuation, immunogenicity, and their ability to protect mice and ferrets. the data indicate that the novel nsd5-laiv was ideally attenuated and elicited strong protective immunity. this study also shows that attenuating mutations can be rapidly engineered into the genomes of emerging ⁄ circulating influenza a viruses in order to produce laivs. the influenza virus exhibits complicated evolutionary dynamics due to multiple reasons, such as diverse hosts, high mutation rates, and rapid replications. in this study, large-scale analyses of 11 454 influenza neuraminidase (na) sequences revealed influenza a and b na genes diverged first around 2641 years ago, and subsequently the na subtypes of influenza a emerged around 1125 years ago. all nine na subtypes of influenza a were genetically distinct from each other, with a total of 23 lineages identified. in addition, five and three sub-lineages were further identified in lineage 1a of na1 and lineage 2b of na2, respectively. the majority of lineages and sub-lineages were found to be host or geographic specific. this study provides not only a better understanding of influenza na evolution, but also a database of lineages and sub-lineages that can be used for early detection of novel genetic changes for improved influenza surveillance. although phylogenetic approaches are commonly used and often found to be powerful, how to accurately identify lineages or sub-lineages of a gene segment of the influenza a virus remains a challenging issue. in this study, we address this issue by analyzing hemagglutinin (ha) sequences using a combination of statistical and phylogenetic methods. following a hierarchical nomenclature system that uses a letter to represent a lineage and a digit for a sub-lineage, we identified 41 distinct lineages and 81 sub-lineages in all 16 ha subtypes through large-scale analyses of 11 821 influenza a hemagglutinin sequences. the majority of the lineages or sub-lineages were host or geographic specific or both. further analysis of other segments will allow us to construct a comprehensive database for influenza a lineages and genotypes, facilitating early detection of new viral strains and genome reassortments and hence improve influenza surveillance. identification of the genetic origin of influenza a viruses will facilitate understanding of the genomic dynamics, evolutionary pathway, and viral fitness of influenza a viruses. the exponential increases of influenza sequences have expanded the coverage of influenza genetic pool, thus potentially reducing the biases for influenza progenitor identification. however, these large amounts of data generate a great challenge in progenitor identification. clinical (nasopharyngeal swabs) and post-mortem materials (fragments of trachea, bronchi, lungs, spleen) were obtained from clinics and ⁄ or out-patients from st. petersburg and from 19 base virological laboratories (bvls) of the research institute of influenza in different regions of the country, which cover approximately 3 ⁄ 4 of the territory of russia. the informed consent for the bio-materials collection and studies was obtained from research subjects or from their relatives in cases of post-mortem materials. isolation of viruses was carried out in the mdck cell culture (cdc, atlanta, ga, usa) and in 10-day-old chicken embryos (e). isolation was done according the standard internationally accepted methods. 3 the reaction of hemagglutination (ha) and the inhibition of hemagglutination (hai) were performed according the who recommended standard method. 3 for the identification of epidemic isolates, we used the hyperimmune diagnostic bovine or ovine antisera annually obtained from the who reference center (cdc). for a detailed antigenic analysis we used the hyperimmune rat antisera against epidemic and reference influenza strains during the period from july 20, 2009 up to april 30, 2010, we have obtained 772 swabs from clinics and out-patients in st. petersburg and 374 swabs from the bvls. in this period, rather high incidence of lethality from pneumonia was observed, which developed on the background of the pandemic flu h1n1v. thus, we received from bvls 173 postmortem materials from 91 deceased patients which manifested pcr+ influenza h1n1v-specific rna. all materials were tested for a possibility of isolation of influenza virus h1n1v both in eggs and in mdck cells. pcr-negative materials were discarded. we isolated 229 strains of pandemic influenza from the materials collected in st. petersburg and region, which comprised 29ae7% of the total number of analyzed samples. at the same time, we did not isolate any other sub-types of influenza in the season 2009-2010 except the pandemic flu. from the swabs purchased from bvls, 47 strains were isolated, which compose 37ae9% of the pcr+ samples, and 35 strains from the post-mortem materials (43ae2% of the pcr+ samples).altogether in the season 2009-2010, we isolated, retrieved, and analyzed in hai 453 influenza strains. 91ae2% of them were pandemic strains a(h1n1)v, and only 7ae9% influenza b viruses. these data together with the epidemiologic data and the results of pcr-diagnostic provide evidence in favor of nearly mono-etiological character of epidemic season 2009-2010 in russia for pandemic influenza a(h1n1)v.though the isolation of pandemic viruses was fulfilled in two traditional model systems, in the case of pandemic 2009 virus, we could observe the tendency of preferential multiplication in embryos compared to mdck, especially in cases of post-mortem material for which chicken embryos are the preferential system of isolation.h1n1v viruses, which were isolated and passaged in mdck, even with significant ha titers, quickly lost their ha activity provided they were kept at +4°c. moreover, some other tested cell lines proved to be practically nonsensitive to the pandemic viruses h1n1v. we used hai reaction for the typing and antigenic characterization of isolated viruses. in the course of isolation of viruses in the reported period, we produced rat polyclonal antisera to the strains a ⁄ california ⁄ 07 ⁄ 09 and a ⁄ st. petersburg ⁄ 56 ⁄ 09 (h1n1)v and the antisera to the strains a ⁄ new jersey ⁄ 8 ⁄ 76 -the virus isolated during the epidemic 1976 in the united states and also of the swine origin -and to the 'swine' strains a ⁄ sw ⁄ 1976 ⁄ 31 and a ⁄ iowa ⁄ 15 ⁄ 30. the hai results of representative strains are given in figure 1 . table 1 shows that the isolated strains were homogenous in their antigenic properties and interacted with the diagnostic antiserum cdc for a(h1n1)v and also with the antisera to the strains a ⁄ california ⁄ 07 ⁄ 09 and a ⁄ st. petersburg ⁄ 56 ⁄ 09 up to 1-1 ⁄ 4 homologous titer. viruses that were isolated from post-mortem materials did not differ by their antigenic characteristics from those isolated from swabs of live patients. only two strains could be attributed to the drift-variants of the strain a ⁄ california ⁄ 07 ⁄ 09 because they reacted with the appropriate antiserum up to 1 ⁄ 8 homologous titer; these strains were a ⁄ pskov ⁄ 1 ⁄ 09 and a ⁄ belgorod ⁄ 6 ⁄ 09. it is interesting that the isolated strains reacted with the antisera to the strains a ⁄ new jersey ⁄ 8 ⁄ 76 and a ⁄ sw ⁄ 1976 ⁄ 31 to 1 ⁄ 4-1 ⁄ 8, and some particular strains even to 1 ⁄ 2 homologous titer. it is even more interesting that some pandemic isolates reacted with the antiserum to the strain iowa isolated in 1930 up to 1 ⁄ 4-1 ⁄ 8 homologous titer. despite of the fact that since the outbreak of 'swine flu' in the usa in new jersey 30 years had gone (and for the strain iowa this period is nearly 70 years) the ha of these viruses and of the pandemic influenza 2009 share some common antigenic determinants as was shown in hai.one more interesting feature of a considerable part of isolated strains is their capability to react with high titers with normal equine serum heated to 56 and to 80°c, while all the strains of swine origin isolated earlier were inhibitor-resistant ( figure 1 ). russian isolates of 2009 divided, in this respect, in two clear and approximately equal in number groups: one of them is similar to the reference strain amino acid substitutions, among them more than 30 were disclosed in antigenic sites, so the degree of similarity to this strain is 78%. a new site of glycosylation was also discovered in the position 276 of ha. 4 essential distinctions of the aminoacid sequence of ha and antigenic properties of the h1n1v strains as compared with actual circulating and vaccine strains is one of the factors that determine the pandemic potential of this new influenza virus.according to the literature, the mutation in the ha gene d222g could cause a broadening of the spectrum of receptor specificity of influenza virus by the acquisition of the capacity to bind both the residues a(2 fi 6) and a(2 fi 3) of the sialic acid of cellular receptors. 5 both types of receptors are present at the human respiratory tract, but in different parts of it, and they exist in different proportions. 6 according to the data of the european center of disease control and prevention (ecdc), the varieties g222 of the h1n1v virus were isolated in 20 countries from subjects deceased of influenza or who suffered a severe form of illness, as well as from those who sustained only a light course of influenza. 7 concerning the strains isolated in rii, this mutation was discovered in nine cases: four were isolated from live patients and five from post-mortem materials. thus, there are no convincing data at present that could prove a causal relationship of the given substitution and the aggravation of a disease course. this is in accordance with previous observations. 7 concerning the resistance of studied strains to the widely used antiviral preparations, it was shown that all tested strains possessed the substitution s31n in the m2 protein that determine the resistance to adamantanes. there was no substitution in the position 275 of neuraminidase (na), which determines the resistance to oseltamivir (h275y). these substitutions are the characteristic indices of the eurasian lineage of swine influenza viruses. 8 thus all studied russian h1n1v isolates were resistant to adamantanes (rimantadine) and sensitive to oseltamivir. respiratory clinical samples taken in 2008 and 2009 that tested positive by real time reverse transcription (rt)-pcr for seasonal influenza viruses (a and b) and pandemic h1n1 2009 respectively were assessed for other respiratory viruses using the resplex ii panel ver2.0 system distributed by qiagen. results showed that co-infections with another 16 respiratory viruses were relatively rare, with a small number of samples having another co-infecting virus present, very few samples having two other viruses detectable in their samples, and none with further viruses. this low number of co-infecting viruses and the ability of certain cell lines not to support infection with particular viruses may make primary isolation of influenza viruses in cell lines easier than might have been thought previously. cm2 is the second membrane protein of influenza c virus and is posttranslationally modified by phosphorylation, palmitoylation, n-glycosylation, and dimer ⁄ tetramer formation. in the present study, we generated rcm2-c65a, a recombinant influenza c virus lacking cm2 palmitoylation site, and examined viral growth and viral protein synthesis in the recombinant-infected cells. the rcm2-c65a virus grew less efficiently than did the wild-type virus. membrane flotation analysis of the infected cells revealed that less np was recovered in the plasma membrane fractions of the rcm2-c65a-infected cells than that in the wild-type virus-infected cells, suggesting that palmitoylation of cm2 is involved in the affinity of the ribonucleoprotein complex to the plasma membrane, leading to the efficient generation of infectious viruses. influenza c virus has seven single-stranded rna segments of negative polarity, encoding pb2, pb1, p3, haemagglutiboth the a2, 6 linkage and its topology on target cells were critical for human adaptation of influenza a viruses. the binding preference of avian flu virus h5n1 ha to the a2, 3-linked sialylated glycans is considered the major factor that limited its efficient infection in human. currently, the switch in binding-specificity of human h5n1 viruses from a2, 3 to a2, 6-glycans did naturally occur, and limited humanto-human transmission was found. to monitor their potential adaptation in the human population, receptor-binding specificity surveillance was made in china. here, the binding specificity of 32 human h5n1 virus strains isolated from 2003 to 2009 was demonstrated. dual binding preference to a2, 3 and a2, 6-glycans were found in a ⁄ guangdong ⁄ 1 ⁄ 06 and a ⁄ guangxi ⁄ 1 ⁄ 08. furthermore, both of them showed a high affinity to the long-branched a2, 6-glycans, which predominate on the upper respiratory epithelial in human. our data suggests that the existence of h5n1 virus with binding specificity to humans should be of concern.introduction via envelope glycoprotein hemagglutinin (ha), influenza viruses bind to cell-surface glycosylated oligosaccharides terminated by sialic acids (sa) where their linkage is celland species-specific. differential receptor binding preference is a host barrier for influenza virus transmission. although most h5n1 viruses have low affinity to neu5aca2, 6gal (human-type) receptor, recent findings suggested that the adaptation of h5n1 virus to human by mutations in the receptor-binding site (rbs) do indeed happen and resulted in enhanced affinity to human-type receptor. [1] [2] [3] in contrast to its putative precursor, a ⁄ gs ⁄ gd ⁄ 1 ⁄ 96, diverse genotypes were presented in currently circulating h5n1 virus, accelerating evolution and widespread occurrence. 4, 5 to date, 10 distinct phylogenetic clades (0-9) were identified based on h5n1 ha, and the confirmed human infections were caused by clade 0, 1, 2ae1, 2ae2, 2ae3, and 7. 6 in china, human h5n1 disease was mostly caused by clade2ae3ae4, which was identified in 28 isolates from 39 confirmed patients from 17 provinces since 2003. clade 7 and clade 2ae2 are responsible for the case in 2003 and 2006, respectively. two current cases of 2009 and 2010 were due to clade 2ae3ae2. now information on receptor property has been documented in some h5n1 viruses of clade 1, 2ae1, and 2ae2. [1] [2] [3] 7 little is known about h5n1 virus of clade 2ae3ae4, particularly from human.recently, a2, 3-specific sialidase-treated red blood cell (rbc) agglutination assay was developed and used for receptor specificity screening of h5n1 virus. 3, 8 the a2, 6 or a2, 3-binding preference can be distinguished by the change of hemagglutination titer reacted with rbcs and enzymatic rbcs. since fine receptor specificity existed in h5n1 viruses, 9,10 the glycan array including sulfated-, fucosylated-, linear sialosides, di-sialosides, or direct binding assay with synthetic polyacrylamide (paa)-based sialylglycopolymers was also recommended for the receptor-specificity surveillance on h5n1 viruses. furthermore, the long-branched a2, 6 sialylated glycans were currently identified to predominate on the upper respiratory epithelial in human and the recognition of this topology, 6¢sln-ln is the key determinant for the human-adaptation of influenza a virus. 11 here, we analyzed the receptor-binding specificity of human h5n1 viruses isolated in china from 2003 to 2009. since 2003, a total of 39 h5n1 infection cases were confirmed in china from 17 provinces. the pharyngeal swabs and lower airway aspirations from the patients were collected within 12 days after disease onset, maintained in viral-transport medium, and tested within 24 hours.options for the control of influenza vii