key: cord-275503-rxjilkff authors: ponkilainen, ville; kuitunen, ilari; hevonkorpi, teemu p.; paloneva, juha; reito, aleksi; launonen, antti p.; mattila, ville m. title: the effect of nationwide lockdown and societal restrictions due to covid‐19 on emergency and urgent surgeries date: 2020-08-07 journal: br j surg doi: 10.1002/bjs.11847 sha: doc_id: 275503 cord_uid: rxjilkff nan due to covid-19 pandemic, surgical societies have recommended to postpone elective and non-urgent surgeries [1] [2] [3] . pandemic potentially prevents patients from seeking medical care also due to acute illnesses. to date, no studies have been conducted on how the pandemic, the resultant social restrictions, and the cancelling of elective operations in hospitals has affected the rate of emergency surgeries. the data for this retrospective study was collected from three finnish the weekly mean incidence of emergency and urgent surgery remained stable after the announcement of the national lockdown (fig. 1a) . however, the weekly mean average was 16 per cent (5 operations per day) less than the average during the previous four years. the incidence of patients aged less than and over 70 years followed a similar trend (fig. 1b) . the number of the most common operation, laparoscopic appendicectomy, decreased by 32 per cent (from 75 to 51, p = 0⋅03) three weeks before the lockdown. however, between three to six weeks after the lockdown, the number rebounded 23 per cent (57 to 70, p = 0⋅25) towards its previous level. the number of hip fracture operations decreased slightly in both the hemi endoprosthesis group (37 per cent, p = 0⋅12, 0-3 weeks after) and the intramedullary nail group (35 per cent, p = 0⋅17, 3-0 weeks before). however, the decrease was followed by a notable rebound (64 per cent, p = 0⋅01) from 37 to 61 operations per three weeks. general mobility in the catchment areas of the participating hospitals decreased notably one week before the declaration of national lockdown. the total incidence of emergency and urgent surgeries was already lower before the lockdown than during the previous four years, and it remained stable during the lockdown. a notable rebound in the rate of appendicectomies and hip fracture operations was seenthree weeks after the lockdown started. the decreasing appendicectomy rate may be the result of citizens avoiding unnecessary healthcare visits, and have therefore delayed their first contact with the ed. conversely, a decreasing trend among hip surgery may be due to senior citizens obeying the general recommendations to stay at home and to avoid falling on slippery roads outside. the general mobility of the population measured by the amount of traffic on the main roads did not result in a decreased rate in emergency or urgent surgery. guidelines for ambulatory surgery centers for the care of surgically necessary/time-sensitive orthopaedic cases during the covid-19 pandemic global guidance for surgical care during the covid-19 pandemic elective surgery cancellations due to the covid-19 pandemic: global predictive modelling to inform surgical recovery plans key: cord-288288-f7yhw3a0 authors: cozzi, giorgio; zanchi, chiara; giangreco, manuela; rabach, ingrid; calligaris, lorenzo; giorgi, rita; conte, mariasole; moressa, valentina; delise, anna; poropat, federico title: the impact of the covid‐19 lockdown in italy on a pediatric emergency setting date: 2020-06-29 journal: acta paediatr doi: 10.1111/apa.15454 sha: doc_id: 288288 cord_uid: f7yhw3a0 italy has been one of countries most affected by the covid‐19 pandemic and the government instituted a strict national lockdown on 9 march 2020 to limit the spread. healthcare services were only open for emergencies or undelayable needs. this study describes the impact of the lockdown on the tertiary level pediatric emergency department (ped) at the institute for maternal and child health irccs burlo garofolo, in trieste, north east italy. it is the only ped in the city and sees 25,000 patients aged 0‐17 annually. were influenced by normal seasonal variations, we also compared the post lockdown data with the same period in 2019. the institute's institutional review board approved the study protocol. the ped database provided the age and gender of the patients, date of the visit, nursing triage category, the patient's main complaint and admission status. triage followed the four-level national triage category system, which ranges from not urgent to emergency and resuscitation ( table 1 ). the primary outcome was urgent visits and admissions before and after the lockdown and the second outcome was the comparison between the post lockdown period in 2020 and the same period in 2019. continuous variables were described as medians and interquartile ranges and the categorical variables as absolute numbers and percentages. the mann-whitney test evaluated the mean difference in the continuous variables before and after lockdown and in 2020 and 2019. fisher's exact test or chi-square test verified the associations between the categorical variables in both 2020 and in 2019. a p-value of <0.05 was considered significant. this article is protected by copyright. all rights reserved the number of ped visits declined considerably after the national lockdown (table 1) . this fell by 76.3%, from 2719 to 646 visits, before and after the lockdown. the decrease from 2019 to 2020 was 77.5%, which confirmed the effect of lockdown. we experienced a decrease in the absolute number of non-urgent and urgent cases. after the lockdown, there was a significant decline in respiratory infections, such as fever, respiratory distress, cough, sore throat and earache and in symptoms related to functional syndromes, such as headache, dizziness and chest pain. injuries also declined. however, the number of agitation symptoms remain stable. the number of patients admitted to the hospital, in particular to the paediatric intensive care unit (picu) also declined. on the other hand, the percentage of urgent triage codes and hospital admissions increased, suggesting that patients in urgent need of medical care came to the ped despite the lockdown. we did not observe a relative increase in emergent codes or in picu admissions, suggesting the lockdown didn't cause a considerable delay in referrals. however, only a limited number of patients needed emergent care during the short study period and the real effects of lockdown on these relatively rare events must be interpreted with caution. after lockdown, a patient affected by intellectual disability arrived dead at the hospital after 10 days of fever and respiratory distress. we can't exclude that this tragic event could have been related to delaying care due to the fear of covid-19 or to the lockdown itself. we know that epidemics can spread easily in emergency settings (1) and limited access to the ped may have helped to limit the spread of covid-19. it is hard to estimate whether it was the fear of contagion or the lockdown that kept people from accessing ped. reports from general emergency departments have showed reduced visits during epidemics, even without lockdown (2) . during the study period, more than 3000 people in our region tested positive for the virus and at least 280 died. in our city, 1250 adults and 60 children and adolescents tested positive, but we only admitted two paediatric patients with covid-19. it is possible that factors apart from the lockdown, such as an independent variation in the spread of infective diseases may have influenced our findings. our data refer to a single ped, limiting the generalizability of our findings. nevertheless, analysing the effects of lockdown on this article is protected by copyright. all rights reserved our clinical setting was important, as it helps us to understand what can be expected when such measures are adopted. this could also help us to plan for future outbreak. the authors have no conflicts of interests to declare. this article is protected by copyright. all rights reserved mers-cov outbreak following a single patient exposure in an emergency room in south korea: an epidemiological outbreak study impact of the 2015 middle east respiratory syndrome outbreak on emergency care utilization and mortality in south korea key: cord-311827-jfdlb2g8 authors: chen, l.-w. antony; chien, lung-chang; li, yi; lin, ge title: nonuniform impacts of covid-19 lockdown on air quality over the united states date: 2020-07-21 journal: sci total environ doi: 10.1016/j.scitotenv.2020.141105 sha: doc_id: 311827 cord_uid: jfdlb2g8 abstract most of the state governments in united states (u.s.) issued lockdown or business restrictions amid the covid-19 pandemic in march 2020, which created a unique opportunity to evaluate the air quality response to reduced economic activities. data acquired from 28 long-term air quality stations across the u.s. revealed widespread but nonuniform reductions of nitrogen dioxide (no2) and carbon monoxide (co) during the first phase of lockdown (march 15–april 25, 2020) relative to a pre-lockdown reference period and historical baselines established in 2017–2019. the reductions, up to 49% for no2 and 37% for co, are statistically significant at two thirds of the sites and tend to increase with local population density. significant reductions of particulate matter (pm2.5 and pm10) only occurred in the northeast and california/nevada metropolises where no2 declined the most, while the changes in ozone (o3) were mixed and relatively minor. these findings are consistent with lower transportation and utility demands that dominate no2 and co emissions, especially in major urban areas, due to the lockdown. this study provides an insight into potential public health benefits with more aggressive air quality management, which should be factored into strategies to reopen the u.s. and global economy. the ongoing pandemic of coronavirus disease has created challenges for governments around the world to balance public safety and economy. in the u.s., following the national emergency declared by president trump on march 13, 2020, states and municipalities have issued various degrees of lockdown and/or stay-at-home policies suiting local specific conditions (lin et al., 2020) . such policies impact air quality through, most notably, declined "non-essential" transportation and energy consumption (le quéré et al., 2020) . among the criteria air pollutants (caps), the u.s. environmental protection agency (epa)'s national emissions inventory attributes 74% of nitrogen oxides (no x , sum of nitrogen dioxide [no 2 ] and nitric oxide [no] ) and 59% of carbon monoxide (co) emissions to on-and off-road traffic and electric generation (u.s. epa, 2016) . ambient levels of the two pollutants might be most affected by the lockdown, compared with primary pm 2.5 and pm 10 (particulate matter with aerodynamic diameters below 2.5 and 10 micrometers, respectively) of which only 10% and 4% result from traffic and electric generation. ozone (o 3 ) is formed in the atmosphere through photochemical reaction of no x and volatile organic compounds (vocs). reduced no x and vocs emissions could either lower or lift o 3 concentrations depending on the local photochemical regime (sillman and he, 2002) . the unprecedented situation of covid-19 pandemic creates an opportunity to assess the contribution of transportation and commercial activities to local air quality and the potential outcome of more stringent emission regulations. such assessments have been carried out for many large cities around the world (kerimary et al., 2020; nakada and urban, 2020; sharma journal pre-proof j o u r n a l p r e -p r o o f 4 et al., 2020; tobías et al., 2020) , but the evidence in the us is lacking. this paper analyzed data from long-term air quality monitoring stations across the u.s. and estimated reductions of caps during the first phase of extensive lockdown. findings can inform future modeling studies that attempt to capture policy outcomes by simulating state-wise emission reductions. such information is also important for the post-pandemic air quality management. the epa national core (ncore) network tracks long-term trends of caps across the u.s. (scheffe et al., 2009) . daily ncore data for january 1, 2020 -april 30, 2020 were acquired from airnowtech (https://www.airnowtech.org/) and cross-verified with those reported to the u.s. epa airdata website (https://www.epa.gov/outdoor-air-quality-data). the six weeks or 42 days between march 15 and april 25, 2020 was designated as the first-phase lockdown period (p1), as many states began restricting businesses and schools in the week of march 15 but relaxed the restrictions somewhat coming into may 2020 (lin et al., 2020; raifman et al., 2020) . a reference period deemed business as usual between january 25 and march 7, 2020 (p0) was also selected, and the relative concentration of a pollutant i, i.e., [i] year should not affect the inter-annual comparison because those effects are mostly canceled in the p1/p0 ratio. p1 and p0 with more than one third of missing data (i.e., >14 days out of the 42-day period) were excluded. for the 28 sites selected, [ ] ′ resulted from at least 2 years of valid data. to estimate the confidence interval of %, a bootstrapping procedure (mooney and duval, 1993 ) based on 12,000 resampling/recalculation of the data were carried out using the matlab® statistics toolbox. the type i error was set to 5%. twenty-eight ncore sites with 2017-2020 no 2 , o 3 , and pm 2.5 data mostly available through the end of april 2020 were identified for this analysis. these sites are in or proximate to 28 different metropolises among 23 states. co and pm 10 were also reported from 21 and 13 of the 28 sites, respectively. there are also significant associations among ∆no 2 %, ∆co%, and ∆pm 2.5 % (see table s2 ). for the top 9 sites in table 1 with the most reductions in no 2 concentration, all the other pollutants except o 3 also declined. as a secondary pollutant, o 3 did not show a clear pattern across the country, with significant increases and decreases observed at 2 and 7 sites, respectively, for the lockdown period. the lockdown appeared to lower no 2 and co more broadly and significantly than pm, consistent with declining mobile and power plant emissions and similar to observations in europe (sicard et al., 2020; tobías et al., 2020) . on a national scale both pm 2.5 and pm 10 are (komenda, 2020) . this is in contrary to states such as north dakota and wyoming where schools were closed but businesses remained open (no stay-at-home order). significant co reductions nonetheless were observed at bismarck, nd and cheyenne, wy during the lockdown (table 1 ). there could also be an urban-rural contrast in how the lockdown affects air quality, as fewer non-essential commercial activities occur in rural and suburban areas than in urban centers. enforcing such policies in rural areas is also more difficult. population density of the zip code where a site is located serves as a surrogate of urbanization and is plotted against ∆no 2 % in population density does explain the different ∆no 2 % between the two oh sites and between the two md sites that bore a uniform lockdown policy within the respective state. in both cases, no 2 reduction increased with the local population density (figure 1) sommer et al. (2020) . the lockdown or stay-at-home orders issued by the u.s. government to counter the covid-19 pandemic has nonuniformly impacted air pollution in the u.s. more consistent no 2 and co declines than other pollutants coincide with reduced transportation and utility demands, while inter-site differences reflect not only the local lockdown policy but also population density. the first phase of lockdown in general affected urban more than suburban air quality. although these effects are temporary, public health benefits from more aggressive air quality management should be considered in the recovery efforts, such as accelerating the transition into cleaner fuels and mass transportation. black and organic carbon emission inventories: review and application to california the impact of covid-19 partial lockdown on the air quality of the city of rio de janeiro assessing air quality changes in large cities during covid-19 lockdowns: the impacts of traffic-free urban conditions in almaty a comparative study of ozone production in five us metropolitan areas covid-19: sisolak bans gatherings of 10 or more people temporary reduction in daily global co 2 emissions during the covid-19 forced confinement statewide stay-at-home directives on the spread of covid-19 in metropolitan and nonmetropolitan counties in the united states sisolak orders statewide closure of nonessential businesses, including casinos, following in footsteps of other states. the nevada independent bootstrapping: a nonparametric approach to statistical inference covid-19 pandemic: impacts on the air quality during the partial lockdown in são paulo state covid-19 us state policy database the national ambient air monitoring strategy: rethinking the role of national networks effect of restricted emissions during covid-19 on air quality in india some theoretical results concerning o 3 -no x -voc chemistry and no x -voc indicators traffic is way down because of lockdown, but air pollution? not so much utah governor asks, salt lake city mayor orders residents to stay home to slow the spread of the coronavirus. the salt lake tribune changes in air quality during the lockdown in barcelona (spain) one month into the sars-cov-2 epidemic national emission inventory (nei) 16) or1 portland -11 (-33, 15) -21 (-19, 74) -6 (-12, 26) wa1 seattle -11 the authors thank staffs from sailbri cooper inc. for collecting and organizing the ncore air quality data, and support from school of public health, university of nevada, las vegas for publishing the paper. key: cord-324708-2ypm0d52 authors: kumar, venkatesan sampath; banjara, roshan; thapa, sushma; majeed, abdul; kapoor, love; janardhanan, ritvik; bakhshi, sameer; kumar, vijay; malhotra, rajesh; khan, shah alam title: bone sarcoma surgery in times of covid‐19 pandemic lockdown‐early experience from a tertiary centre in india date: 2020-07-13 journal: j surg oncol doi: 10.1002/jso.26112 sha: doc_id: 324708 cord_uid: 2ypm0d52 background and objectives: coronavirus disease 2019 (covid‐19) lockdown has presented a unique challenge for sarcoma care. the purpose of this study is to evaluate the early results and feasibility of surgeries for bone sarcomas during the covid‐19 lockdown. methods: our prospectively collected orthopaedic oncological database was reviewed to include two groups of patients‐ those who underwent surgery in the immediate 4 weeks before lockdown (non‐lockdown group) and those operated in the first 4 weeks of lockdown (lockdown group). all patients were followed‐up clinically and telephonically to collect the outcome data. results: out of the 91 patients who qualified for inclusion, fifty were classified into the non‐lockdown group while 41 patients formed the lockdown group. both the groups were comparable with respect to baseline demographic parameters. however, during the lockdown period 37 patients (90%) had undergone a major surgical intervention as against 24 patients (48%) in the non‐lockdown group (p < .001). there was no significant difference in type of anaesthesia, median estimated blood loss and procedure duration. none of the patients/health care workers had evidence of severe acute respiratory syndrome‐coronavirus 2 infection at 15 days follow‐up. conclusion: our study results suggest that appendicular bone tumours can be safely operated with adequate precautions during the lockdown period. the coronavirus disease 2019 (covid-19) pandemic continues to spread across the globe. nationwide lockdown has been the most common response by majority of countries across the world to contain the spread. despite lockdown being the commonest means, there is considerable variability in the response of healthcare systems to the pandemic between different countries. healthcare systems in advanced countries like sweden continued to function unabated during the pandemic, while most others have ceased normal functioning due to nation-wide lockdown. 1, 2 in india too, a nationwide lockdown was imposed from the midnight of 24 march 2020. this was supposed to be one of the strictest lockdowns in the world. 3 like in rest of the world, during the nationwide indian lockdown, it was agreed that elective surgeries had to be deferred and emergency life-saving procedures should be performed with adequate personal protective equipment (ppe) if covid-19 test results are awaited. yet there was considerable ambiguity on semi-elective procedures such as bone sarcoma surgery, as it is well known that delay in surgery can significantly increase the mortality risk in these patients. [4] [5] [6] as a tertiary bone sarcoma referral centre catering to a large population, we continued to perform oncological procedures during the lockdown period with strict adherence to local protocols. the present study is an evaluation of bone sarcoma patients operated during the lockdown period and their comparison to the cohort operated in the month immediately before the lockdown. we suppose that this study would be useful in providing inputs in formulating guidelines for bone sarcoma surgery in times like the current covid-19 pandemic. the study was conducted at new delhi where lockdown was in effect from 6 am on 23rd march 2020. we identified two groups of patients, namely the "non-lockdown group" which was defined as those patients who were operated between 24th february and 22nd march 2020; and the "lockdown group" included those patients operated between 23rd march and 18th april 2020. data regarding patient demographics, tumour characteristics and intraoperative details were collected from a prospectively maintained, dedicated musculoskeletal oncological database. as the lockdown was sudden and unprecedented, some changes were made in patient selection for surgical procedures. ever since lockdown, all patients and their caretakers were thoroughly screened for covid-19 symptoms through a safety checklist ( figure 1 ). those patients/caretakers, who failed the checklist, were immediately referred to designated covid-19 testing area for further evaluation. we did not perform routine covid-19 testing on all our preoperative patients as per existing guidelines during that period. all healthcare workers were following the ppe protocols as per hospital guidelines. for those patients who had to undergo a chest computed tomography scan for preoperative staging work up (as per standard sarcoma guidelines), radiologists were asked to specifically look for evidence of covid-19 infection to rule out asymptomatic carriers. during surgery, pulse lavage irrigation of the wound was avoided to reduce droplet contamination. 7 however, we continued to use electrocautery albeit minimally so as to reduce intraoperative blood loss. in those procedures where intramedullary reaming was required, power reamers were used as necessary. tele consultation services were provided for sarcoma patients apart from routine hospital visits for wound check. all patients included in the study had a minimum follow-up of 15 days following the procedure. during follow-up, apart from routine examination, each patient was specifically asked for symptoms/ evidence of covid-19. statistical analysis was performed using r statistical software version 3.6.1. categorical data was analysed using the χ 2 test. fisher's exact test was employed if the expected frequency in any cell was less than 5. test for normality was employed for continuous data. those following normal distribution were analysed using the student t test. non parametric data were analysed using wilcoxon test. statistical significance was attributed when p value is less than or equal to 0.05. a total of 91 patients had been treated during the entire study period. of these, 50 patients had been classified in the 'non-lockdown' group; while the remaining 41 as 'lockdown' group. table 1 compares the baseline characteristics between the groups. during the lockdown period, malignant tumours were operated in preference to benign tumours and biopsy procedures (p < .001). the only benign tumour patient that had been operated during lockdown had a displaced pathological fracture of femur secondary to solid abc for which she had undergone bone grafting and plating. out of the five giant cell tumor (gct) patients who had surgery during this period, two had pathological fracture (in proximal femur), two had impending fungation and one nursing mother had distal radial gct with severe pain restricting her baby care. hence, these procedures qualified for urgent intervention. the most common location was distal femur followed by proximal tibia in both groups ( figure 2 ). there was no statistically significant difference in tumour location between the two groups. however, due to logistic reasons, we could not operate upon pelvic and sacral tumours during the lockdown period. there was a significant increase in number of major surgeries performed during lockdown (37 out of 41) as compared with normal period (24 out of 50). this is because of availability of extra theatre time during the lockdown period as other elective orthopaedic procedures like arthroplasty, arthroscopy, etc. were deferred. we had a significantly higher proportion of patients who had pre-operative chemotherapy in the lockdown group (49% vs 18%). also, the american society of anesthesiologists grades, intraoperative blood loss and duration of procedures were not different compared with nonlockdown group ( table 2) . none of the patients had symptoms/ evidence of covid-19 at a minimum of 15 days follow-up. out of the major procedures (n = 37) performed during the lockdown period, 31 (84%) were limb salvage procedures ( table 2 ). our choice of reconstruction did not defer during the lockdown period as compared with normal group. however, we had faced certain challenges which will be discussed under a separate he recovered uneventfully. we had two complications in each group-two vein injuries which were repaired, one patient had common peroneal nerve coursing through the tumour that had to be sacrificed and one had an iatrogenic cautery burn. completion of neo-adjuvant chemotherapy increases recurrence risk by 1.14 times. also, at a mean follow-up of 2.8 years, just 20% patients with recurrence were surviving as against 60% of survivors in the group without local recurrence. hence, we continued to serve our patients with due precautions. even the nhs england report says "cancer services will need to continue …." and elective surgeries with expectation of cure, prioritised to priority level 2 so as to save life and to prevent progression of disease to beyond operability. 11, 12 out of the 41 patients who had been treated during the lockdown, three patients were referred for covid-19 screening. in all these three patients, these symptoms were picked up in the waiting area before they entered the ward or operation theatre with the help of the checklist in figure 1 . out of these three patients, two patients did not require testing as per standard indian council of medical research guidelines. hence, they were taken up for surgery. in the third patient, covid-19 testing was done and found to be negative. also, as mentioned earlier, every non-contrast computerized tomography chest scan was discussed with a radiologist to rule out radiological evidence of covid-19. although there is no fool proof mechanism to identify asymptomatic carriers, a proper checklist and strict adherence to local infectious disease protocol can help in reducing inadvertent disease transmission. this is especially important in a country like india with 1.3 billion population where national lockdown was declared well before the disease entered the community transmission stage. while it is well known that lockdown can plateau the epidemic curve, it should be understood that the disease may not be eradicated in the near future in our setting. hence, deferring surgery to a later date for sarcomas will only risk patient's life and may put undue pressure on the system when national lockdown is relaxed. we performed a significantly higher number of major surgeries (90%) during the lockdown. while we had reported increased availability of theatre slots under the "results" section, another important factor is that patients could not reach our tertiary centre due to transport lockdown (figure 3) . we had performed only four biopsy procedures during the 4 week lockdown period as against 26 biopsy procedures during the preceding 4 weeks. our hospital caters to covid-19 pandemic in sweden. wikipedia, 2020 comparing six health-care systems in a pandemic. council on foreign relations covid-19 pandemic: a litmus test of trust in the health system french sarcoma group proposals for management of sarcoma patients during the covid-19 outbreak surgical management of bone and soft tissue sarcomas and skeletal metastases during the covid-19 pandemic management of cancer surgery cases during the covid-19 pandemic: considerations covid-19: potential transmission through aerosols in surgical procedures and blood products preparing for a covid-19 pandemic: a review of operating room outbreak response measures in a large tertiary hospital in singapore survey of covid-19 disease among orthopaedic surgeons in wuhan, people's republic of china factors associated with local recurrence in operated osteosarcomas: a retrospective evaluation of 95 cases from a tertiary care center in a resource challenged environment cancer guidelines during the covid-19 pandemic document library bone sarcoma surgery in times of covid-19 pandemic lockdown the authors declare that there are no conflict of interests. the data that support the findings of this study are available from the corresponding author upon reasonable request. http://orcid.org/0000-0001-7457-4080 key: cord-283416-dhtintid authors: cheli, marta; dinoto, alessandro; olivo, sasha; tomaselli, marinella; stokelj, david; cominotto, franco; brigo, francesco; manganotti, paolo title: sars-cov-2 pandemic and epilepsy: the impact on emergency department attendances for seizures date: 2020-08-22 journal: seizure doi: 10.1016/j.seizure.2020.08.008 sha: doc_id: 283416 cord_uid: dhtintid abstract introduction the risk of acquiring sars-cov-2 in a hospital setting and the need of reorganizing the emergency departments (eds) to cope with infected patients have led to a reduction of ed attendances for non-infectious acute conditions and to a different management of chronic disorders. methods we performed a retrospective study evaluating the frequency and features of ed attendances for seizures during the lockdown period (march 10th april 30th 2020) in the university hospital of trieste, italy. we studied the possible pandemic impact on the way patients with seizures sought for medical assistance by comparing the lockdown period to a matched period in 2019 and to a period of identical length preceding the lockdown (january 18th – march 9th 2020). results a striking decrease in total ed attendances was observed during lockdown (4664) compared to the matched control (10424) and to the pre-lockdown (9522) periods. a similar reduction, although to a lesser extent, was detected for seizure attendances to the ed: there were 37 during lockdown and 63 and 44 respectively during the two other periods. intriguingly, during the lockdown a higher number of patients attended the ed with first seizures (p=0.013), and more eegs (p=0.008) and ct brain scans (p=0.018) were performed; there was a trend towards more frequent transport to the ed by ambulance (p=0.061) in the lockdown period. conclusions our data suggest that the pandemic has affected the way patients with seizures access the health care system. the first reported cases of unexplained severe pneumonia in wuhan date back to december 31 st , 2019 [1] . on february 20th, 2020, in codogno hospital, the first italian case of sars-cov-2 was diagnosed. the virus has rapidly spread throughout the whole peninsula and on march 9th, the italian government imposed a national lockdown in order to reduce the viral transmission and to avoid overloading the national health system. the urgent need of reorganizing emergency departments (eds) to cope with the rising number of infected patients had, and it is still having, a significant impact on the management of both acute and chronic conditions. in fact, the risk of acquiring sars-cov-2 in hospital-based setting has deeply shaped the access of patients to the health care resources [2] [3] . the net reduction of patients admitted for acute coronary syndromes in northern italy is a striking example of how the fear of the pandemic could overtake the need of medical assistance. despite the growing evidence of neurological involvement during the sars-cov-2 infection [4] , very little is known on how the pandemic modified the access to health care in patients with neurological conditions. recently, a taskforce of experts has published an evidence-based consensus [5] on the management of epileptic patients during the pandemic, highlighting that those patients should receive as much care as possible at home since sars-cov-2 infection could pose a life-threatening risk. the university hospital of trieste, in northern-east of italy, is part of the italian national health system and directly serves a population of 234.493 people. the hospital is the hub centre of the "giuliano-isontina" area (total population 373.839, it is composed of one hub and two spoke hospitals, one of which has a neurology ward) in the friuli-venezia giulia region (fvg). the population has a median age of 48 years and 47,9% of them are males. the ed guarantees a 24-hours free-of-charge admission. our neurological ward has 32 beds, 8 of them dedicated to sub-intensive care monitoring. patients with seizures are firstly evaluated in the ed by a general neurologist who may prescribe, if necessary, eeg and brain computed tomography (ct) scans in the emergency setting. patients are then referred to the epilepsy unit, where more specific examinations may be requested. during the pandemic, nasal swabs were performed in the ed and positive patients were admitted to dedicated wards, intensive and sub-intensive care units. in march 2020, 1593 positive patients in fvg (545 in trieste) and 113 deaths (60 in trieste) were recorded. by the end of april, the total number of positive patients in fvg increased to 3008 (1252 in trieste) and the deaths increased to 289 (153 in trieste) [6] . we performed a retrospective study evaluating the frequency and features of ed attendances for seizures during the lockdown period, in order to verify if the pandemic could have had an impact on how patients with seizures sought medical assistance. the aim of our analysis is to describe the frequency and characteristics of ed attendances for seizures during the nationwide lockdown in the university hospital of trieste, and to compare it to a similar period immediately preceding the lockdown ("pre-lockdown") as well as to the same period in 2019 ("matched control"). we defined the "lockdown period" j o u r n a l p r e -p r o o f from the 10 th of march to the 30 th of april (52 days). as a "matched control period" we picked the same days in 2019. we also analysed the "pre-lockdown period", with the same duration, from the 18 th of january to the 9 th of march. data were retrospectively collected by reviewing medical charts of each patient admitted for seizure in the ed. age, sex, triage code, attendance by ambulance, required hospitalization, first seizure, seizure semiology and diagnostic tests performed (eeg or ct) of each patient were collected. seizures were classified, following ilae 2017 recommendations, in focal onset, with or without awareness impairment, generalized onset and status epilepticus [7] . we also reported the total number of ed attendances for all causes (data provided through ed registry by one of the authors, f.c.) . continuous variables are presented as median (range) and non-continuous variables as number (percentage). a statistical analysis was performed in order to compare the distribution of variables between the "lockdown", the "pre-lockdown" and "matched control" periods: chi-squared and kruskal-wallis tests were used as appropriate. p-values <0.05 were considered as statistically significant. the study was conducted according to the declaration of helsinki. the study was approved by the local ethics committee. a total number of 144 patients with seizures attended the ed during the whole period considered (lockdown, prelockdown and matched control). data regarding the three periods are summarized in table 1 and figure 1. the number of total ed attendances decreased from 10424 in the "matched control" and 9522 in the "pre-lockdown" to 4664 in the "lockdown" period. in parallel, the number of ed attendances for seizures decreased with the spread of sars-cov-2: during the "lockdown period" only 37 patients went to the ed for seizures, compared to 69 and 44 patients who sought for medical assistance for seizures during the "matched control period" and "pre-lockdown period". when comparing the "lockdown period" with "matched control period", we found a significant increase in the number of first seizures (p=0.013), eeg (p=0.008) and ct scans (p=0.018) performed in the ed. a trend favouring attendances by ambulance (p=0.061) was also detected during the "lockdown period" compared to the matched control period. surprisingly, during the "lockdown period" we found an increase in the admissions to the ward for seizures (p=0.043) and a more severe triage code (p=0.027) if compared to the "pre-lockdown period". furthermore, comparison of the two control periods showed no differences, except for the number of eeg performed, which was higher in 2020 (p=0.005). the total number of ed attendances dramatically decreased during the lockdown, raising concerns about the reduced seek of medical assistance for sars-cov-2 unrelated, yet life-threatening, conditions; indeed, a drastic reduction of hospitalization for acute coronary syndromes has been reported in several italian hospitals during the lockdown (547 admissions compared to 889 in the previous months and 756 in the previous year) [3] . seizures play a major role in the ed since it has been estimated that about 1% of all evaluations are related to them [8] . our study shows that sars-cov-2 pandemic had an impact on the ed attendances for seizure, although their reduction was not as dramatic as reduction of total ed attendance. in fact, the number of seizure evaluations was nearly halved during the "lockdown" when compared to the "matched control period" in 2019 (37 vs 69). a slight reduction of attendances was also noted between the "pre-lockdown" and the "lockdown" periods" (44 vs 37), and between the "matched control" and "pre-lockdown" periods (69 vs 44). those findings suggest that the pandemic, even before the lockdown, has reduced the seek for medical assistance, even in patients with this acute, life-threatening condition. regarding neurology ward admissions, we noted a higher number of admitted patients during the lockdown when compared to the preceding months (36,1% vs 15,8%). the increase could be related to a more severe clinical picture of ed attending patients, as highlighted by the difference in triage codes between the two periods, or by the need of further diagnostic tests that could not be easily performed in an outpatient setting during the lockdown (e.g. mri). intriguingly, we also found that patients who attended the ed during the "lockdown" for seizures were mostly experiencing their first epileptic event. it has been estimated that 46% of patients presenting with seizures in the ed are known epileptics [9] . in our population, 77.8% and 65.9% of patients were known epileptics in the "matched control" and "pre-lockdown" periods, respectively. the percentage of chronic patients presenting to the ed decreased to the 54.1% during lockdown. we hypothesize that in patients with known epilepsy and in their relatives, who have already experienced and managed seizures, the fear of contagion tended to scale back the perceived benefit of seeking medical aid. finally, despite the limitations imposed by the reorganization of eds, the number of eegs and ct scans performed increased during the lockdown, probably due to the higher number of attending first seizures, which required further testing for diagnostic and therapeutic purposes [10]. in our experience, dedicated pathways for sars-cov-2 patients were developed rapidly, in order to limit the exposure of health care workers and provide to infected patients the needed diagnostic procedures [11] . all patients who attended to the ed and required diagnostic assessment or ward admission, underwent to sars-cov-2 nasal swab. this procedure allowed us to complete the diagnostic assessment in most of the patients presenting with first seizures and to safely admit to the ward those patients who required further medical care. attendances to the emergency department for seizures in the "matched control", "pre-lockdown" and "lockdown" periods a familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster no lockdown for neurological diseases during covid19 pandemic infection reduced rate of hospital admissions for acs during covid-19 outbreak in northern italy the lives of neurologists in the era of covid-19: an experience from the trenches keeping people with epilepsy safe during the covid-19 pandemic data acquired from the italian ministry of health ilae classification of the epilepsies: position paper of the ilae commission for classification and terminology emergency department seizure epidemiology evidence-based guideline: management of an unprovoked first seizure the authors would like to thank marta tison for english proof-reading.j o u r n a l p r e -p r o o f key: cord-287676-qh7zeyyx authors: angoulvant, françois; ouldali, naïm; yang, david dawei; filser, mathilde; gajdos, vincent; rybak, alexis; guedj, romain; soussan-banini, valérie; basmaci, romain; lefevre-utile, alain; brun-ney, dominique; beaujouan, laure; skurnik, david title: covid-19 pandemic: impact caused by school closure and national lockdown on pediatric visits and admissions for viral and non-viral infections, a time series analysis date: 2020-06-03 journal: clin infect dis doi: 10.1093/cid/ciaa710 sha: doc_id: 287676 cord_uid: qh7zeyyx a time series analysis of 871,543 pediatric emergency visits revealed that the covid-19 lockdown and school closure were associated with a significant decrease in infectious diseases disseminated through airborne or fecal-oral transmissions: common cold, gastro-enteritis, bronchiolitis, acute otitis. no change was found for urinary tract infections. m a n u s c r i p t in late december 2019, patients with viral pneumonia due to an unidentified microbial agent were reported in wuhan, hubei province, central china. this disease outbreak, covid-19, then grew substantially and was declared a pandemic by who on march 11, 2020 . [1] in 1995, a major french nationwide strike paralyzed france. for 19 days from november 30, 1995 to december 18, people stayed at home, including children who normally went to day-care centers and a significant decrease of bronchiolitis cases was observed. this decline might have been caused by workplace and school absenteeism, and lower attendance of day-care centers. [2] likewise, during the 2013-2014 measles epidemic, a reduction in contact rate during school vacations was associated with 4900 averted cases in the netherland. [3] after reaching france on january 24, 2020, a major progression of covid-19 from february to march lead to public health interventions. partial lockdown and school closure were initiated early march and a national lockdown was officially started on march 17 th , 2020. [4, 5] no previous public health intervention can be compared to the extent of the lockdown established for the covid-19 epidemic. we hypothesized that this unusual situation in france would be associated with a sharp decrease in pediatric infectious diseases that usually disseminate through social contacts, with schools at their center. being able to prevent these infections, responsible for many pediatric hospitalizations, would be an unwanted direct benefit of the lockdown for children, that seem otherwise significantly more protected than the adult from the sars-cov-2 infection. [6] even more, this could open the road for future guidelines to control future major health issues once the covid-19 pandemic is under control. therefore, the evolution of several major diseases usually correlated with dissemination through contact, such as gastroenteritis, common cold, and acute otitis media, were investigated before and after the start of the national lockdown. urinary tract infections, which are not reported to be correlated with contacts in children, were used as control outcome. we conducted a quasi-experimental interrupted time series analysis based on multicenter prospective french surveillance data for pediatric emergency department (ped) visits and related hospital admissions. the regional centre of observation and action on emergencies e-cerveau, agence régionale de santé, is an official network of emergency departments dedicated to public health that automatically transmit a summary of anonymized data from all their visits to the regional database. the database has been approved by the french data protection authority. these data include discharge diagnosis coded by the physicians in a c c e p t e d m a n u s c r i p t charge of the patient at the end of the peds visits according to icd-10 th revision and hospital admission or discharge. this study covers 6 peds from academic hospitals being part of assistance publique -hôpitaux de paris, located in and around paris, gathering 250,000 annual visits that daily transmitted their from january 1 st , 2017 to april 19 th , 2020. we used the e-cerveau database for this research. data are anonymous. patient informed consent is not required according to current dispositions. groups of diagnosis extracted were: gastroenteritis, common cold, bronchiolitis, acute otitis media, considered as infectious diseases thriving through social contact, and urinary tract infection (table s1). visits were grouped by calendar weeks for each year. the main outcome was the evolution of the number of hospital admissions following the french decision to close schools and start a lockdown for the whole country. [5] the secondary outcomes were the number of peds visits for gastroenteritis, bronchiolitis, common cold, acute otitis media. as recommended to prevent potential confusion, [7, 8] urinary tract infections were analyzed as a control outcome, given that this common pediatric infectious disease is not expected to be impacted by social distancing, although indirect effect such as stress or diet change cannot be excluded. this outcome was already used as a control concerning previous acute respiratory tract infections studies. [9] statistical analysis outcomes were analyzed by quasi-poisson regression, accounting for seasonality, secular trend before and after lockdown, and overdispersion of data. [7, 8, 10, 11] seasonality was taken into account by including harmonic terms (sines and cosines) with 12-months and 6months periods to adjust for the seasonal pattern. [11] the time unit chosen was one week to provide optimal precision to the model. [8] we hypothesized that the intervention would have an immediate impact, meaning after one analysis. analysis of acute gastroenteritis was performed on data from four hospitals combining 81% of the visits during the study period. all statistical tests were two-sided, and we considered a result as "significant" when the p-value was <0.05. all statistical analysis involved using r v3.6.1 (http://www.r-project.org). a total of 871,543 peds visits in the six participating centers from january 1 st , 2017 to april (table s2 , figure 1 and figure s1 ). we found a significant decrease of acute gastro-enteritis, common cold and acute otitis media (table s2, table s3 ) with a sharp decrease over 70% compared to the expected values (figures 1b-d) . decrease of bronchiolitis was also significant (-63.5% 95ci [-101.8 ; -25.9]) (table s2) . consistent both with our hypothesis and the incubation time of these different diseases, a dramatic decrease of overall peds visits (-68.5%) and hospitalization (-44.7%) was observed as soon as one week after the start of the lockdown ( figure s1 ). by contrast, urinary tract in this time series analysis of 871,543 peds visits, the number of peds visits and admission after the lockdown decreased by -68% and -45%, respectively. we found a significant decrease over 70% of acute gastro-enteritis, common cold, bronchiolitis and acute otitis media compared to the expected values. unprecedented public health interventions were ordered to reduce the risk of sars-cov-2 transmission. [4] our data suggest that these measures have also a critical impact on the transmission of numerous infectious diseases, more specifically on viral or viral-induced pediatric diseases. this major achievement may also play a critical role in making more health resources with adults admitted in icu and health care workers available to fight covid-19 pandemic. [12] in paris area, children with acute illness could be seen not only in ped but also in private office (general practitioners or pediatrician), general practitioner house calls, and communities' center. however most of these sites of care do not work 24/7 and rarely perform additional tests especially for the younger children. our study have limitations; we cannot exclude a change in clinical management such as avoidance of ent examination because of covid-19 fear which could have influenced diagnosis coding; we did not collect data regarding severity and so we cannot exclude that reduction in presentations was associated with children presenting later in their illness. while the dramatic decrease in peds could be partially due to transportation limitations and a fear of going to the hospital and increase of telemedicine, the stability in the number of urinary tract infections cases that we used as control outcome and the significant decrease in hospital admissions do not favor this hypothesis. moreover visits to private doctors' offices decrease by 40%. [13] while the role of the children in the dissemination of sars-cov-2 is still being discussed, finding that school closure and national lockdown were linked to a dramatic decrease in pediatric emergency visits concerning gastroenteritis, acute otitis media, bronchiolitis, and other viral diseases could be not only an unexpected benefit for the children but could also raise the question of the impact on the health care system of starting lifting the french national lockdown by reopening the schools. [14] complementary studies using more granular data such as severity could be useful to better understand lockdown's impact on children's health. this nation-level quasi-experiment is unprecedented in the modern era. it provides unique evidence which could be key in the post covid-19 era, to implement new guidelines and new routines in our way of life, and in order to fight past but also potential future infectious diseases threats reaching both children and adults. m a n u s c r i p t impact of lockdown on weekly pediatric emergency department visits and major pediatric infectious diseases, from january 1 st , 2017 to april 19 th , 2020. -a: overall peds visit (n=871,543). the black line shows the observed data. the bold red slope shows the model estimates based on observed data (quasi-poisson regression modeling). the bold blue slope shows the expected values without lockdown in the post-intervention period (quasi-poisson regression modeling). the start of the lockdown is indicated by the vertical black arrow. a c c e p t e d m a n u s c r i p t figure 1 world health organization. who director-general's opening remarks at the media briefing on covid-19 -11 coincidence of public transport strike with bronchiolitis epidemic the reduction of measles transmission during school vacations adaptation of the national plan for the prevention and fight against pandemic influenza to the 2020 covid-19 epidemic in france. disaster medicine and public health preparedness décret n° 2020-260 du 16 mars 2020 portant réglementation des déplacements dans le cadre de la lutte contre la propagation du virus covid-19 mmwr morbidity and mortality weekly report interrupted time series analysis in drug utilization research is increasing: systematic review and recommendations segmented regression analysis of interrupted time series studies in medication use research impact of implementing national guidelines on antibiotic prescriptions for acute respiratory tract infections in pediatric emergency departments: an interrupted time series analysis. clinical infectious diseases : an official publication of the infectious diseases society of america regression based quasiexperimental approach when randomisation is not an option: interrupted time series analysis interrupted time series regression for the evaluation of public health interventions: a tutorial paediatric intensive care society. pics and ics joint position statement 14. french government. french government measures m a n u s c r i p t a c c e p t e d m a n u s c r i p t key: cord-277667-vclij9ax authors: glancy, d.; reilly, l.; cobbe, c.; glynn, m.; punchoo, s.; foy, k. title: lockdown in a specialised rehabilitation unit: the best of times date: 2020-05-21 journal: irish journal of psychological medicine doi: 10.1017/ipm.2020.50 sha: doc_id: 277667 cord_uid: vclij9ax specialised rehabilitation units offer inpatient multi-disciplinary rehabilitation for individuals with severe and enduring mental illness. a cornerstone of therapy is the work in the community through further education and community organisations. however, coronavirus restrictions have meant that such external supports are no longer available for the duration of the crisis. this has led to opportunities for developing new ways of offering rehabilitation within hospital environments. this article describes some of the new initiatives developed. the benefits of the lockdown for service users are also discussed. many found the cessation of visits from family members with whom they had an ambivalent relationship helpful. the lockdown improved relationships between patients on the unit and encouraged a greater feeling of community. the lockdown has also emphasised the importance of team self-awareness and an awareness of the nature of the treatments offered. the national mental health division established placements at specialised rehabilitation units for individuals with severe chronic and enduring mental health illnesses at bloomfield hospital, dublin (hse mental health services, 2018) . the patients referred to the unit typically have a history of complex treatment refractory psychiatric symptoms and multiple prolonged admissions to acute mental health units. they have reduced ability to manage in the community despite intensive management from their local community mental health teams. the service was only established in 2018 and since then has accepted referrals from all parts of the country. whilst the majority of the service users have a diagnosis of schizophrenia, all have additional mental health needs and most have a history of complex trauma. our multidisciplinary team consists of specialists from psychiatry, psychology, occupational therapy, nursing as well as a peer support worker. the rehabilitation offered is individualised, goal orientated and led by service user-generated goals. prior to the national coronavirus-19 emergency, a fundamental pillar of the rehabilitation offered was an emphasis on activities in the community and outside the unit. as a result, our clients attended a variety of local services including adult education, men's sheds group, tidy towns groups, voluntary work in charity shops and fitness activities in local sport and leisure facilities. the lockdown restrictions meant that such local supports and activities were no longer available. the multidisiplinary team (mdt) in conjunction with service users therefore had to develop additional activities to support the rehabilitative programme. previously, most patients on the unit had weekly visits from family members. these visits were generally perceived to be helpful, and many of the patients had close relationships with their relatives. as family contact was curtailed due to the lockdown, many service users felt better able to reflect and empowered to speak with therapists about the nature of those relationships. familial constellation and the role of the patient within that system became much more apparent. patients opened up more during psychotherapeutic sessions about significant trauma or attachment issues. trauma and attachment issues within families can contribute to high levels of unhealthy enmeshment creating chaotic boundaries, difficulties with emotional regulation and poor sense of self for the individuals. one of the service users summarised the new world of the lockdown as 'no visits, no callspeace and space'. whilst families may be supportive, the family can also be the source of trauma triggers for those with complex trauma history. as stated by aldersey & whitley (2015) , 'families both facilitate and impede recovery process'. the lockdown created an opportunity to explore these challenges in a safe and non-threatening environment during 1:1 sessions and group sessions. the reduced visits from relatives meant that patients felt better able to examine patterns of communication, relationships, power structures and other aspects of family systems. many of the service users are highly self-critical, perfectionistic and have experienced considerable rejection and perceived failure throughout their time in psychiatric services. one of our patient summarised this as 'my doctor said that there was nothing he could do for me : : : i'm not like the others'. in essence, they have been through a revolving door with multiple relapses and readmissions. during the covid-19 pandemic, they described their setting as 'safe' and that 'everyone is in the same boat'. they reported feeling less expectations being placed on them, both by themselves and others. the team noted that it was easier to collaboratively work on development of grounding techniques, addressing internal critic, practising mindfulness, exploring past trauma and integrating self, improving self-care and improving awareness of emotions. prior to the restrictions, self-reflection could be easily avoided through engagement in a myriad of distraction and avoidance techniques. getting in touch with internal processes to integrate mind and body awareness has been an enriching though difficult journey for patients. for most, self-awareness improved, and there was increased hope for the future by addressing the past with one of our clients saying 'i am fearful but hopeful that i will see the light at the end of the tunnel'. during the lockdown, some reflected on their rehabilitation journey and one stated 'i can't believe i was allowed out on my own, and now i appreciate it even more, i will make full use of it in the future'. according to herman (1997) , 'safety, remembering, mourning and reconnection are essential trauma resolution preambles' which summarise the self-reflection for many during the lockdown. with the onset of covid-19, previous routines have been thrown into chaos. the service has now adapted the therapeutic programme to enable this period to be a learning experience of self-discovery and examine our values, beliefs and raison d'etre. the focus has shifted to increasing group-based activities on the ward and looking at innovative ways to occupy this time based on service user needs. the team has committed to provision of traumafocused therapy and developed educational processes with staff psychology book clubs and lunchtime education groups. a tai chi group is delivered daily on the unit focusing on grounding, breathing and self-soothing (kong et al. 2019 ). the group is attended by everyone involved in the unit, staff and service users alike. a 'time to talk' group was established within the service to address social and relationship skills for service users with co-morbid learning disability or developmental disorders. the group was designed as an open forum to explore gender identity, expression and sexuality in the context of their own lives and society as a whole. a separate psychotherapy group was established for individuals with higher level functioning using the yalom model (yalom & lesczc, 2005) . the ethos of the wellness recovery action plan (wrap) (copeland, 2011) and the decider skills group (ayres & vivyan, 2019) fosters personal responsibility, developing coping skills and use of grounding techniques. the delivery of other existing group programmes has also undergone changes with the music therapist now delivering their sessions over the zoom application. an on-site greenhouse has afforded the opportunity for one service user to lead a gardening segment to fellow interested service users called 'how's it growing'. there are increasing tangible examples of shared camaraderie evident on the unit including service users sharing their cooked food and blossoming friendships among others. while some of these interactions have occurred spontaneously, the unit itself has focused on creating a therapeutic community with organisation of activities such as afternoon tea and outdoor hikes within the 2 km distance. like the rest of the country, time was spent to enhance the physical environment by improving the garden and outdoor spaces with planting. the focus of this was to generate a common purpose and ownership of the shared space through engagement in meaningful pursuits. however, within any confined space, it is anticipated that conflict will occur, and healthy outbursts have been welcomed as individuals are encouraged to share how they openly feel about situations and resolve it accordingly. at the weekly peer support group, residents were offered the opportunity to express their thoughts and feelings about the restrictions in place as a result of the coronavirus outbreak. despite the increased focus on group and 1:1 sessions, maintaining social contact with the outside world has remained pivotal. service users were offered the same rights as everyone else to access the community in line with the national lockdown regulations. service users can access essential services within the community such as post office, bank and essential shops. a preventative approach to covid-19 was adopted with emphasis placed on hand hygiene and social distancing measures and adapted education sessions for this were developed and delivered. wilcock (2002) discusses the theory of 'doing, being and becoming' as central components for achieving wellness and realising self-actualisation. as a service in its infancy, this period of lockdown has encapsulated a greater balance between the three aspects, moving from primarily doing to being and becoming. as the psychiatrist victor frankl (1984) once said, 'when we are no longer able to change a situation, we are challenged to change ourselves'. the team believes that the end of the lockdown shall present its own challenges. the possibility of reinstatement of contact with family members, returns to activities in the community and increased external distractions have the potential to be testing. as a service, we plan to devote a number of group therapy and individual sessions to reflect on the past number of months and the return to a state of normality. using questionnaires, we hope to assess the service users and staff attitudes to the changes made to the programme during the lockdown. the results of these questionnaires along with wider patient and staff discussions will inform how we integrate the lockdown programme into a post-lockdown world. the lockdown allowed the team the space and opportunity to self-reflect on the essence of what defines our work. it allowed both service users and staff the opportunity to reflect on the shared journey that we are taking together and the necessity for collaboration, honesty and open dialogue. whilst families can often be seen as an important resource for service users, we became more aware of the double-edged nature of family relationshipsparticularly in individuals who have traumatic, ambivalent or challenging relationships with their relatives. all too often, in healthcare, we tend to prioritise action at the expense of reflection. the covid-19 emergency allowed this service to challenge that and to instead focus on developing a true therapeutic community from within. family influence in recovery from severe mental illness the decider skills for self help: cbt and dbt skills to increase resilience, coping and confidence wellness recovery action plan bloomfield sru (unpublished) man's search for meaning: an introduction to logotherapy trauma and recovery guidelines for the management of national specialised rehabilitation unit placements treating depression with tai chi: state of the art and future perspectives reflections on doing, being and becoming theory and practice of group psychotherapy, 5th edn the authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committee on human experimentation with the helsinki declaration of 1975, as revised in 2008. the authors assert that ethical approval for publication of this manuscript. this research received no specific grant from any funding agency, commercial or not-for-profit sectors. none.ethical standards key: cord-335679-dpssd1ha authors: rawson, t.; brewer, t.; veltcheva, d.; huntingford, c.; bonsall, m. b. title: how and when to end the covid-19 lockdown: an optimisation approach date: 2020-05-02 journal: nan doi: 10.1101/2020.04.29.20084517 sha: doc_id: 335679 cord_uid: dpssd1ha countries around the world are in a state of lockdown to help limit the spread of sars-cov-2. however, as the number of new daily confirmed cases begins to decrease, governments must decide how to release their populations from quarantine as efficiently as possible without overwhelming their health services. we applied an optimal control framework to an adapted susceptible-exposure-infection-recovery (seir) model framework to investigate the efficacy of two potential lockdown release strategies, focusing on the uk population as a test case. to limit recurrent spread, we find that ending quarantine for the entire population simultaneously is a high-risk strategy, and that a gradual re-integration approach would be more reliable. furthermore, to increase the number of people that can be first released, lockdown should not be ended until the number of new daily confirmed cases reaches a sufficiently low threshold. we model a gradual release strategy by allowing different fractions of those in lockdown to re-enter the working non-quarantined population. mathematical optimisation methods, combined with our adapted seir model, determine how to maximise those working while preventing the health service from being overwhelmed. the optimal strategy is broadly found to be to release approximately half the population two-to-four weeks from the end of an initial infection peak, then wait another three-to-four months to allow for a second peak before releasing everyone else. we also modelled an ''on-off'' strategy, of releasing everyone, but re-establishing lockdown if infections become too high. we conclude that the worst-case scenario of a gradual release is more manageable than the worst-case scenario of an on-off strategy, and caution against lockdown-release strategies based on a threshold-dependent on-off mechanism. the two quantities most critical in determining the optimal solution are transmission rate and the recovery rate, where the latter is defined as the fraction of infected people in any given day that then become classed as recovered. we suggest that the accurate identification of these values is of particular importance to the ongoing monitoring of the pandemic. . schematic diagram depicting the movement of individuals through the seir network. the function u describes the action of the strategy employed to end lockdown, as people are released from the quarantined group. the arrows linking the two groups operate in both directions, to allow for any "on-off" strategy where people are returned to quarantine. 106 the lowercase greek letters in equations (1) -(8) represent our rate parameters. firstly, β represents the transmission rate . cc-by-nc-nd 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 may 2, 2020. . https://doi.org/10.1101/2020.04.29.20084517 doi: medrxiv preprint informing policy. for this reason our sensitivity analyses (below) also consider transmission rates up to twice as high as these 114 values. note that we consider the population of both i and i q to impact the spread of disease, as the quarantined group are still 115 assumed to occasionally mix with the population (for instance, when leaving their homes to shop for essential items). the 116 parameter c is a scalar between 0 and 1 that captures how effective the self-isolation (i.e. lockdown) measures enforced are in 117 reducing the the rate of sars-cov-2 transmission. 118 119 µ represents the natural, background death rate of the population regardless of the impact of covid-19, and can have 120 important implications for the strength of herd-immunity effects on disease dynamics, as this is the only mechanism in our 121 model through which the recovered population is reduced. the parameter α represents the rate of death directly attributed to 122 sars-cov-2. while the mortality rate of sars-cov-2 has been demonstrated to vary substantially between age classes 10-12 , 123 in its current form our model does not incorporate age-structure and we therefore adopt an age-invariant mortality rate. the parameter σ represents the incubation rate. the exposed population classes, e/e q , capture the effect of the lag be-126 tween people becoming infected (and incubating the disease for several days) and becoming infectious. understanding the size 127 of this effect is of great importance when assessing strategies in which a second lockdown may be enforced because efforts to 128 monitor the subsequent spread of infection must consider the upcoming, but lagged, threat posed by the exposed class. lastly, γ 129 represents the recovery rate and describes how long individuals remain infectious. . cc-by-nc-nd 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 may 2, 2020. the parsimonious nature of our model was chosen to enhance the ease of interpretation of our results and, most impor-148 tantly, to enable the model to be quickly adapted to non-uk populations. different countries currently provide varying levels of 149 epidemiological detail in their reporting of covid-19 cases. by reducing the number of classes and parameters considered, our 150 model is amenable to a wider range of countries and scenarios than the more specific model structures currently published 9, 17 . the result of this modelling choice is that our system captures the broad-scale dynamics of the disease resulting from different 152 lockdown exit-strategies rather than making accurate predictions of the number of infected individuals, which will require 153 continuous, data-driven adaptations applied to our framework. the primary challenge facing policy makers currently is in devising how to return the population to work most safely, ending 156 the lockdown and its detrimental consequences on the economy. the objective is to release as many people from lockdown, as release which, even if gradual, will still be managed with distinct groups of people leaving at different times. our primary 179 results presented in the following section are instead derived from an iterative process in which multiple different release times 180 and portions of the population are trialled across various ranges, with the optimal choice being that which maximises our 181 objective function. all code used to perform these optimal control approaches was performed in matlab, and is available at: 182 https://osf.io/hrt2k/. definition initial conditions and definition of n and n q reference source non-quarantined exposed. a gradual release strategy aims to end the lockdown of the the public from quarantine through multiple staggered releases. expressed mathematically, we seek to release m 1 people at time t 1 , while ensuring that i + i q < i thresh at all times. we . cc-by-nc-nd 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 may 2, 2020. . https://doi.org/10.1101/2020.04.29.20084517 doi: medrxiv preprint therefore, the optimum choice of m 1 and t 1 are those which maximise c 1 . in short, this approach calculates how to release as 192 many individuals as possible, as early as possible, without breaking the infection carrying capacity. after this optimum solution 193 is found, a second release of m 2 people at time t 2 can be similarly calculated after the first release, if people still remain in 194 quarantine. to calculate these outputs, we used ode45, a fourth-order runge-kutta solver in matlab, to solve the system of equa-197 tions (1)-(8) using the initial conditions in table 1 for t from 0 to t 1 . at this point we subtracted m 1 individuals proportionally 198 from s q , e q , i q and r q and added these to s, e, i and r. the system was then solved again from these new points for t from t 1 199 to 400 days. to allow understanding of the effect of different values of some of the parameters presented in table 1 , we operate table 1 . . cc-by-nc-nd 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 may 2, 2020. table 1 . figures 2 and 3 are example simulations, to illustrate general model behavior, but are not optimal solutions. we now consider outcome considered is the objective function, c, for our optimum strategy. defined formally, the total sensitivity index for , where y is the model outcome monitored, and x i is the parameter considered. to determine the optimal timings for an "on-off" lockdown release strategy, both the times at which quarantine was ended, t off i , 258 and the times at which it was reinstated, t on i , were iterated on a mesh of 500 evenly spaced points across a timespan of 0 to 400. once one optimum release pair was found, the process was repeated up to two further times to identify subsequent optimum . cc-by-nc-nd 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 may 2, 2020. . here we have investigated the optimal release of individuals from a state of lockdown. the primary conclusion of our work 282 is that a gradual release strategy is far preferable to an on-off release strategy. we conclude this from the finding that a . cc-by-nc-nd 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 may 2, 2020. . https://doi.org/10.1101/2020.04.29.20084517 doi: medrxiv preprint of time. any decision to begin easing lockdown measures will require constant monitoring and a high-level of population 285 testing to track the likely rise towards a second-peak of infections. we show that employing a gradual release strategy, where 286 groups of the population are slowly released from quarantine sequentially, will slow the arrival of any subsequent infection 287 peaks compared to an on-off strategy, where lockdown is ended for all individuals imminently and reinstated when subsequent 288 infections begin to increase. in all considered instances (i.e. parameter variations), it will not be possible to end lockdown 289 for the entire population for any longer than two weeks, as the number of infected individuals is then expected to quickly 290 overwhelm the health service following such a release. by ensuring that the increase in the number of infected individuals is as 291 slow as possible, this will enable health officials to monitor more accurately the evolving situation, and provide more time to 292 respond to unexpected increases in the number of infected individuals. we note that our approach does not consider the ethical 293 responsibilities that will also impact any policy decision. if enough hospital provision was available, many more people can 294 return to employment, but we recognise this will result in increased risk of further mortalities. as many governments state 295 however, a functioning economy is more able to provide health provision to those with non-covid19 life-threatening illness. for a gradual release strategy, our simulations broadly suggest that a large section of the population should be released 298 from lockdown initially, after the first peak of infections has fully passed. the rest of the population may then be released three 299 to four months later following a likely second peak in infections. again, in a general context, it is optimal to wait for one-to-two 300 weeks after the end of an infection peak before releasing any of the population from lockdown. while it is desirable to return 301 the population to work as early as possible, our optimal calculation states that this one-to-two week "wait" period is crucial in 302 ensuring that the number of infected individuals is as low as possible when ending any lockdown measures, to reduce the growth 303 of new cases. after this sufficient, cautious, wait period has ended, people should then be released from quarantine, with the 304 knowledge that as many as 1 in 100 of them (under the worst-case scenario) may require critical care 10 in the coming months. 305 it is expected that a second peak in infections may be observed one to two months after this release date, and that the remaining 306 population in quarantine should remain so until, once again, several weeks of low newly infected cases daily have been observed. in conclusion, using an optimal control methodology, we have shown that a gradual staggered release of individuals out 362 of lockdown is recommended to ensure that health systems are not overwhelmed by a surge in infected individuals. it has been 363 well observed that older individuals are more likely to require critical care as a result of covid-19 10 . although our analysis 364 does not as yet differentiate by age who should be in any partial lockdown releases, this does indicate that, potentially, the 365 younger population could be the first to be released from lockdown. this would further ease any subsequent strain on the 366 health system, and potentially further bolster a herd-immunity effect. similarly, our analysis does not model the capability of 367 businesses and individuals who have the infrastructure and availability to continue to work remotely. the ongoing threat of covid-19 will require continual monitoring and study in the coming months. it is important to 370 ensure that infections are kept to a minimum, and that the government and relevant services are given enough time to prepare for 371 increases in infections. the findings of this study stress that gradual and cautious action must be taken when easing lockdown 372 measures, to save resources, and lives, while adding to the evidence base of possible routes out of lockdown. . cc-by-nc-nd 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 may 2, 2020. . https://doi.org/10.1101/2020.04.29.20084517 doi: medrxiv preprint estimating the asymptomatic proportion of coronavirus disease 387 2019 (covid-19) cases on board the diamond princess cruise ship epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in wuhan, 390 china: a descriptive study economic effects of coronavirus outbreak (covid-19) on the world economy the mathematics of infectious diseases early dynamics of transmission and control of covid-19: a mathematical modelling study. the 395 modelling the covid-19 epidemic and implementation of population-wide interventions in italy impact of non-pharmaceutical interventions (npis) to reduce covid19 mortality and 399 healthcare demand real estimates of mortality following covid-19 infection estimation of sars-cov-2 mortality during the early stages of an epidemic: a modelling study in hubei, 402 china and northern italy key workers: key facts and questions estimating the number of infections and the impact of non-pharmaceutical interventions on key: cord-258072-6d5ieakl authors: kochhar, anuraj singh; bhasin, ritasha; kochhar, gulsheen kaur; dadlani, himanshu; mehta, viral vijay; kaur, roseleen; bhasin, charanpreet kaur title: lockdown of 1.3 billion people in india during covid-19 pandemic: a survey of its impact on mental health date: 2020-06-18 journal: asian j psychiatr doi: 10.1016/j.ajp.2020.102213 sha: doc_id: 258072 cord_uid: 6d5ieakl nan covid-19 pandemic presented as a black swan event, and as a measure to curtail it the governments of different countries took various approaches (tandon r). however, most countries announced complete lockdown, with draconian travel and social restrictions. on march 24, 2020, the government of india ordered a nationwide lockdown for 21 days, limiting movement of the entire population of 1.3 billion. though this was probably a requisite, even short term lockdowns, quarantine and social distancing can precede long term effects such as symptoms of mental stress and disorder, including insomnia, anxiety, depression, and post-traumatic stress symptoms (wang et al, 2020; brooks et al, 2020) . although there is a deluge of studies published regarding covid-19, there is a paucity of published data regarding the mental health status of the general population. therefore the present survey was intended to assess the mental health impact of the current lockdown on the population of new delhi, india, a week after its imposition to assist the government agencies and healthcare professionals in safeguarding the mental health wellbeing of the community. after reviewing the pertinent literature which included scales such as, impact of event scale -revised (ies-r), perceived stress scale 10 (pss-10), and the international guidelines, a self administered, pre-validated web-based questionnaire with 18 questions in english, exploring age and domains of knowledge about covid-19, understanding reasons for lockdown, stressors which included fear of infection, helplessness and boredom, scarce supplies, and post quarantine stressors related to work, finances and stigma was floated on a famous social media site, facebook (groups from new delhi). those who accepted the invitation and confirmed they were residents of new delhi, were above 20 years of age and had a minimum graduate level education received the questionnaire. the study population consisted of 992 participants (out of the 2876 invited) of which 11.2% were aged between 21-35 years, 46.3% between 36-50 years, 6.5% were over 65 years of age and the remaining 35% were between 51 to 65 years. ethical clearance and consent were obtained for the same. the data was subjected to statistical analysis using spss j o u r n a l p r e -p r o o f 20.0. shapiro-wilk test indicated the data to be normally distributed. results were segregated based on the variable of age and domains that indicated significant association with age have been highlighted in table 1 (p<0.05). chi-square indicated significant associations were observed between age groups and the multivariate variables of knowledge about covid-19, effects of covid-19 on humankind and how the participants felt about the present situation. of the bivariate variables, significant association of age was noted with effects on work and income with sleep and diet pattern changes (graph-1). pearson's correlation analysis of bivariate variables indicated that an increased understanding of the meaning of social distancing enabled its increased practice. also, negative correlation between age and travel plans indicated that decreasing age was associated with increasing possibility of modifications in travel plans. while an increase in fear of acquiring covid-19 by meeting people was correlated with an increase in dietary pattern changes, it displayed negative correlation to sleep pattern changes. the covid-19 outbreak has led to diversified mental health responses depending on the individual's strengths and weaknesses (wang et al, 2020) . knowledge about the disease and the reasons for lockdown are vital tools for successful disease containment (brooks et al, 2020) . in the present study, 92.5% and 97.5% respondents were well acquainted with the knowledge of covid-19 and social distancing or isolation respectively, with 98.2% following it (p<0.05). there was a positive correlation between the awareness of social distancing or isolation and those who were practicing it. present survey revealed 78.5% participants were optimistic and believed that together the disease could be curtailed (p<0.05). however, it has been contemplated that greater the number of individuals following updates about the same, higher are the anxiety levels (moghanibashi-mansourieh, 2020). although social media has played a pivotal role in these times with a splurge of knowledge, information must be gathered from health authorities in order to help distinguish facts from rumours, as facts can help minimize fears. 12.1% participants felt helpless & depressed suggesting that containment, loss of daily schedule, and diminished social and physical contact with others can cause mental fatigue, dissatisfaction, and a feeling of confinement from the remainder of the world. 12.1% participants felt helpless & depressed suggesting that containment, loss of daily schedule, and diminished social and physical contact j o u r n a l p r e -p r o o f with others can cause mental fatigue, dissatisfaction, and a feeling of confinement from the remainder of the world (brooks et al, 2020) . following the imposition of the lockdown, 91.6% of respondents altered or canceled their travel plans. when asked about effects of the disease on mankind, 60.2% believed that it would lead to loss of human lives, 29.8% people were concerned about possible economic slowdown while 1.9% people believed it to be a mere social media hype. when enquired about the activities during the lockdown, only 11% were occupied with hobbies whereas who advises engaging in regular exercising, daily chores, and hobbies during the present covid-19 pandemic for mental health well-being (who, 2020). significant associations were observed between age groups and the multivariate variables of knowledge about covid 19, effects of covid 19 on mankind and how the participants felt about the present situation. sleep disturbances have been shown to be a risk factor for mental disorders. also, sleep quality has been found to be dependent on anxiety, stress and self-efficacy (xiao et al, 2020). moreover, short sleep duration in some individuals has been associated with suicidal tendencies (weber et al, 2020). in current study 55.3% complained of trouble sleeping during this lockdown period, of which people who feared contracting the disease and participants between the age group 35-50 years witnessed maximum alteration in sleep. this could be associated with high levels of anxiety and stress because of isolation, indicating adverse mental health. (rajkumar, 2020) nutritional factors are interlaced with human behaviour, and emotions while playing a critical role in not just the initiation, but also the severity and duration of depression. numerous reports support the view that stress can either increase or decrease caloric intake, and chronic stress exposure can lead to either obesity or anorexia (sathyanarayana rao et al, 2008) . variations in eating patterns were observed in the present study in 79.5% of participants, with people aged 35-50 years experiencing maximum alterations in diet patterns that were positively correlated with the practice of social distancing and fear of acquiring the disease on meeting someone. though not significant, 26.3% of respondents stated that they started consuming more alcohol/drugs/tobacco, raising an alarm, as isolation might lead to an escalation in alcohol misuse and probable development of alcohol use disorder in high-risk persons during and after the pandemic (clay & parker, 2020) . it has been reported that during quarantine, inadequate basic supplies can cause resentment. 35.4% of the participants commented that they could sustain lockdown; however majority of the respondents could not manage either due to lack of basic amenities and medical needs (31.8%) or due to emotional reasons (32.8%) blendon rj also stated that lack of regular medical care was a concern for participants (brooks et al, 2020) . financial hardships are often experienced by individuals during quarantine. lockdown majorly affected the work and income of 63.4% of the population that was positively related to people aged 35-50 years, followed by 50-65 years, impacting their financial status. even though 49.7% of the study population claimed to have sufficient funds to manage the lockdown situation, remaining were either uncertain about it or did not have the resources to sustain it. monetary loss is a stressor during and post isolation because people are unable to work and professional activities are interrupted unprecedentedly; the effects appear to be long lasting. financial loss due to quarantine created severe socioeconomic distress (pellecchia et al, 2015) and was established to be a contributing factor for symptoms of mental health disorders, anger and anxiety (mihashi et al, 2009) . although stigma has aggravated the anguish from many major epidemics in the past, with various studies suggesting that patients were being treated differently, being avoided, treated with fear and suspicion and received critical comments (wilken et al, 2017) , in the current study 71.7% believed that acquiring covid-19 was not a social taboo. the present lockdown which was initially proposed for 21 days was eventually extended for another 3 weeks. longer quarantines have a direct correlation with the mental health outcomes and the duration of the lockdown is a predominant stressor affecting the mental health of individuals (brooks et al, 2020) .during the sars outbreak, many studies investigated the mental health impact on the non-infected community, revealing significant psychiatric morbidities (sim et al, 2010) . even when people adhere to the lockdown, rather than comply, various moral and mental health issues are raised, creating a debate about individual rights versus such public health interventions during a crisis. however, if authorities are strategic in their planning, lockdowns may become more effective and with fewer effects on the mental health of people. author contributions: 1. dr. anuraj singh kochhar: conception and design of the study, acquisition of data and analysis of the same. drafting the article, critically evaluating and giving final approval for the same. 2. dr. ritasha bhasin: conception of design, drafting the article, and critically evaluating it. approving the final version. 3. dr. gulsheen kaur kochhar: conception of design for the study, analysis of the data, drafting the article, revising it and critically evaluating it. approving the final version. 4. dr. himanshu dadlani: conception and design of the study, acquisition of data, or analysis and interpretation of data, drafting the article, critically evaluating and giving final approval. 5. dr. viral vijay mehta: the conception and design of the study supervising the acquisition of data along with analysis and interpretation of data. critically evaluating the article and giving the final approval. 6. ms. roseleen kaur: conception and design of the study, acquisition of data and analysis of the same. drafting the article, revising it, critically evaluating and giving final approval for the same. 7. ms. charanpreet kaur bhasin: condensing the article in its present form. no fees and grants from, employment by, consultancy for, shared ownership in, or any close relationship with, an organisation whose interests, financial or otherwise, has been received by any of the authors. 1. authorship of the paper: authorship has been limited to only those who have made a significant contribution to the conception, design, execution, or interpretation of the reported study. credit author statement attached with the cover letter. 2. originality and plagiarism: the authors have ensured that they have written entirely original manuscript, and if the authors have used the work and/or words of others, that has been appropriately cited or quoted. 3. data access and retention: the datasets used and analyzed during the current study are available from the corresponding author on reasonable request. 4. multiple, redundant or concurrent publication: this manuscript has not been published before and is neither under consideration for publication elsewhere since it was originally submitted to asian journal of psychiatry. the manuscript in its current form is approved by all authors. 5. acknowledgement of sources: proper acknowledgment of the work of others has been specified wherever used. 6. disclosure and conflicts of interest: submitted separately 7. fundamental errors in published works: when ever such an error would be noted / discovered, it will be promptly notified to the journal editor or publisher and cooperated with the editor to retract or correct the paper. 8. reporting standards: work is accurately performed and discussed in its significance. 9. hazards and human or animal subjects: there was no indication of unfavorable effects on safety/risk to the participants of the study. 10. use of patient images or case details: ethical committee approval and informed consent of volunteers obtained, which is documented in the manuscript. no conflict of interest reported by any of the authors. the psychological impact of quarantine and how to reduce it: rapid review of the evidence alcohol use and misuse during the covid-19 pandemic: a potential public health crisis? the lancet public health j o u r n a l p r e -p r o o f key: cord-318437-tzp33iw7 authors: lovrić, mario; pavlović, kristina; vuković, matej; grange, stuart k.; haberl, michael; kern, roman title: understanding the true effects of the covid-19 lockdown on air pollution by means of machine learning() date: 2020-11-06 journal: environ pollut doi: 10.1016/j.envpol.2020.115900 sha: doc_id: 318437 cord_uid: tzp33iw7 during march 2020, most european countries implemented lockdowns to restrict the transmission of sars-cov-2, the virus which causes covid-19 through their populations. these restrictions had positive impacts for air quality due to a dramatic reduction of economic activity and atmospheric emissions. in this work, a machine learning approach was designed and implemented to analyze local air quality improvements during the covid-19 lockdown in graz, austria. the machine learning approach was used as a robust alternative to simple, historical measurement comparisons for various individual pollutants. concentrations of no(2) (nitrogen dioxide), pm(10) (particulate matter), o(3) (ozone) and o(x) (total oxidant) were selected from five measurement sites in graz and were set as target variables for random forest regression models to predict their expected values during the city's lockdown period. the true vs. expected difference is presented here as an indicator of true pollution during the lockdown. the machine learning models showed a high level of generalization for predicting the concentrations. therefore, the approach was suitable for analyzing reductions in pollution concentrations. the analysis indicated that the city’s average concentration reductions for the lockdown period were: -36.9 to -41.6%, and -6.6 to -14.2% for no(2) and pm(10,) respectively. however, an increase of 11.6 to 33.8% for o(3) was estimated. the reduction in pollutant concentration, especially no(2) can be explained by significant drops in traffic-flows during the lockdown period (-51.6 to -43.9%). the results presented give a real-world example of what pollutant concentration reductions can be achieved by reducing traffic-flows and other economic activities. the covid-19 pandemic has caused disastrous health and socio-economic crises across the 50 globe (alabdulmonem et al., 2020; mckee and stuckler, 2020). questions have been raised 51 whether atmospheric pollution is a co-factor in disease development causing a higher lethality 52 rate, especially in highly populated and polluted areas such as those in italy (conticini et al., 53 2020; fattorini and regoli, 2020) . a study from china suggests there is a statistically confirmed 54 relationship between air pollution by means of elevated concentrations of pm 2.5 , pm 10 , co, no 2 55 and o 3 and the covid-19 infection rate (zhu et al., 2020) . another study from italy supports the 56 insight by providing causal relationships between the covid-19 spread and air quality (delnevo 57 et al., 2020 ). an interplay of air quality and the pandemic seems obvious. 58 on the other side, lockdowns have caused significant changes in air quality (dutheil et al., 59 2020). a study on 44 chinese cities (bao and zhang, 2020) showed a decrease in main air 60 pollutants from 5.93-24.67% during the lockdown while megacities such as sao paulo showed 61 even higher concentration drops (40-70%) for some pollutants (krecl et al., 2020) . a study on 62 pm 2.5 in capital cities showed concentration drops of 20-60% during the covid-19 crisis 63 (rodríguez-urrego and rodríguez-urrego, 2020). it is suggested that the pollution drop was 64 mainly driven by a reduction in traffic (kerimray et al., 2020) and industrial activities (li et al., 65 2020). even if lockdowns hinder economic growth and might cause various negative effects in 66 the long term, drops in pollution concentrations may act as another factor which slows disease 67 transmission in tandem with limiting human contact. lockdowns in europe were instituted 68 gradually by means of governmental interventions (desvars-larrive et al., 2020). this massive 69 intervention also poses a unique opportunity to study the change in various aspects of air quality, 70 thus motivating our study. 71 we discuss and explore that for complete understanding of the true factors influencing 72 pollutant concentrations, pure statistical tests or observational comparisons might be inadequate 73 since weather conditions, particle persistence and seasonality affect concentrations by linear and 74 non-linear processes (šimić et al., 2020) . furthermore, transport pathways and source distribution 75 can also play a role in analyzing the effects of the lockdown on pollution by means of trajectory 76 models (zhao et al., 2020) .therefore, a comparison of air quality in 2020 vs other years may be 77 biased since other independent factors such as shifts in heating seasons or weather conditions can 78 affect air quality (schiermeier, 2020) . to be able to solve independent factor some authors 79 proposed fixed effect models (liu et al., 2020; venter et al., 2020) . our proposal is that the 80 pollution level can be solved as a multivariate problem predicted by independent variables 81 elevated from environmental variables and seasonal trends, i.e. there are many effects and they 82 might not have fixed effects because the atmosphere is a very dynamic system. moreover, if one 83 wants to return a full time series, observational and fixed effect methods might fail if not 84 accounted for environmental dependencies. for predicting the pollutant concentrations, we 85 employed the random forest algorithm, a non-linear regression method which has the power to 86 solve multivariate problems independent of the variable type. a complementary approach found 87 j o u r n a l p r e -p r o o f in literature is a forecasting method which accounted for atmospheric and other effects but using 88 a mechanistic instead of a data-driven machine learning approach (menut et al., 2020) . 89 we investigate the effects of lockdown on air quality in an urbanized area in graz, styria, 90 austria. due to the high degree of traffic influence, we have included traffic data into our 91 analysis. furthermore, we have investigated in detail which of the pollutants' concentrations were 92 influenced by the lockdown. as such, the outcome of our study serves as a guide for future 93 interventions and their expected associated change in the pollutants' concentration changes. 94 our study contains traditional exploratory statistical analysis, including the utilization of principal 96 component analysis (pca) to explore key attributes. however, the primary analysis is based on 97 machine learning (ml) models which were used to capture historical relationships between the 98 attributes and compare the predictions to true pollution values after the covid-19 lockdowns 99 were imposed. we utilize historical data which matches the time frame of the lockdown for the 100 preceding years, but also include traffic flow data to represent the drop in mobility. 101 data description 102 we collected environmental, pollution and weather data from publicly available sources provided 103 by the austrian government 1 . in order to obtain a realistic picture of air quality during the 104 lockdown, we analyzed the long term measurement data from january 2014 to may 2020 from 105 five measurement sites in the austria city of graz (süd (eng. south) -s, nord (eng. north) -n, 106 west (eng. west) -w, don bosco -d, ost (eng. east) -o); figure 1 ). graz is a medium-sized 107 european city which has much in common in respect to size and layout to many other european 108 urban areas. the latter two measurement sites are situated on arterial roads with high traffic 109 volumes, especially during morning and evening rush hours. the most polluted measurement site 110 of graz is don bosco that struggles to meet the annual no 2 and pm 10 regulatory limits of the 111 eu-council directive 96/62/ec. this is primarily because of the traffic related emissions, but 112 also because of the emissions from a nearby steel-and iron-mill (hinterhofer, 2014) . although 113 graz east is located at a heavily frequented commuter-arterial, mean pollutant concentrations are 114 lower than at don bosco. graz south is situated at a secondary road segment but also records 115 higher pollutant concentrations due to an industrial complex nearby. graz north and west are 116 classified as urban background sites and are located near minor roads with no specific emission 117 contributors in immediate vicinity. a more detailed site description, photos of the sites and 118 historical overview of the sites is given in moser et al., 2019. 119 with the intention of understanding the potential effects of traffic, the traffic flow for the 120 city of graz was accessed. the traffic flow data were mainly measured with inductive loop 121 detectors where the detectors measure the change in field when objects pass over them. once a 122 vehicle drives over a loop sensor, the loop field changes which allows the detection of the 123 presence of an object (a vehicle). the "traffic control and street lighting unit of the city of graz" 124 monitors and records the data at one-minute time frequency and provided data from january 2017 125 to may 2020 for two sites, namely don bosco and ost. 126 to determine the start, end, and duration of the austrian lockdown, we extracted these 127 data from a dataset which contains a collection the air quality data covers pm 10 and no 2 , from five sites (d, n, o, s, w) described in figure 1 values were imputed by backfilling (see missing value counts in table 1 ). the processed data 160 consists of 2324 days and 60 variables in total and is provided in table format within a persistent 161 data repository (lovrić et al., 2020b) . the traffic data were aggregated to a daily frequency and 162 stored as a time series for the two sites (o, d). the processed traffic data ranges from january 163 2017 to may 2020. values (values from the previous two days). these predictive variables allowed the machine 190 learning model to capture seasonal behavior from activities such industrial production and traffic 191 flows and therefore, can be thought of as surrogate variables. 192 the machine learning algorithm used was random forest regression (rf) (breiman, 193 2001) which has been utilized in a number of previous air pollution models and air quality data shown in figure 4 . 273 one can see that for pm 10 the ml models show a larger concentration drop whereas for no 2 we 345 see a smaller concentration drop with the ml models (table 3, which are located close to roads (figure 1 ). the traffic measured at the detector loops at these 356 measurement sites showed a reduction of 45.6% at o and 51.6% at d respectively (table 3, industrial processes were also significantly curtailed in and around graz during the lockdown 362 period, it could be expected that pm 10 would decrease more drastically. the lack of reduction 363 outlines the more numerous and complicated processes which drive pm 10 concentrations, and/or 364 a high lag in the relationships involved in the pm 10 related variables, i.e., the duration of the 365 intervention was too short to lead to a significant drop in this particular pollutant. 366 our results agree regarding a traffic-related drop in no 2 and an increase in o 3 with a 367 variety of studies conducted on air quality issues regarding the covid-19 lockdown. a study 368 from northern china shows that a reduction in no 2 and pm was most likely caused by a 369 reduction in traffic and industrial activities . furthermore, they also show an 370 increase in o 3 consistent with our analysis as well. another study from china where lockdown 371 measures were introduced earlier than in austria (1 st january -29 th february 2020) (shi and 372 brasseur, 2020) reveals a reduction of no 2 up to 60% and a o 3 increase of 40-100%. a two-stage 373 lockdown in india, which was introduced concurrently with the austrian lockdown (mahato et 374 al., 2020), and data from this region shows a pm 10 concentration drop around 60% compared to 375 the same period in 2019. these results are contrary to ours, since graz did not experience such a 376 pm 10 drop, revealed by both historical comparison and machine learning prediction. reasons for 377 that may be domestic heating which is difficult to evaluate since data on heating and stays in the 378 city are difficult data to obtain. the same study observed concordant results in the drop of no 2 (-379 52.68%). a study from the uk, which employed also a historical comparison shows a reduction 380 of 48% in no 2 concentration and an average increase in o 3 concentration of 11% across 126 381 urban sites, which is also consistent with our analysis. a maybe more relevant comparison to our 382 study is given by (menut et al., 2020) shows a decline in pm 10 during the lockdown we believe the fixed-effect and observational 397 methods may not be enough to deliver a definite conclusion on the concentration reductions. it is 398 a limitation of this study that long-range transport data and chemical speciation of the particulate 399 matter is not available. more efforts must be put into chemical speciation of pm at the individual 400 sites, especially measurement techniques which deliver "online" data. 401 in this work, we have explored the changes in air pollutant concentrations during the covid-19 403 lockdown for the city of graz, austria. the exploration illuminated the relative influences of 404 observed meteorological variables on a wide range of pollutants for an unpresented historic event 405 of human society. besides using explorative methods, we employed random forest regression to 406 analyze the differences between predicted (expected) and observed (true) pollution levels based 407 on environmental data. 408 our prediction models showed good generalization and performance for the analyzed 409 pollutants indicating that the selection of independent variables (predictors) was sufficient to 410 explain changes in pollutant concentrations. for pm 10 principal component analysis covid-19: a global public health disaster does lockdown reduce air pollution? evidence from 44 cities in 503 northern china random forests can atmospheric pollution be considered a co-factor in 508 extremely high level of sars-cov-2 lethality in northern italy? particulate matter and covid-19 disease diffusion 511 in emilia-romagna (italy). already a cold case? computation a structured open dataset of 518 government interventions in response to covid-19, medrxiv. cold spring harbor 519 laboratory press covid-19 as a factor influencing air pollution? 521 role of the chronic air pollution levels in the covid-19 outbreak 524 risk in italy using meteorological normalisation to detect interventions in 529 air quality time series random forest 532 meteorological normalisation models for swiss pm10 trend analysis random forest 535 meteorological normalisation models for swiss pm10 trend analysis more dust blows out from north africa [www document anteil der verkehrsbedingten pm10 und pm2,5 emissionen aus abrieb 541 und wiederaufwirbelung an der feinstaubbelastung in österreich assessing air quality changes in large cities during covid-19 lockdowns: 545 the impacts of traffic-free urban conditions in almaty drop in urban air pollution 548 from covid-19 pandemic: policy implications for the megacity of são paulo air quality changes during the covid-19 lockdown over the yangtze 553 river delta region: an insight into the impact of human activity pattern changes on air 554 pollution variation abrupt decline in tropospheric nitrogen dioxide over china after the outbreak of 558 covid-19 prediction of 560 anode lifetime in electro galvanizing lines by big data analysis (including covid-19 lockdown) data from graz 566 machine learning in prediction of intrinsic aqueous solubility of drug-like compounds: 567 generalization, complexity or predictive ability? chemrxiv effect of lockdown amid covid-19 pandemic on air 570 quality of the megacity delhi if the world fails to protect the economy, covid-19 will damage 573 health not just now but also in the future impact of 576 lockdown measures to combat covid-19 on air quality over western statistische analyse der luftqualitätin graz anhand von 579 feinstaub und stickstoffdioxid air quality during the covid-19: pm2.5 581 analysis in the 50 most polluted capital cities in the world why pollution is plummeting in some cities -but not others the response in air quality to the reduction of chinese 586 economic activities during the covid-19 outbreak applying machine learning methods 589 to better understand, model and estimate mass concentrations of traffic-related pollutants at 590 a typical street canyon covid-19 lockdowns cause global 593 air pollution declines source apportionment of particulate matter in europe: a review of methods and results changes in air quality related to 601 the control of coronavirus in china: implications for traffic and industrial emissions air pollution episodes during the covid-604 19 outbreak in the beijing-tianjin-hebei region of china: an insight into the transport 605 pathways and source distribution association between short-term exposure to air 608 pollution and covid-19 infection: evidence from china ☒ the authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐the authors declare the following financial interests/personal relationships which may be considered as potential competing interests:j o u r n a l p r e -p r o o f key: cord-330562-dabjcvno authors: poli, piero; boaga, jacopo; molinari, irene; cascone, valeria; boschi, lapo title: the 2020 coronavirus lockdown and seismic monitoring of anthropic activities in northern italy date: 2020-06-10 journal: sci rep doi: 10.1038/s41598-020-66368-0 sha: doc_id: 330562 cord_uid: dabjcvno in march/april 2020 the italian government drastically reduced vehicle traffic and interrupted all non-essential industrial activities over the entire national territory. italy thus became the first country in the world, with the exception of hubei, to enact lockdown measures as a consequence of the covid-19 outbreak and the need to contain it. italy is also a seismically active area, and as such is monitored by a dense permanent network of seismic stations. we analyse continuous seismic data from many stations in northern and central italy, and quantify the impact of the lockdown on seismic ambient noise, as a function of time and location. we find that the lockdown reduces ambient noise significantly in the 1–10 hz frequency range; because natural sources of seismic noise are not affected by the lockdown, the seismic signature of anthropic noise can be characterised with unprecedented clarity, by simply comparing the signal recorded before and after the lockdown. our results correlate well with independent evaluations of the impact of the lockdown (e.g., cell phone displacements), and we submit that ambient-noise seismology is a useful tool to monitor containment measures such as the coronavirus lockdowns. noise is of interest to geoscientists, as it can be used at relatively small scales, for instance in mapping and monitoring efforts 9, 10 . while earlier studies have attempted to characterise high-frequency seismic noise 7, [11] [12] [13] [14] , the current lockdown of industrial activities and reduction in road and train traffic in italy is an unprecedented opportunity to discriminate it from ambient noise of natural origin. italy is a highly industrialized and urbanized country, densely covered with non-stationary noise sources 15 , such as traffic and industry-induced vibration 16 . this is particularly true in its northern regions, which account for 70% of the country's entire industrial output, and where lockdown measures have been enacted earlier than everywhere else in europe. we analyse continuous data from an array of broadband seismic stations, located in the vicinity of known industrial districts in lombardy, emilia-romagna and tuscany (fig. 1) ; we identify the spectral signature of the march 2020 lockdown, and take advantage of the lockdown to quantify and evaluate the spectral signature of anthropic activities. importantly, measuring the overall reduction in seismic energy associated with the lockdown is also a way to quantify its effects; this is relevant to governmental entities, wishing to monitor the effectiveness of the measures being taken. we downloaded publicly available, continuous, three-component seismic recordings from a set of permanent broadband stations, part of the italian national seismic network operated by the istituto nazionale di geofisica e vulcanologia 17 . all instruments have a flat response at frequencies between ~0.01 and ~10 hz, or broader; we remove ("deconvolve") instrument response from the data prior to our analysis. the locations of stations employed in most of our study are shown in fig. 1 . stations were selected based on their proximity to industrial districts; in particular, miln is located near the city of milano, with a particularly high concentration of vehicle traffic and industrial activities. the seismic signature of the containment measures in italy is apparent from a relatively simple analysis of continuous recordings at station miln, located within the city limits of milano, in a busy area near the university of milano campus and the lambrate train station. we compute spectrograms (fig. 2) by fourier-transforming 1-hour-long segments of continuous signal, with a 30-minute overlap between subsequent segments; for each calendar day, all segments are then averaged, and the squared modulus of the resulting average fourier transform is computed: this way, a single "power-spectral density" (psd) function is obtained, for each station, component (east-west, north-south, vertical) and calendar day. figure 2 shows clearly that the lockdown has a relevant impact on recorded seismic noise over a broad frequency range; its effect disappears at frequencies below 1 hz, where anthropic noise is weaker. the energy drop associated with the lockdown is comparable with that occurring every weekend and during the winter break, both in 2018/19 and 2019/20. interestingly, loss of energy is gradual over time, starting with the first lockdown measures on february 21, and increasing with time until a plateau is reached around march 22 (interruption of non-essential industrial activities). a trend similar to that seen in fig. 2 has also been found through the analysis of cell phone displacements 18 . this suggests that vehicle traffic, which was significantly reduced (particularly in and around milano) already with the february measures, contributes significantly to the entire spectrum of anthropic noise; there is also episodical evidence from the press that a number of factories were closed based on the unilateral decision of their owners, before the government-imposed lockdown. the analysis applied to station miln is repeated for all seismic stations of fig. 1 , and the results are illustrated in figs www.nature.com/scientificreports www.nature.com/scientificreports/ lockdown measures apparently impact all stations under consideration, but the character of their effects changes in various ways with station location. in the case of fir, located in the city of florence, the signature of the winter break is almost negligible, while the february/march lockdown still has a prominent effect; it might be possible to interpret this observation through the analysis of anthropic activities usually taking place in the area (e.g., tourism, which is presumably not reduced by the holiday). the drop is gradual at all stations, with no specific governmental decree standing out with respect to the others. at station prma, a slight increase in ambient noise occurs after february 21 and before march 9. we next characterise anthropic noise by evaluating variations in the spectra of seismic ambient noise before and after the implementation of lockdown measures. we compute the ratios of the psd measured (as described in sec. 3) on tuesday march 31 2020, to that measured at the same station on tuesday december 3 2019. we carry out this calculation separately for each component, and for all stations analysed thus far; the results of this exercise are shown in fig. 5 . the energy associated with ambient signal is clearly reduced for all stations, at all frequencies in the range of interest. at each station, psd ratios change with frequency almost exactly in the same way for all components. near 1 hz, all stations show a more or less rapid decline in the psd ratio, with ambient noise being more effectively reduced as frequency grows. this trend continues all the way to 20 hz at stations euct and prma, while other stations show a more complex behaviour. above 10 hz station miln stands out, its psd ratio growing quickly with frequency. anthropic noise is known to be relevant at frequencies above 1 hz, and to consist of a range of different excitation mechanisms 7, 15, 19, 20 . natural sources such as rain, wind 21 and sea/ocean waves are typically characterized by frequencies below 1 hz, and are obviously not affected by the lockdown. we infer that, by taking the ratio of noise spectra before and after the lockdown, an estimate of the spectral character of anthropic noise is obtained, and the spectra in fig. 5 can help us estimate the nature of anthropic noise in the region of interest, independent of the lockdown; the frequencies where the psd of ambient signal is most reduced by the lockdown are those where, in normal times, the contribution of anthropic activities to seismic ambient noise is most important. the fact that most energy loss associated with the lockdown is at frequencies between 1-10 hz is coherent with what is known of the typical signature of industrial activity and vehicle traffic 22 . seismic data recorded during the lockdown might be particularly useful in identifying sources of anthropic noise, which could be employed by geophysicists, after the lockdown, e.g. to characterise the upper subsoil by cross correlation of ambient signal 23, 24 . we further analysed the relationship between ambient noise recorded on different components, finding the "h/v" ratio between the psds of horizontal-component and vertical-component signals: first, the psd of each component of signal recorded on a given day is averaged in the frequency range 1-10 hz; then the arithmetical average of the resulting east-west-component and north-south-component values is taken; finally, the ratio of the resulting horizontal psd to the vertical one is computed. the procedure is iterated for each station and for each day between december 1 and march 31, and the results are shown in fig. 6 . in general, the value of h/v is related to how seismic energy in the ambient-noise field is distributed in the form of compressional, shear and surface waves 7, 25, 26 ; changes in h/v after vs. before the lockdown would reveal whether the reduction in anthropic noise affects one of these seismic phases/components more or less importantly than the others; in other words, whether traffic and industry-induced vibration can be associated to one particular constituent of the seismic field. figure 6 shows that the lockdown measures have no effect on h/v, and we infer that, while anthropic noise is reduced significantly by the lockdown (fig. 5) , the relative contributions of compressional, shear and surface waves remain approximately constant: the noise wave field is stable in the fig. 3 are plotted here on a single graph, for comparison; for each station, the average value of the psd observed in the time interval of interest is subtracted from the corresponding curve, prior to plotting, as this can change significantly from station to station, but is not relevant to our analysis. each colour corresponds to one station, as specified. again, the dates of mentioned governmental decrees are highlighted as in fig. 2 . the italian territory is densely covered by seismic instruments, and by repeating our analysis on the entire network of available stations we are able to quantify the spatial dependence of anthropic noise reduction. for each station, for each day, the psd of signal recorded 6 am to 8 pm is computed, and averaged over different frequency bands. in practice, we employ the direct fourier method 27 , as implemented in the obspy package 28, 29 : this is standard procedure to identify artefacts related to station operation, episodic cultural noise, overall station quality and level of earth noise at each site. to emphasize the change in ambient noise with the lockdown, we plot the difference between the values so obtained on three dates in 2020, and reference values obtained conducting the same calculation on data recorded for five months until the lockdown, and averaging. we include as supplementary material s1 an animated version of fig. 7 , showing the psd at the same stations, october 7, 2019 through april 1, 2020; through this time-dependent visualization, the drastic effects of the lockdown are further emphasized. our main result, that noise be strongly reduced after the lockdown in the "cultural" frequency range, is confirmed by fig. 7 , and extended to most of northern italy. between 1-3 hz, the lockdown effects are more pronounced in the lombardy and veneto regions than in central italy and along the apennine range. the most important reductions in ambient noise are recorded by stations along the alpine arc, near torino, milano and verona, and in the city of florence. we have analysed continuous data from northern italy, and quantified the effects of the march 2020 coronavirus lockdown on the seismic ambient noise field. we confirm that this effect is significant, and easily observed in our data: see in particular figs. 3 and 4. the italian government first imposed a reduction of people (and therefore vehicle) movement, on march 9; we find that this date marks the beginning of a gradual loss in ambient-noise energy at all frequencies, which we attribute to the reduction of road and railroad traffic in the region of interest. depending on the station, the energy curve flattens out, or starts to decline more slowly towards the beginning of april, despite the more stringent measures imposed at that time (interruption of all non-essential industrial activities). a similar trend has been found from cell-phone displacement data 18 . one implication of our observations is that seismic data could be useful for governmental institutions to monitor the effectiveness of measures involving a reduction or interruption of human activity in a given area. it is understood that the lockdown only reduces noise of anthropic origin; it follows that by comparing the fourier spectrum of seismic ambient noise before and after the lockdown (fig. 5) , one can attempt to characterise anthropic noise. we find that, confirming earlier estimates 30,31 , anthropic noise becomes dominant at frequencies coronavirus lockdowns have changed the way earth moves seismic imaging and monitoring with ambient noise correlations stationary-phase integrals in the cross-correlation of ambient noise a theory of the origin of microseisms seismic noise in fennoscandia, with emphasis on high frequencies variations in broadband seismic noise at iris/ida stations in the ussr with implications for event detection the nature of noise wavefield and its applications for site effects studies: a literature review emergence of broadband rayleigh waves from correlations of the ambient seismic noise faster, better: shear-wave velocity to 100 meters depth from refraction microtremor arrays rain and small earthquakes maintain a slow-moving landslide in a persistent critical state characterization of and correction for cultural noise h/v ratio: a tool for site effects evaluation. results from 1-d noise simulations on the stability and reproducibility of the horizontal to vertical spectral ratio on ambient noise: case study of cavola, northern italy cultural noise and the night-day asymmetry of the seismic activity recorded at the bunker-east (bke) vesuvian station maninduced low frequency seismic events in italy a catalogue of non-tectonic earthquakes in central-eastern italy italian seismological instrumental and parametric database (iside) the reduction of social mixing in italy following the lockdown observations and modeling of seismic background noise recent advances in seismology spectral analysis of seismic noise induced by rivers: a new tool to monitor spatiotemporal changes in stream hydrodynamics sources of long range anthropogenic noise in southern california and implications for tectonic tremor detection shear wave structural models of venice plain, italy, from time cross correlation of seismic noise train traffic as a powerful noise source for monitoring active faults with seismic interferometry observation of equipartition of seismic waves seismic velocity change patterns along the san jacinto fault zone following the 2010 m 7.2 el mayor-cucapah and m 5.4 collins valley earthquakes an algorithm for the machine calculation of complex fourier series obspy: a bridge for seismology into the scientific python ecosystem seismic noise analysis system, power spectral density probability density function: stand-alone software package. united states geological survey open file report seismic noise level variation in south korea global oceanic microseism sources as seen by seismic arrays and predicted by wave action models rete sismica nazionale (rsn) mediterranean very broadband seismographic network (mednet) north-east italy seismic network. international federation of digital seismograph networks regional seismic network of north western italy. international federation of digital seismograph networks the generic mapping tools version 6 we downloaded and analysed continuous seismic data provided by the istituto nazionale di geofisica e vulcanologia, the osservatorio geofisico sperimentale, the university of genova. the generic mapping tools were used to generate the map in figure 1 . piero poli was supported by the european union horizon 2020 research and innovation programme (grant agreements, 802777-monifaults). the authors declare no competing interests. supplementary information is available for this paper at https://doi.org/10.1038/s41598-020-66368-0.correspondence and requests for materials should be addressed to p.p.reprints and permissions information is available at www.nature.com/reprints.publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. key: cord-263174-dpa6yjao authors: christoforidis, athanasios; kavoura, evgenia; nemtsa, aggeliki; pappa, konstantina; dimitriadou, meropi title: coronavirus lockdown effect on type 1 diabetes management οn children wearing insulin pump equipped with continuous glucose monitoring system date: 2020-07-08 journal: diabetes res clin pract doi: 10.1016/j.diabres.2020.108307 sha: doc_id: 263174 cord_uid: dpa6yjao abstract aims on the 10th of march, greece imposed the closure of schools and universities and a full lockdown a few days later in order to counter the spread of the coronavirus outbreak. our aim was to monitor the effect of the coronavirus lockdown in diabetes management in children with type 1 diabetes mellitus (t1dm) wearing insulin pump equipped with continuous glucose monitoring system. methods in 34 children with t1dm on medtronic 640g insulin pump equipped with the enlite sensor uploaded carelink data were categorized in 2 three-week periods before and after the 10th of march. results mean time in range (tir) did not significantly differ between the two periods. however, a significantly higher coefficient of variation (cv) indicating an increased glucose variability in the pre-lockdown period was observed (39.52% versus 37.40%, p=0.011). blood glucose readings were significantly fewer during the lockdown period (7.91 versus 7.41, p=0.001). no significant difference was recorded regarding the total daily dose of insulin and the reported carbohydrates consumed. however, the meal schedule has changed dramatically as the percentage of breakfast consumed before 10.00 a.m. has fallen from 80.67 % to 41.46 % (p<0.001) during the lockdown. correspondingly, the percentage of dinner consumption before 10.00 p.m. significantly fell during the lockdown period (60.22 % versus 53.78 %, p=0.019). conclusions glycemic control during the coronavirus lockdown can be adequately achieved and be comparable to the pre-lockdown period in children with type 1 diabetes mellitus wearing insulin pump equipped with sensor. at the dawn of 2020, central china faced the outbreak of a highly transmittable, novel strain of the coronavirus, causing severe illness that was subsequently named sars-cov-2 [1] . the coronavirus disease is characterized by severe acute respiratory syndrome and has a high mortality rate especially among the elderly and people with serious underlying medical conditions irrespectively of age [2] . by the 11 th of march and as more than 110 countries have reported numerous cases of covid-19, who declared this outbreak as a pandemic [3] . as of the 13 th of march, europe was the active center of the pandemic and shortly after, all european countries began reporting confirmed cases and deaths. one after the other, most european countries have implemented various degrees of lockdowns to counter the spread of the coronavirus outbreak [4] . greece imposed the closure of all educational institutions on the 10 th of march after only 89 confirmed cases of covid-19 and a full lockdown two weeks later, much earlier than many of its european neighbors. so far, these quick reflexes have been effective as the number of covid-19 cases and deaths have been one of the lowest in europe, nonetheless, there is a long way ahead. the lockdown and the subsequent social distancing practices, which, in some cases, meant complete isolation, has caused a major change in people's daily living routines. emerging evidence supports that people have become less physically active and frequently consume a nutritionally unbalanced diet [5] . people with chronic health conditions might have limited access to the healthcare system and even to health supplies and equipment. for children and adolescents with type 1 diabetes mellitus (t1dm) all these factors may lead to impaired glycemic control as physical activity, healthy eating and even a steady daily routine contribute substantially in a more effective diabetes management [6, 7] . the aim of this study was to monitor the effect of the coronavirus lockdown in glycemic variability, insulin requirements and eating portions and habits in children with t1dm wearing insulin pump equipped with continuous glucose monitoring system. thus, a direct comparison of all these parameters was conducted between a three-weeks period during the lockdown and an equal time period prior to the lockdown. we invited children with type 1 diabetes, followed in our pediatric diabetes outpatient clinic and wearing a medtronic minimed 640g insulin pump accompanied with enlite tm sensor and guardian tm 2 link transmitter to upload their data on the carelink system one month since the initiation of the coronavirus lockdown. the study was performed in accordance with the helsinski declaration of 1975 and was approved by the scientific and administrative council of hippokration general hospital of thessaloniki. children and their caregivers were informed for the nature and the purpose of the study and a verbal consent was obtained for every participant. exclusion criteria included: i) recent diagnosis of t1dm (less than 6 months), ii) chronic or acute medical condition or medication that would be likely to interfere with glucose metabolism for 2 weeks prior and 2 weeks after the study period, iii) less than 3 months experience with the pump and the sensor iv) sensor duration for the study period of less than 75%, v) incomplete or missing data, or inability to upload data using the carelink system and v) unwillingness to participate in the study. figure 1 shows a flow chart of the study population selection. carelink data that were uploaded by the patients and their caregivers were analyzed as categorized into 2 periods: i) lockdown period (3 weeks period starting from the 11 th of march 2020) and ii) period prior to the lockdown (3 weeks period ending on the 10 th of march 2020). demographic data regarding date of birth, date of t1dm diagnosis and date of medtronic 640g pump first use were extracted from patients' medical files. additionally, anthropometric parameters including weight and height and pubertal status were extracted from the records of patient's last visit to the clinic, and their bmi was calculated as the ratio weight/height 2 (kg/m 2 , quetelet index). for every anthropometric parameter, z-scores were calculated according to cdc standards [8] . recent hba1c values were also recorded for every patient. shapiro-wilk test was used for assessing the normality of the studied parameters. wilcoxon signed ranks test for parameters with normal and skewed distribution respectively. in parameters with normal distribution linear correlations were calculated with the pearson's correlation coefficient, whereas spearman's correlation coefficient was employed for non-parametric variables. a p value of 0.05 or less was considered statistically significant. thirty-four children (16 boys, 47,06%) with t1dm were finally enrolled in the statistical analysis. their decimal mean age at the time of the study was 11.37 ± 4.45 years, ranging from 2.52 to 18.59 years. mean age at the time of the diagnosis of t1dm was 6.23 ± 3.43 years. mean time using insulin pump was 2.66 ± 2.06 years. descriptive demographic and anthropometric characteristics of the studied population are presented in table 1 . following data stratification in the two studied periods, a few significant differences arose. table 2 , whereas figure 2 illustrates the significant changes in meal timing. our data showed that glycemic control during the coronavirus lockdown period can be adequately achieved and be comparable to the pre-lockdown period in children with type 1 diabetes mellitus wearing insulin pump equipped with sensor. our results are in accordance with those reported in 2 emerging studies showing no difference in glycemic control during coronavirus lockdown period in both adults [9] and adolescents [10] with t1dm on hybrid closed loop system. interestingly, bonora et al have just shown improved metabolic control in adult patients with t1dm that have stopped working, in contrast, no difference in glycemic control was reported in those that continue to work during the lockdown period [11] . although the reported amount of carbohydrates consumed did not differ significantly between the two periods in our data, meal timing has drastically moved to a looser routine with frequent late-night eating and a significantly increased percentage of consuming the first meal of the day later that 10.00 a.m. in a large, recent study on adult population with t1dm, skipping breakfast was associated with lower odds of reaching good glycemic control and higher mean blood glucose values [12] . however, in our study, we could not declare our population as "breakfast skippers" or "late-awakers", as the information regarding the actual wake up time is not available through the report obtained from the carelink system. in the previous study, higher number of reported meals during the day was also associated with a higher variability in blood glucose measurements; nevertheless, with a better overall glycemic control [12] . better glycemic control was also associated with higher number of eating occasions in two studies performed in adolescents with t1dm [7, 13] , whereas skipping meals was associated with higher odds of suboptimal hba1c in a study of 655 children with t1dm [14] . of particular interest, skipping breakfast has been associated with increased postprandial glycemic response after lunch in a recent, experimental study of healthy young individuals [15] . regarding physical activity, in a recent study assessing the health and wellbeing of normal chinese adults living and working after one month of restrictions to contain the covid-19 outbreak results showed that for those who exercise regularly and over 2.5 h per day, life satisfaction was negatively associated with the level of restrictive measures [5] . on the other hand, individuals who exercise less than half an hour a day, life satisfaction was significantly positively associated with the level of restrictive measures in more severely affected locations, reflecting a better justification or rationalization of their inactive lifestyles [5] . in patients with t1dm, regular physical activity associates with several positive physical health effects including improvement of cardiovascular function and blood lipid profile as well as enhancement in psychological well-being [16] . however, these beneficial effects on health-related outcomes do not perfectly coordinate with improvements in glycemic control something which is mainly attributed to increased glucose variability during exercise [17] . in accordance to that, our results showed a statistically decreased coefficient of variation (cv) of mean blood glucose values, an indirect and inverse indicator of glucose variability, during the lockdown period. on the other hand, in the recent study by tornese et al, 8 adolescents with t1dm that continued their physical activity during the lockdown period showed improved glycemic control compared to 5 adolescents who discontinued their regular physical activity [10] . additionally, in our study, the percentage of basal insulin requirements was increased during the lockdown period indicating increased sedentary behavior, however, with no increase in the total daily insulin dose indicating a compensatory bolus reduction as a result of reduced carbohydrates consumption. finally, our data showed that there was a significant reduction in blood glucose readings during the lockdown period. as expected, the number of blood glucose readings was inversely correlated to age (r=-0.400, p=0.019, data are not shown), and also the difference in blood glucose readings performed during the lockdown and the period before the lockdown was inversely correlated to the frequency of testing prior to lockdown period with a correlation that was approaching significance (r=-0.326, p=0.060, data are not shown). could this reduction in using glucose strips reflect an endogenous stressor associated with fear of running out of medical supplies? shortages in basic and medical supplies have been considered a major stressor during quarantine and continues to be associated with anxiety and anger even 4-6 months after release [18] . in order to be prepared and provide effective pharmaceutical care for the general population and patients with chronic diseases during the coronavirus outbreak, chinese pharmacists have recently published recommendations and guidelines [19] . one of the study's main limitation derives from the subjective nature of information provided in the carelink system, especially in the reported carbohydrates consumed and the absence of reporting physical activity and sleep patterns. additionally, our study population includes only patients wearing insulin pump and sensor, representing a group of patients and caregivers familiar with technology and usually aiming at achieving the best possible glycemic control; thus, generalization of our results to all patients with t1dm should be avoided. furthermore, as this is a single-center study, readers should be cautious not to expand our conclusions in different populations and clinical scenarios. also, the nature of traditional control study does not allow the adjustment of some potential confounders such as accompanying policies related to the outcomes. on the other hand, our study's statistical strength stems from the paired comparison of each patient serving as a control to themselves. thus, we manage to show that during the coronavirus lockdown period, children with type 1 diabetes on insulin pump with continuous glucose monitoring and an automated insulin suspension system can achieve adequate glycemic control regardless of decreased physical activity, and possibly a nutritionally imbalanced diet and unpredictable meal schedules. despite all these limitations, our patients and caregivers proved that they are well-trained and experienced in dealing with any possible unpredictabilities that might come in their way and maintain a good glycemic control. this research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. table 2 . parameters of glycemic control, insulin requirements and carbohydrates consumption during the three studied (results are expressed as mean ± standard error).  glycemic control during the coronavirus lockdown can be comparable to the pre-lockdown period in children with type 1 diabetes mellitus wearing insulin pump equipped with sensor  mean time in range (tir) did not significantly differ between the two periods. however, higher coefficient of variation (cv), indicating an increased glucose variability in the pre-lockdown period, was observed  meal schedule has changed dramatically during the lockdown with a decrease in the percentages of early breakfast and early last meal of the day sars-cov-2: virus dynamics and host response covid-19: risk factors for severe disease and death health organization holds news conference on the coronavirus outbreak covid-19: towards controlling of a pandemic unprecedented disruption of lives and work: health, distress and life satisfaction of working adults in china one month into the covid-19 outbreak children and adolescents: standards of medical care in diabetesd2019 eating patterns in adolescents with type 1 diabetes: associations with metabolic control, insulin omission, and eating disorder pathology centers for disease control and prevention 2000 growth charts for the united states: improvements to the 1977 national center for health statistics version effects of covid-19 lockdown on glucose control: continuous glucose monitoring data from people with diabetes on intensive insulin therapy glycemic control in type 1 diabetes mellitus during covid-19 quarantine and the role of in-home physical activity glycaemic control among people with type 1 diabetes during lockdown for the sars-cov-2 outbreak in italy meal timing, meal frequency, and breakfast skipping in adult individuals with type 1 diabetesassociations with glycaemic control metabolic control and diet in finnish diabetic adolescents sweets, snacking habits, and skipping meals in children and adolescents on intensive insulin treatment association between breakfast skipping and postprandial hyperglycaemia after lunch in healthy young individuals continuous glucose monitoring and exercise in type 1 diabetes: past, present and future the psychological impact of quarantine and how to reduce it: rapid review of the evidence recommendations and guidance for providing pharmaceutical care services during covid-19 pandemic: a china perspective the authors would like to thank mrs vivian pavlopoulou for the linguistic revision of the manuscript. the authors declare no conflict of interest. key: cord-288851-lz5qc1f3 authors: rastogi, ashu; hiteshi, priya; bhansali, anil title: improved glycemic control amongst people with long-standing diabetes during covid-19 lockdown: a prospective, observational, nested cohort study date: 2020-10-21 journal: int j diabetes dev ctries doi: 10.1007/s13410-020-00880-x sha: doc_id: 288851 cord_uid: lz5qc1f3 background and aims: covid-19 is likely to affect the lives of individuals with type 2 diabetes. however, the effect of covid-19 lockdown on physical activity and glycemic control in such individuals is not known. we studied the physical activity and glycemic control during lockdown in comparison to pre-lockdown parameters in individuals with long-standing type 2 diabetes. methods: this prospective, observational study includes 2240 people with t2dm regularly attending diabetes clinic prior to lockdown. glycemic record, hba1c, and physical activity assessed with global physical activity questionnaire (gpaq) as metabolic equivalents (mets min/week) were obtained during lockdown (minimum duration of 3 months). results: a total of 422 out of 750 participants (nest) responded. the median (iqr) for age was 58 (52 to 64) years, duration of diabetes 11 (6 to 16) years, prevalent foot complications in 59.7%, and atherosclerotic cardiovascular disease in 21.3% of participants. there was a decrease in hba1c from 7.8% (6.9 to 9.4) prior lockdown to 7.4% (6.6 to8.7) during lockdown [δhba1c − 0.41 ± 0.27% (p = 0.005)] and postprandial blood glucose 200.0 mg/dl (152.0 to 252.0) to 158.0 (140.0 to 200.0) mg/dl (p < 0.001). the physical activity increased during lockdown from a gpaq score 140 (0.0 to 1260) mets to 840 (0.0 to 1680) mets (p = 0.014). the improvement of glycemic control was observed in either gender and independent of the presence of foot complications or increase in physical activity. conclusions: there is an overall improvement of glycemic control during covid-19 lockdown independent of increase in physical activity in people with long duration of diabetes. introduction covid-19 due to sars-cov-2 infection was declared as global pandemic by who on 11 march 2020. it was suggested that the transmission may be significantly curbed by limitation of outdoor activities through the imposition of strict lockdown [1] . subsequently, complete lockdown was enforced in india on march 25, 2020, until may 4, and partial lockdown is in place limiting daily activities at the time of writing the manuscript. a significant restriction of outdoor physical activity during lockdown may have perpetuating influence on lifestyle disorders including obesity, hypertension, and diabetes. sedentary lifestyles, poor dietary habits, and sleep deprivation are known potentially modifiable risk factors for poor glycemic control in people with diabetes. hence, lockdown during covid-19 pandemic may be associated with poor glycemic control in people with diabetes. however, there is no evidence set forth for this presumption except for the experiences from the past natural disasters which mimic the similar difficulties and limitations of daily activities [2, 3] . isolated studies in type 1 diabetes individuals have conflicting reports of worsening or no impact of lockdown period on glycemic control [4] [5] [6] . it is also evident that glycemic control may worsen due to the direct effect of sars-cov-2 infection in individuals affected, and people with diabetes are likely to have poorer outcomes from sars-cov2 infection [7] . therefore, we prospectively studied the effect of lockdown on physical activity and glycemic control in people with pre-existing type 2 diabetes mellitus. we invited 750 participants out of 2240 people with preexisting type 2 diabetes who were regularly attending diabetes clinic at pgimer, chandigarh, prior to covid-19 lockdown and had access to home-based capillary glucose monitoring by glucometer during the lockdown period. we have complete demographic and disease-related detail in electronic case record system. patients with type 1 diabetes, incomplete records, or not having facility for self-monitoring of blood glucose during the lockdown period or not accessible for telemedicine counselling or consultation and covid-positive patients were excluded from the study. their demographic characteristics, duration of diabetes, physical activity, microvascular and macrovascular complications, and glycemic parameters including hba1c were evaluated and entered in the electronic database at each follow-up visits. subsequently, the scheduled visits to the hospital were not possible due to lockdown; therefore, they were approached telephonically for consultation and guidance for titration of the medication doses including oral anti-diabetic drugs and/or insulin. they were requested to share glycemic records of fasting (fbg) and postprandial (1-2 h after major meal) blood glucose (ppbg) by home available glucometers and obtain hba1c at the nearest available laboratory facility after a minimum of 3-month duration of lockdown. the physical activity pattern of the participants during lockdown was enquired telephonically by global physical activity questionnaire (gpaq) that has been validated earlier in indian population [8, 9] and represented as metabolic equivalents (mets min/week). body weight prior to lockdown was obtained from the electronic repository, and weight during lockdown was recorded from the home-based weighing scales or at nearest available health facility. the primary outcome was change in hba1c, fbg, and ppbg compared to the last observed value before the lockdown in the electronic database. the other outcome measure was the change in gpaq scores (mets min/week). the evaluation for micro-and macrovascular complications was performed as per existing protocol of the institute that included annual (more frequently, if needed) fundus examination, neuropathy and vascular assessment, urine protein, creatinine (egfr) estimation, and fasting lipids. data analysis was performed using the statistical package of social sciences (spss) version 23 (ibmcorp, armonk, ny). normality was examined using the shapiro-wilk test. the data is expressed as the median and interquartile range (iqr) as most of the data variables were non-parametric. the pre-and post-lockdown glycemic variables were compared using wilcoxon signed-rank t test and fischer exact test/chi-square test for categorical variables. a sub-group analysis by stratifying data according to gender and the presence or absence of foot complications (active pedal ulcer or foot with deformities limiting physical activity) was performed. the correlation between change in glycemic control (hba1c) with the modification of weight, bmi, and physical activity (gpaq) during the lockdown period was analysed. the change in hba1c during the lockdown was considered as dependent variable with the change in fbg, ppbg, weight, bmi, and physical activity (gpaq) as independent variables. p < 0.05 was considered significant for the study. a total of 422 of the 750 individuals (303 male and 119 female) with diabetes responded with the requisite glycemic parameters within the stipulated duration. the median age of the participants was 58 (52 to 64) years, duration of diabetes of 11(6 to 16) years, and body mass index of 25.6 (22.7 to 28.7). prevalent microvascular complications include neuropathy in 58.3%, retinopathy in 30.1%, and nephropathy in 27.0% of participants (table 1) . foot complications were prevailing in 59.7% and atherosclerotic cardiovascular disease in 21.3% of participants. overall, 22.7% of participants are on insulin, and the rest are on oral anti-diabetic drugs (fig. 1 ). last observed hba1c before covid-19 lockdown is 7.8 (6.9 to 9.4%), and a hba1c of 7.4 (6.6 to 8.7) after 3 months of lockdown, with an overall hba1c reduction of 0.41 ± 0.27% (p = 0.005) ( table 2) . overall, 35.1% participants had hba1c < 7%, prior to lockdown as compared to 38. table 3 . participants with foot complications constituted more than half (59.7%) of the respondents. participants with foot complications had a higher baseline hba1c 7.9% (6.9 to 9.4) compared to those without foot complications 7.3% (6.6 to 8.3) (p = 0.180) with a decrease in hba1c of − 0.4% (− 1.7 to 0.9) and − 0.3% (− 1.0 to 0.5) (p = 0.341) in the two groups, respectively (table 4 ). we did not find significant correlation between change in glycemic control (δhba1c) with either age (p = 0.549), duration of diabetes (p = 0.416), change in weight (p = 0.597), or physical activity by gpaq scores (p = 0.128). we observed an overall improvement of glycemic parameter in people with long-standing type 2 diabetes associated with an increase in physical activity as assessed with gpaq score during the lockdown period unlike the conventional belief of worsening of glycemic control and limitation of physical activity. the decline in hba1c was independent of the increase in physical activity and was observed in either gender and irrespective of the presence or absence of diabetic foot complications. covid-19 pandemic has necessitated lockdown to limit the sars-cov2 infection and shown to be effective in reducing the r0, i.e. number of people infected by each infected person [10] . while lockdown slows the spread of infection, it is likely to have adverse influence on lifestyle patterns contributing to weight gain. a failure to adhere to lifestyle recommendations for diabetes during lockdown due to a significant curb of outdoor physical activity along with psychological stress related to pandemic may be associated with worsening of glycemic control. the stress of acquiring covid has also been ascribed as one of the reasons for poor glycemic control. a predictive modelling using a simulation model created with the aid of a multivariate regression analysis has shown that the predicted increment in hba1c from baseline at the end of 30 days and 45 days lockdown could be 2.26 and 3.68%, respectively [11] . however, this prediction was based on data from similar natural disasters but not exactly the same scenario as covid-19 lockdown and is likely to overestimate the risk because of modelbased risk prediction. a cross-sectional study in type 1diabetes individuals observed an increase in average blood glucose 276.9 ± 64.7 mg/dl as compared to 212.3 ± 57.9 mg/dl and hba1c of 10 ± 1.5% compared to 8.8 ± 1.3%) (p < 0.05) during and before lockdown, respectively [4] . the major reason attributed to worsening of glycemic control was the nonavailability of insulin in rural and semi-urban areas. we prospectively studied glycemic parameters in people with diabetes along and a change in their physical activity consequent to lockdown. unlike the belief, we observed an improvement in glycemic parameters compared to the last available pre-lockdown with a significant reduction in hba1c and postprandial blood glucose after a minimum of 3 months of lockdown. there was an increase in fasting blood glucose but an overall decrease in hba1c that was likely contributed by a considerable decrease in postprandial blood glucose during the lockdown phase. our results are consistent with recent studies predominantly in type 1 diabetes people that noticed no effect of lockdown on glycemic control [5, 6] . italian authors observed a decrease in time spent in hypoglycemia (time below range) during lockdown in insulin-treated people [5] . the possible reasons for better glycemic control in our study could be a decrease in work-related stress, adequate time for self-care, better compliance to medications, adherence to dietary recommendations (home cooked food), lack of availability of outside calorie-dense diet, and an increase in physical activity though indoors. though ghosh et al. observed an increase in carbohydrate consumption and snacking in people with type 2 diabetes from north india [12] , recurrent contact through teleconsultations may have helped in allaying fear and stress of acquiring covid in the present cohort. excessive sedentary behavior and lack of exercise are a problem area in management of diabetes due to lack of adherence which is likely to be further worsened by covid-19 pandemic. however, we observed that most of the respondents engaged themselves in physical activity doing household chores and indoor exercise consequent upon availability of time that was reflected in a significant increase in gpaq scores during the lockdown. all the respondents were motivated individuals having long duration of diabetes, attending diabetes clinic regularly, and were knowledgeable of lifestyle recommendations and glycemic targets. moreover, they were regularly counselled telephonically and encouraged to limit calorie intake and sedentary behavior during lockdown. it has been observed that unstructured physical activity like performing household chores is known to help in weight management, controlling postprandial hyperglycemia, and overall improved glycemic control by reducing the total sedentary time, increasing the energy expenditure that may [13, 14] . thus, despite a significant limitation of outdoor activities during lockdown, an increase in gpaq scores suggests that increasing indoor activities and limiting sedentary time are also beneficial for people with diabetes in improving glycemic control. our results also suggest that people with significant comorbidities of diabetes that limit outdoor activities like foot complications are also able to achieve good glycemic control. knowing that people with foot complications like neuropathic foot ulcers or charcot neuroarthropathy and foot deformities are likely to have higher mortality as compared to individuals with diabetes without foot complications [15, 16] , good glycemic control in this cohort is more desirable. the improvement in glycemic parameters associated with an increase in physical activity and weight loss was observed irrespective of gender. covid-19 is associated with significant psychosocial impact on people with type 2 diabetes related to concerns about worsening of glycemic control. however, improvement noticed in glycemic control in the present study will help to counsel the patients for better self-care during covid-19 pandemic [17] . this is the first large, prospective study amongst people with long-standing type 2 diabetes to assess the effect of more than 3 months duration of lockdown on glycemic control. however, certain potential biases cannot be ruled out in the present study including that all the respondents in our study were self-motivated, had long duration of diabetes (> 10 years), were under clinic follow-up for long duration, and aware of lifestyle recommendations and glycemic goals. moreover, only the motivated patients are likely to respond with glycemic parameters that might have contributed to most patients having improved glycemic control. during lockdown, gpaq survey was conducted telephonically; various kinds of glucometers were used for capillary glucose that might have an inherent bias. the reliability and reproducibility of the home-based weighing scales cannot be vouched, but it helped us in understanding the trend of weight change in real life pandemic situation. the dietary change, macronutrient composition, and calorie intake were not recorded. the results of our study may not be generalized to those with shorter duration of diabetes or with limited healthcare teleconsultation access. in conclusion, the present study assures that lockdown period may not be associated with worsening of glycemic control in people with long-standing diabetes. limiting sedentary time and increasing indoor activities also help in achieving better glycemic control during covid-19 lockdown. awareness of glycemic goals, access to self-monitoring of blood glucose, and ability to cope with restrictions of lockdown by rigorously following lifestyle recommendations and engagement in some form of physical activity are beneficial. the efficacy of lockdown against covid-19: a cross-country panel analysis. appl health econ health policy impact of a natural disaster on diabetes impact of psychological stress caused by the great east japan earthquake on glycemic control in patients with diabetes impact of lockdown in covid19 on glycemic control in patients with type 1 diabetes mellitus no deleterious effect of lockdown due to covid-19 pandemic on glycaemic control, measured by glucose monitoring, in adults with type 1 diabetes effects of covid-19 lockdown on glucose control: continuous glucose monitoring data from people with diabetes on intensive insulin therapy practical recommendations for the management of diabetes in patients with covid-19 validity of the global physical activity questionnaire (gpaq) in assessing levels and change in moderate-vigorous physical activity and sedentary behavior physical activity and inactivity pattern in india result from the icmr-indiab study (phase-1) how and when to end the covid-19 lockdown: an optimization approach. front public health estimation of effects of nationwide lockdown for containing coronavirus infection on worsening of glycosylated haemoglobin and increase in diabetes-related complications: a simulation model using multivariate regression analysis effects of nationwide lockdown during covid-19 epidemic on lifestyle and other medical issues of patients with type 2 diabetesin north india [published online ahead of print the role of free-living daily walking in human weight gain and obesity breaking up prolonged sitting reduces post-1prandial glucose and insulin responses long term outcomes after incident diabetic foot ulcer: multicenter large cohort prospective study (edi-focus investigators) epidemiology of diabetic foot complications study mortality in asian indians with charcot's neuroarthropathy: a nested cohort prospective study psychological adaptive difficulties and their management during covid-19 pandemic in people with diabetes mellitus publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations acknowledgments we thank miss raveena, mrs. kusum, and mrs. reshma for data collection. conflict of interest none.informed consent a written informed consent was obtained from all participants (signed digitally) and the study was approved by the institute ethics committee. key: cord-346664-ilebaqx3 authors: rahul; verma, alka; yadav, priyank; sharma, vijay kumar; sanjeev, om prakash title: non-covid surgical emergency during the nationwide lockdown due to corona pandemic: a critical appraisal date: 2020-08-10 journal: indian j surg doi: 10.1007/s12262-020-02549-5 sha: doc_id: 346664 cord_uid: ilebaqx3 the world health organization (who) declared corona infection as a pandemic in february 2020. a nationwide lockdown was enforced by indian government on 25 march 2020. separate health facilities were developed to handle the confirmed and suspected cases of covid-19 (coronavirus disease). other than emergency services and care of cancer patients, all remaining healthcare activities were curtailed. through this study, we intend to assess any change in number and pattern of non-covid surgical emergencies during the lockdown as well as the interventions required. this was an observational study which included all patients with surgical emergencies who presented during the study period (25 march to 24 april 2020) after two stage screenings for corona infection (group 2). the results obtained from analysis of prospectively collected database were compared with a similar period (group 1) prior to the onset of pandemic in india using appropriate statistical tests. in group 2, an increase (17%) in number of patients was noted. the need of organ support was more than 4 times the usual period. an upsurge in neurosurgical emergencies was noted, though the number of interventions decreased by 40%. a significant decrease in hospital stay was also documented (7 days vs 12 days). the nationwide lockdown led to an increase and change in pattern of surgical emergencies, though the interventions required were less. effective management entails appropriate preparedness. an outbreak of viral infection emerged in china in december 2019. a novel coronavirus named covid-19 (coronavirus disease) was isolated as the causative agent. despite containment measures taken by china and other countries, the case count soared so high that the disease was declared a pandemic by the world health organization (who) in february 2020 [1] . the first case in india was recorded on 30 january 2020 in the southern state of kerala. the virus spreads through direct contact, fomites, and droplets [2] . a slow but steady increase in the number of cases throughout the country was noted in the months of february and march. to prevent the exponential growth and a sudden outburst of patients, a nationwide lockdown was enforced by the indian government from 25 march 2020, when the case count in the country was 500. the primary intention was social distancing to put a check on the transmission and buy time for preparedness. as compared with the western countries, the early enforcement of lockdown in india diminished the initial rate of progression [3] .the recorded patients of coronavirus and mortality associated with the disease remained substantially low as compared with european and american countries. in order to ensure that the focus of the healthcare system remains undivided in the fight against coronavirus, except for emergency services and care of cancer patients, all remaining healthcare activities were curtailed during the lockdown. the outpatient department activities were put on hold (other than malignancy and end-stage organ disease). hospital admissions and elective surgeries were postponed to safeguard the manpower and resources. the essential emergency services remained functional. at many centers, a separate health facility was created within a span of 1-2 weeks to handle the confirmed and suspected cases of covid-19 [4] . the response to coronavirus outbreak in the province of uttar pradesh with a population of over 200 million was swift, and as the lockdown was enforced, a 3-tier triage system was introduced [4] . the trauma cases during the lockdown were expected to be low. the trauma center at our institute was converted to a corona hospital within 2 weeks. all non-covid emergency services continued to be delivered by the department of emergency medicine in the main hospital. during the lockdown period in india, the services to patients who were not at imminent risk took a serious toll [5] . the lack of standard guidelines and dearth of eloquent personal protection equipment for screening and management of patients in the first month impeded the healthcare services offered by the nursing homes and private practitioners, a trend that mimicked the one seen in italy earlier [6] . as a result, many patients who would not normally come to the emergency department at a government center were forced to do so. the increased load on emergency during a contagious pandemic conjures efficient screening system of covid suspects and triage to bestow maximum benefit to the patients. organizing separate traffics for the epidemic patients and non-covid casualties entails huge redistribution of manpower. in the period of crisis, it is prudent to judiciously utilize the limited resources. the available staff members need to be trained and reorganized to manage increasing number of concurrent emergencies. with limited number of operation theaters and available anesthesia support, it is pragmatic to learn about changes in number and type of surgical emergencies. this will aid in formulating effective plans. through this study, we intend to highlight the difference in patterns of patients who presented as a surgical emergency during the lockdown period (covid outbreak). the primary end points of the study were (a) to assess any change in number and pattern of patients with surgical emergency during the lockdown and (b) to assess the impact of lockdown on the duration of symptoms before reaching the emergency. the secondary endpoint of the study was to assess any change in the number of interventions. this was an observational study conducted in the department of emergency medicine at a tertiary care center in northern india. institute's ethical committee approval was obtained to conduct the study (iec code: 2020-134-ip-exp-20). the emergency department (ed) has 30 beds which cater to both surgical and medical emergencies. during the lockdown period, the patients were admitted to the department only after thorough screening with questionnaire (symptoms and travel history) and temperature probes for suspicion of corona. the suspected cases were evaluated at corona center in a holding area. patients who were not suspected to be positive for corona and hailed from a low transmission area were directly admitted at the ed in the main hospital. all patients were tested for corona before any radiological or surgical intervention or before being shifted to their respective wards. in this study, we intended to evaluate the patients with surgical emergencies visiting the ed (in the non-covid hospital) during the first month of national lockdown (25 march 2020 to 24 april 2020). patients who were brought dead were excluded. the data was collected retrospectively from the prospectively maintained hospital records. the management of the patients was not altered by the study. the data of patients during the lockdown period was compared with the profile and outcomes of patients referred to the respective departments over a similar period (1 month) prior to the onset of the pandemic in india. any change in the characteristics of patients and their management were recorded and compared using appropriate statistical tests. sample size estimation to detect the 0.8 effect size (≥ 0.8 effect size between two independent groups is considered large effect) between two independent groups, at minimum two-sided 95% confidence interval and 80% power of the study, calculated sample size of the two groups came out to be 26 each. in this study, we have included 29 patients in group 1 (pre lockdown) and 34 in group 2 (during lockdown). sample size was estimated using software g power version 3.1.9.2 (düsseldorf university, germany). statistical analysis normality of the continuous variables was tested, and a variable was considered normally distributed when z score of the skewness was ± 3.29.continuous variables were presented in mean ± standard deviation/median (interquartile range), whereas categorical data was presented in frequency (%). to compare the mean, median, and proportions between two groups, independent sample t test, mann-whitney u test, and chi-square test/fisher exact test were used respectively. error bar graph was used to present the distribution of means. a p value < 0.05 was considered statistically significant. statistical package for social sciences version-23 (spss-23, ibm, chicago, usa) was used for the analysis. the differences in the profile and management of the patients (surgical emergencies) who were managed in the ed of the non-covid hospital during the first month of the lockdown period and a similar period in the non-pandemic era (before january 30th in india) have been highlighted in tables 1 and 2 . there was a marginal increase (17%) in the total number of surgical emergencies during the lockdown period, and the average distance covered by the patients to reach the hospital was around 25 km more (13% more) than the pre-covid era, though the differences were not statistically significant. the duration of symptoms before reaching the hospital was on an average 6 days in group 2 against 5 days in group 1. though the most common complaint with which the patients were admitted in both the groups was pain in the abdomen, there was significant decrease in the number of patients with acute abdomen overall (82% vs 58% with significant decrease in group 2, p value 0.039). the type of abdominal emergencies was also different. in group 1, the majority of cases included perforation peritonitis, biliary peritonitis following cholecystectomy, necrotizing pancreatitis, liver abscess, or intestinal obstruction necessitating surgical/radiological intervention. in group 2, mild acute pancreatitis, biliary colic, and advanced malignancy with jaundice or ascites were the commonest abdominal emergency. most of them could be managed conservatively. two patients required surgery: one underwent nephrectomy for renal cell carcinoma with intractable hematuria, and another underwent percutaneous nephrostomy for obstructive uropathy. a significant increase in the number of patients with neurosurgical emergencies (most commonsubarachnoid hemorrhage) was noted. all patients had severe headache and altered sensorium. they were stabilized, resuscitated, preferably tested for covid, and taken up for intervention. two patients required craniotomy and three underwent coiling for aneurysm. no differences in groups 1 and 2 were recorded in terms of number of patients managed successfully and number of blood transfusion required or mortalities. the major difference noted was in the status of patients at arrival to the ed: more than 4-fold increase in requirement for organ support in group 2 (9 vs 2 in favor of group 2), though the difference was not statistically significant. six patients with neurosurgical issues required intubation with due precaution. an increase in number of radiological evaluation was noted in group 2. this was because all the patients referred to neurosurgery underwent cross-sectional imaging of the head. the major difference in groups 1 and 2 was the median hospital stay (12 days vs 7 days with significant decrease in group 2, p = 0.007; fig. 1 ). the covid-19 pandemic has made a true global footprint and has strained the healthcare facilities across the world including india [7] . knowledge regarding the presentation and possible management of the covid infection is improving constantly. new corona hospitals are being set up in every city. the hospital staffs including doctors, nursing personnel, and supporting crew managing these patients are being rotated frequently. the constantly rising number of cases has engaged more than one-third of the workforce in care of the infected. during the lockdown period, we observed a certain change in the patient load in ed at our center. a minor rise (17%) in the number of surgical emergencies was noted. the patients had to travel a longer distance (13% more) to avail the health facilities. the most important difference was increase (4-fold increase) in number of seriously ill patients requiring organ support. majority of patients with neurological complains required ventilatory support. the increase in requirement of vasopressor support among the patients may be due to various reasons. in pre-lockdown period, the patients were referred to our center by peripheral centers after providing primary care and initial resuscitation, but in the pandemic era, majority of the patients directly came to tertiary care centers. moreover, the average time to reach the hospital from the start of illness was greater during the lockdown. the health facilities extended by the peripheral private setups remained non-functional due to lack of personal protection equipment and the fear of pandemic in the first month of the lockdown. in group 2, majority of patients who arrived with abdominal complaints could be managed conservatively. the number of interventions needed was less (26% vs 65%). this was again because majority of the patients came with symptoms of biliary colic or mild pancreatitis which resolved on medications. in normal circumstances, they are usually treated at peripheral centers and need not travel distance to avail health facilities. further, the strict lockdown also affected the diet habits and social behavior due to isolation that could partially explain the reduction in the incidence of bowel obstruction, severe pancreatitis, and perforation. similar findings were documented by patriti et al. in their study [6] . in the pre-lockdown period, many neurosurgical emergencies would be operated in smaller hospitals across the city and the state. in the country, neurosurgery units run in trauma centers which take care of the non-traumatic neurosurgical emergencies as well. with the start of the pandemic, trauma centers in the city were converted into covid care facilities. this added to the load of neurosurgery cases in the ed which was very visible in the limited duration of the present study. they were managed successfully by the neurosurgery team. at our center, the patients treated in the ed were all low risk for covid infection admitted after thorough 2 stage screenings (detailed history and thermo scan). none of them tested positive for the virus during the stay. all the patients requiring intervention were first tested for covid antigen, and those requiring early ventilation were intubated with all due precautions. the overall emergency and hospital stay in group 2 was significantly less as the decision to shift to respective units and intervene were swiftly taken. the upsurge in medical and surgical emergencies necessitated prompt treatment. moreover, majority of the patients (with mild abdominal complains-biliary colic and mild pancreatitis) improved with conservative management over a short period. the substantial decrease (40%) in the requirement of intervention was responsible for reduced hospital stay. the present study points towards the likelihood of upsurge in emergencies presenting to the ed, especially the neurosurgical emergencies. with the ease in travel restrictions, the number of emergencies (including trauma) is set to rise, and this will add to the burden of ed. it is important to make periodic appraisal of the hospital services and the changing patient load in order to apprise the concerned authorities regarding the changing needs. the lacunae need to be highlighted and necessary adjustments made. this will also help to reactivate and coordinate with the local healthcare centers for mutual support and mitigate the undue load on a tertiary care center. the use of referral and counter-referral systems can help to efficiently distribute the work load, provide home care (with the help of local hospitals), and avoid unnecessary admissions in the hospital. hospital emergency management plan during the covid-19 epidemic feng z (2020) early transmission dynamics in global comparison of changes in the number of test-positive cases and deaths by coronavirus infection (covid-19) in the world ministry of health and family welfare government of india (2020) updated containment plan for large outbreaks novel coronavirus disease 2019 (covid-19) lockdown deals deadly blow to kidney patients, the hindu what happened to surgical emergencies in the era of covid-19 outbreak? considerations of surgeons working in an italian covid-19 red zone an interactive web-based dashboard to track covid-19 in real time publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations conflict of interest the authors declare that they have no conflict of interest.ethics approval this study was approved by the institute's ethics committee and patient consent was waived off (iec code: 2020-134-ip-exp-20). key: cord-263248-8y1u0h6y authors: ediev, d. m. title: population heterogeneity is a critical factor of the kinetics of the covid-19 epidemics date: 2020-06-26 journal: nan doi: 10.1101/2020.06.25.20140442 sha: doc_id: 263248 cord_uid: 8y1u0h6y the novel coronavirus pandemic generates extensive attention in political and scholarly domains. its potentially lasting prospects, economic and social consequences call for a better understanding of its nature. the widespread expectations of large portions of the population to be infected or vaccinated before containing the covid-19 epidemics rely on assuming a homogeneous population. in reality, people differ in the propensity to catch the infection and spread it further. here, we incorporate population heterogeneity into the kermack-mckendrick sir compartmental model and show the cost of the pandemic may be much lower than usually assumed. we also indicate the crucial role of correctly planning lockdown interventions. we found that an efficient lockdown strategy may reduce the cost of the epidemic to as low as several percents in a heterogeneous population. that level is comparable to prevalences found in serological surveys. we expect that our study will be followed by more extensive data-driven research on epidemiological dynamics in heterogeneous populations. because of the novelty and urgency of the situation, epidemiological models inform decision-making 1,7-13 in addressing the covid-19 pandemic. those models indicate high contagiousness of the virus and raise concerns about the majority of the population to be infected (if not vaccinated). the basic reproduction number ܴ of the pandemic at its beginning was estimated to be around 3 9,14-16 , which implies 1 െ 1 ܴ ⁄ , i.e., about 67 percent of the population must be infected or vaccinated before the infection may be controlled without lockdown measures. this conclusion has affected mitigation policies in many countries, it has also contributed to expectations of recurrent waves of the epidemic. those models, however, ignore varying social engagement, epidemic-awareness, and hygiene preparedness that, along with other factors, contribute to the varying propensity of contracting the disease and spreading it to others. various reports suggest 10-20 percent of cases may be responsible for 80 percent of the covid-19 transmissions [17] [18] [19] . these findings illuminate the fact that while the majority of people may barely contribute to the spread of the epidemics thanks to either limited social engagement or higher alertness and better hygiene, few others may become superspreaders infecting dozens of people. an essential practical conclusion from this conclusion was a call to aim the mitigation policies at superspreaders to reduce the basic reproduction number (average number of secondary infections per one initial infected person) and contain the spread of the infection. another aspect of the heterogeneity, however, may demand to revise that conclusion and readdress the prospects of the pandemic and mitigation policies. the population heterogeneity is an essential player in the kinetics of the epidemic, because when the minority who contributes most to the spread of the virus contracts the disease and develops immunity, the outbreak may abruptly come to an end before the expected majority gets affected. differential contagiousness also matters for how to manage the lockdown policies and whether to assume the recurrent waves of the epidemic after the lift of the social isolation measures or autumn cooling. furthermore, population heterogeneity may shorten the course of the outbreaks, because those with higher social engagement will also be the first to catch the infection. in figure is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) reproduction number of three, the epidemics could have been checked after 67 percent of the population gets infected. in reality, that threshold may be surpassed thanks to the gained momentum of the spread of the infection. in the heterogeneous cases, the total numbers infected are also substantial (27.6 and 14.1 percent, respectively) but much lower than in the homogeneous case. the peak levels of the infected population are also much higher (56.1 percent) in the homogeneous population than in the heterogeneous ones (12.1 and 5.8 percent). also note that the more heterogeneous is the population, the earlier is the peak of the epidemic. an intuition to this observation is that the faster infection (and recovery) of the superspreaders accelerates the epidemics in its early phase while slowing it down in a later phase. in panes b-d, we present results for three timing options for the lockdown that lasts over 28 days and reduces the spread of the virus by 90 percent. when started too early (day 30, pane b), the lockdown leaves too many people susceptible to the virus and facilitates a substantial second wave. the total infected population is nearly the same as in the no-lockdown variant for the homogeneous population but considerably lower for the heterogeneous populations. with a better timing of the lockdown, the long-term costs of the epidemics are much lower. the lockdown presented in pane c (starts in day 39) is optimal for the more heterogeneous population that experiences, with the optimal lockdown timing, no second wave (and the total number infected is minimal at 4.9 percent). that lockdown, however, is yet too early for the less heterogeneous population where a moderate second epidemic wave develops and leads to a total of 15.8 percent infected (a substantially higher cost as compared to the minimal cost of 11 percent associated with the lockdown starting in day 44). in the homogeneous case, the second wave is even higher, and almost everybody (93.3 percent) is, again, gets infected. only a later lockdown that starts in day 55 (pane d) produces the optimal result for the homogeneous population (68.8 percent infected). while the first wave of the epidemics in the heterogeneous cases is earlier and more compressed as compared to the homogeneous case, the second wave, on the contrary, is later and more stretched out. even if beneficial in terms of a lower peak, an extended small second wave may misguide the policymaker about the long-term efficiency of the lockdown measures. heterogeneous scenarios show much lower long-term costs of the epidemics and peak levels of the infected as compared to the traditional homogeneous case. if the lockdown had been more selective, better protecting the non-spreading population, those numbers could have been even smaller. indeed, the epidemics could, in principle, be contained after most of the superspreaders were infected (bringing total infected populations down to about 1. 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. (which was not certified by peer review) the copyright holder for this preprint this version posted june 26, 2020. . expectations that about 70 percent of people may be infected before containing the pandemic were implicitly based on assuming population homogeneity. contrary to those expectations, we show that the population heterogeneity may bring that threshold level down to as few as 14 percent with a similar basic reproduction number. population heterogeneity, it appears, may even outweigh the vaccination in its importance as a factor checking the spread of the disease. we urgently need to fully understand the extent and nature of how people differ in susceptibility to the infection and the ability to spread it and appreciate that in our decision-making. in the long run, a lower number of people infected means fewer causalities to the virus. in the short run, however, lockdown policies around the world take the capacity of the healthcare system into account too. in that context, it is notable that population . 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 june 26, 2020. . heterogeneity also reduces the peak levels of the infected population (from 56 percent, as in the homogeneous case, to 6-12 percent). lockdown, when well-scheduled, is capable of substantially reducing the cost of the outbreak. the timing of the lockdown is crucial in all scenarios. when prematurely implemented, the lockdown leaves too large a portion of the population susceptible to the infection, which results in the second wave of the epidemic. in such cases, the epidemic may gain momentum and eventually lead to nearly the same total number of infected persons as in the case of no lockdown. the second wave appears to stretch over a more extended period in the heterogeneous cases, which may misguide policymakers in their assessment of the efficiency of the lockdown. too late a lockdown, however, is also inefficient, because it allows for many avoidable infections. in the optimal lockdown strategy, one should wait until the proportions susceptible fall to levels where the instantaneous reproduction number turns unity. after reaching that threshold, the lockdown measures should be implemented with maximal possible strength to cool off the epidemic' momentum and halt the further spread of the virus. to design such an optimal policy response, however, it is mandatory to understand the kinetics of the epidemic well and assess the threshold correctly. that includes accounting for the role of population heterogeneity. designing policy responses to the covid-19 and other epidemics. our results imply that we should further extend those models to include different predispositions to catch and spread the infection. with optimal lockdown strategy, the total number of infected people may be reduced to as low as five percent in the heterogeneous population. notably, such level of prevalence is of the same magnitude as was found in serological surveys 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 june 26, 2020. . https://doi.org/10.1101/2020.06.25.20140442 doi: medrxiv preprint will also be less affected by lockdown measures. that may contribute to the selectivity of the first type and reduction of the eventual cost of the epidemic. on the other hand, closing down public workplaces, introducing strict social distancing, isolation, or public hygiene measures may affect more the superspreaders while having limited effect, if at all, on socially less engaged and initially more prepared people. that may form lockdown selectivity of the second type and exaggerate the cost of the epidemic. which scenarios develop in reality needs urgently being examined while countries move into the postlockdown phases. long-term effects of the population heterogeneity reported here also call for revisiting the policy recommendations with respect to the superspreaders. the usual policy recommendation with respect to the superspreaders is to maximize lockdown efficiency among the superspreaders. yet, we indicate that such a policy may delay but not prevent the second wave of the epidemic and spread, unnecessarily, the infection more into the non-spreaders population. we need to address this issue in designing social isolation policies. the extent and type of population heterogeneity depend on many factors that need to be studied. those include demographic factors such as age and sex, kinship structures and relations, household sizes and roles within them. factors of heterogeneity also include biological predispositions, behavioral patterns (that, in turn, may depend on demographic circumstances, such as the presence of persons vulnerable to the disease in the household or kinship networks), educational, occupational, and income differentials, and others. a better understanding of these relations is instrumental in combating both the current urgency and other communicable diseases. to address those issues, however, we need representative and comparable statistics on how we differ in odds to catch and then to spread the virus. such data are barely available, and we call statistical and healthcare agencies to urgently fill the gap in data on population heterogeneity. . 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 june 26, 2020. . we incorporate population heterogeneity into the discrete version of the kermack we assume a symmetric model where both the propensity of catching the virus and the propensity of spreading it are proportional to the communicability parameter ݇ . hence, we model new infections as follows: in modeling the course of recovering, we trace the duration of the infection period for the infected people and assume every infected person to recover in time ߬ after getting infected. that is, the number of people recovering in the period ‫ݐ‬ equals: . 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 june 26, 2020. we neglect r-s transitions from recovered to the susceptible population because such transitions have not yet been reported to play a substantial role in the covid-19 epidemics. we also neglect the fatality of the disease because we intend to highlight the primary effects of population heterogeneity upon the overall course of the epidemic. introducing r-s transitions, mortality, and more realistic demographics should pose no difficulty in future research. assuming the entire original population is susceptible, the number of secondary infections per one initially infected person of type ݆ over the communicability period ߬ may be found from (3) here, 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 26, 2020. . https://doi.org/10.1101/2020.06.25.20140442 doi: medrxiv preprint that halt the new infections completely. eqs. (5) and (10) lead to the following closedform solution for the evolution of the susceptible population of type the higher the communicability parameter ݇ , the faster is the fall of the susceptible population in (11) . that creates compositional change in the remaining susceptible population, a change that suppresses the communicability-weighted susceptible population ‫ܬ‬ ሺ ‫ݐ‬ ሻ and checks the spread of the epidemics. in generating and interpreting results of simulation scenarios, it is useful to relate the model parameters to the commonly used basic reproduction number ܴ . to establish the relation, assume the initial distribution of infected people follows the model relation (3) and is proportional to the weighted populations of each type: that is, the basic reproduction number is the weighted average of the communicability parameter with weights equal to the weighted susceptible populations of each type. at an advanced phase of the epidemic, substantial portions of the population move to infected or recovered compartments, and the instantaneous reproduction number of a new outbreak decreases to: , a new outbreak may be contained without a lockdown. subpopulation is set at such a level that the population-average basic reproduction . 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 june 26, 2020. . https://doi.org/10.1101/2020.06.25.20140442 doi: medrxiv preprint number (12) equals 3. we form the heterogeneous populations in such a way that 10 or 20 percent of infected people are responsible for 80 percent of the further spread of the epidemic, similar to what was reported in the literature [17] [18] [19] . assuming that ‫ݔ‬ percent of infected are responsible for ‫ݕ‬ percent of transmissions, the communicability parameter of non-spreaders (݇ ଵ ) and superspreaders may be found from (12) as: , in which case our model turns to the conventional sir model. in fig. s1 , we present sizes of the three population compartments in four selected simulations for the homogeneous population: no lockdown intervention (pane (a)); lockdown reducing the . 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. similarly timed (pane (c)) leads to near-optimal results. indeed, such an 'optimality' of the lockdown ignores infection fatality and healthcare systems' capacity that has become a concern in many countries. in fig. s2 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 26, 2020. . table s1 . even with only half of the people being non-spreaders, and with no lockdown, the long-term cost of the epidemic and the peak number of the infected people decrease by more than 30 percent as compared to the homogeneous population. in an extreme case where 99.9 percent of people are non-spreaders, the long-term cost of the epidemic is only about five percent without any policy intervention. the peak level of the infected population also falls dramatically as the population heterogeneity increases. . 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 june 26, 2020. . (pane (c)), and the optimal lockdown reducing the spread by 99.1 percent that starts at ‫ݐ‬ ൌ 5 7 (pane (d)). all lockdowns last for 14 days. vertical axis: population size starting with 1000 original population. horizontal axis: time in days from the original infection of 0.01 percent of people. is the strength parameter of the lockdown; ‫ܫ‬ is the eventual proportion of the population infected throughout the course of the epidemic. . 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 june 26, 2020. . https://doi.org/10.1101/2020.06.25.20140442 doi: medrxiv preprint . 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 june 26, 2020. . https://doi.org/10.1101/2020.06.25.20140442 doi: medrxiv preprint ) 4. elife. covid-19: a collection of articles a contribution to the mathematical theory of epidemics lessons from a rapid systematic review of early sars-cov-2 serosurveys experts' request to the spanish government: move spain towards complete lockdown first-wave covid-19 transmissibility and severity in china outside hubei after control measures, and second-wave scenario planning: a modelling impact assessment how will country-based mitigation measures influence the course of the covid-19 epidemic? the incubation period of coronavirus disease 2019 (covid-19) from publicly reported confirmed cases: estimation and application social network-based distancing strategies to flatten the covid 19 curve in a post-lockdown world estimation of country-level basic reproductive ratios for novel coronavirus (covid-19) using synthetic contact matrices health and sustainability in post-pandemic economic policies early transmission dynamics in wuhan, china, of novel coronavirus-infected pneumonia epidemiological and clinical features of the 2019 novel coronavirus outbreak in china the reproductive number of covid-19 is higher compared to sars coronavirus clustering and superspreading potential of severe acute respiratory syndrome coronavirus 2 (sars-cov-2) infections in hong kong estimating the overdispersion in covid-19 transmission using outbreak sizes outside china epidemiology and transmission of covid-19 in 391 cases and 1286 of their close contacts in shenzhen, china: a retrospective cohort study #propoirtion of non-spreaders; =0 for the homogeneous case k1<-0.5 #communicability parameter for the non-spreaders k2<-r0*(1+sqrt(1+4*beta*k1*(r0-k1)/(r0^2*(1-beta))))/2 #communicability parameter for the superspreaders k<-c(k1,k2) #communicability parameters #ditribution of the population *tau))-1 #period of the lockdown qq<-0.1 #lockdown efficiency coef. 1 -no lockdown; 0 -absolute halt of epidemic q<-ifelse(horizon %in% quaranteen,qq,1) #generate scenario fun=sum) plot(pt.agr$t,pt.agr$s,type="l",col="darkgreen r if(min(q)<1) rect(min(horizon[q!=1]),-10,max(horizon[q!=1]), 1100,density=30,col="grey #function generating the epidemic scenario for the given set of parameters covgen<-function ncol=tau) p.k<-p*distr n0<-sum(k*p.k) alpha<-step/(tau*n0) k/n0) if(sum(i.k)<i0) { k-i.k i.k<-i.k+(i0-sum(i.k))*p 1:tau]<-i.k/tau i.k<-rowsums(i.matr) effective) for (t in horizon) { q<-q[horizon==t] #lockdown parameter: the lockdown reduces alfa q-fold s.prev<-pkt new i.k<-rowsums(i.matr) s<-s.prev-i.k.new r<-r.prev+i.rec i<-sum(i.k) j<-sum(k*i.k) n<-sum(k*s) r0.effective< the epidemic is minimal ‫ܫ(‬ ൌ 9 . 7 percent) when the lockdown is both optimally timed, and strict enough (pane (d)).an even more heterogeneous case where 10 percent of infected people spread the virus in 80 percent of the secondary cases is presented in fig key: cord-354475-im2py2or authors: ioanna, giannopoulou; vasiliki, efstathiou; georgia, triantafillou; panajota, korkoliakou; athanasios, douzenis title: adding stress to the stressed: senior high school students’ mental health amidst the covid-19 nationwide lockdown in greece. date: 2020-11-06 journal: psychiatry res doi: 10.1016/j.psychres.2020.113560 sha: doc_id: 354475 cord_uid: im2py2or the two-year preparation for the national university entrance exams in greece is one of the most trying periods in a young person's life, physically and emotionally. the present study reports the results from 442 last year senior high school students who completed an online survey (16-30 april 2020) concerning the lockdown impact on their mental health. overall, the rate of positive screen for depression (phq-9 score ≥ 11) significantly increased from 48.5% to 63.8% and of those scoring within severe depression range (phq-9 ≥20) from 10% to 27%; for anxiety (gad-7 score ≥ 11) increased from 23.8% to 49.5% and of those scoring within severe anxiety range (gad-7≥17) from 3.8% to 20.5%. after taking sex and baseline (one month prior to the lockdown) levels of depression and anxiety into account, the level of lockdown experienced distress was predictive of depression and anxiety levels in time of home confinement, accounting for about 30% of variance in symptoms severity scores. although our results may be subject to sampling and recall bias, the unexpectedly high rates of anxiety and depression warrant an urgent call to action aiming at mitigating and managing mental health risks of senior high school students in future waves of pandemic. on march 11th, who declared the disease caused by covid-19 a pandemic due to its widespread and rapid rate of transmission. the first covid-19 case was diagnosed in greece on february 26th. on march 10th, with officially 89 confirmed covid-19 cases and 0 deaths, the greek government implemented national school and university closures and following this, new measures were gradually introduced, every 2-3 days, to mitigate the risk of exponential virus transmission. on march 23rd with 695 confirmed cases and 17 deaths, nationwide lockdown (i.e. restriction of movement, whereby citizens could leave their house only for specific reasons and with a special permit) was enforced as a key emergency public health measure (giannopoulou & tsobanoglou 2020) . studies have shown that prolonged school closure and home confinement during a disease outbreak might have negative effects on children's physical and mental health (for review guessoum et al., 2020) but no studies to date have reported on mental health of final year high school students facing university entrance exams. successful admission to a higher education institution in greece is determined through the combination of: (a) the candidate's score on the national higher education entrance exams, known as panhellenic exams, administered by the ministry of education, (b) the candidate's choices (by order of preference) of desired degree courses and academic institutions, and (c) the number of places available in each academic department. greek society places a high value on tertiary education, and psychological pressure for success in university entrance exams is great given that it is perceived as having major consequences for determining the young person's future life and career prospects (oecd, 2018) . the two-year preparation for the panhellenic university entrance exams is one of the most trying periods in a young person's life, physically and emotionally. almost every student is overburdened by a considerable workload and long hours of study due to enrollment in two parallel educational systems -regular school and private coursework delivered by a cram school, so called "frontistirio" and/or individual tutoring (giannopoulou, 2019; liodaki & liodakis, 2016) . the pressure to score high in these exams in combination with the greek society's view that every student should obtain a university degree makes the panhellenic exams a highly stressful and often traumatic event for many greek youngsters, and particularly for those who wish to be admitted to a highly ranked academic institution (e.g. medicine, law, electrical & computer engineering). the money invested by parents in their child's extra private tuition adds more financial strain to the family, raises parental expectations and puts some young people under enormous stress to succeed. this year, students were faced with abrupt coronavirus-related disruption adjustment in their daily life and new challenges, including the adaptation to the sudden switch to online learning, combined with uncertainty about the timing of the exams, the conditions under which these will take place, the exact material they will be tested on, and the date of any return to school. the aim of the present study was to examine the impact of the nationwide lockdown on the levels of anxiety and depression among last year senior high school students preparing for the upcoming panhellenic university entrance exams. we expected to find an increase in anxiety and depression scores from one month prior to the lockdown to past two weeks during the lockdown. we anticipated that the level of lockdown experienced distress impinging on students' preparation for national exams would be significant predictors of anxiety and depression levels during the lockdown. a brief survey was constructed and made available in electronic form, using a secure web-based application redcap (research electronic data capture) hosted at the national and kapodistrian university of athens. self-selection sampling procedure was followed; the study with its link was publicized, through social media and websites of educational news, but also through sending an email invitation to students from cram schools (frontistiria). students were informed about the goals of the study, its rationale, and the process by which data would be collected, and about confidentiality and anonymity of their voluntary participation. for those under 18 years of age informed parental consent was sought, in addition to the young person's consent. young people were given an option to contact our service through e-mail or telephone if they felt they needed guidance/psychological support. the study was approved by the ethics committee of the attikon university general hospital. data was collected over two weeks from april 16 to april 30. a total of 459 students completed the survey; 4 students who reported that they will not take part in the panhellenic exams and 13 students who indicated that they were sitting the exams for the second time were excluded from the data analysis, as the numbers were too low to allow for separate groups analysis. only one student requested psychological help via sending an e-mail to our service. the questions were presented in four sections: (a) demographic characteristics, (b) anxiety and depressive symptoms over the first two weeks of february (prior to nationwide lockdown -baseline); (c) anxiety and depressive symptoms during the past two weeks (during lockdown -current), using the greek versions of gad-7 and phq-9 (the patient health questionnaire screeners, www.phqscreeners.com); (d) distress experienced during lockdown. anxiety was measured using the 7-item generalized anxiety disorder scale (gad-7), which has been validated in adolescent populations (mossman et al., 2017; tiirikainen et al., 2019) . the respondents were asked to indicate on a 4-point likerttype scale (0= not at all, 1 = several days, 2= over half of the days, 3=nearly every day) the frequency of each anxiety symptom over two weeks the scores on these seven items were summed to generate the total anxiety symptoms severity score (range 0-21). the internal consistency of the gad-7 in the present study was good in the total sample (α = 0.81 for the baseline -prior to the lockdown), and α = 0.88 for the current -during the lockdown) measure, with cronbach's alpha ranging from 0.82 to 0.91 in the sex subgroups. consistent with research literature we used a gad-7 score of 11 or greater as indicating a positive screen for anxiety. in addition, we applied the cut-off score of 17 as indicating severe anxiety, which coincided with ≥96th percentile at baseline. this is a 9-item self-report questionnaire assessing depression symptoms over the past two weeks; the phq-9-modified includes minimal adjustments to the original phq-9 (spitzer, williams, & kroenke, 1999) to incorporate characteristics of depression in adolescents and age-appropriate language. specifically, the phq-9-modified includes irritability in the item assessing depressed mood, and weight loss in the item assessing appetite. throughout this paper the abbreviation "phq-9" is used to indicate the phq-9-modified. the phq-9 has been validated among adolescents across various cultures and settings (adewuya et al., 2006; burdzovic andreas and brunborg, 2017; fatiregun and kumapayi, 2014; richardson et al., 2010; tsai et al., 2014) . each item is rated on a 4-point scale ranging from 0 ("not at all") to 3 ("nearly every day"). in our study, cronbach's alphas were 0.86 prior to the lockdown and 0.90 during the lockdown; 0.86 and 0.91 respectively for male sex, and 0.85 and 0.89 respectively for female sex. consistent with research literature we used a phq-9 score of 11 or greater as indicating a positive screen for depression. in addition, we applied the cut-off score of 20 as indicating severe depression, which coincided with ≥90th percentile at baseline. for the purpose of the study we developed a brief measure of distress experienced during lockdown, impinging on the students' preparation and revision for the exams. respondents were asked to rate on a 5-point scale (1=not at all, 2= slightly, 3=moderately, 4=very, 5= extremely) how often, since the nationwide lockdown, they had been experiencing heightened anxiety/insecurity about the panhellenic exams, feeling loneliness, feeling boredom, feeling abandoned by their teachers, feeling estranged from their friends/school peers, feeling worried about future, being troubled by negative thoughts/images, experiencing difficulty to feel close to others, experiencing a fear of death. the scores on these nine items were summed to generate the total lockdown-experienced distress score, which ranges from 9 to 45. the cronbach's alpha of this measure for the entire sample was 0.82, indicating good internal consistency. descriptive statistics were calculated and presented as mean and standard deviation (m, sd) for continuous measures and absolute and relative (%) frequencies for categorical variables. simple comparisons between groups were performed using independent samples t tests for quantitative variables and the chi-squared test for categorical variables. a two-way mixed analysis of variance (anova) was performed to assess the within-person changes, i.e. prior to the lockdown (baseline) versus past two weeks (current) and the main effect of sex (male vs female) on the levels of anxiety and depression symptoms. in addition, mcnemar test was used to examine within individual changes in symptom severity according to measures' cut off points. two-stage multiple regression analyses were carried out to identify significant predictors of self-report gad-7 and phq-9 current scores (dependent variables). independent predictor variables included in the analysis were sex, baseline depression symptoms severity and baseline anxiety symptoms severity (stage one) and lockdown distress measure (stage two); interactions were tested and found non-significant. prior to conducting the multiple regressions, the relevant assumptions of this statistical analysis were tested, i.e. correlations between independent variables, the collinearity statistics (i.e., tolerance and vif), residual and scatter plots, indicating whether the assumptions of normality, linearity and homoscedasticity were all satisfied. all the analysis was done at 95% level of confidence using spss software version 24. a two-way mixed anova revealed a significant increase in gad-7 and phq-9 severity scores from one month prior to the lockdown to during the lockdown period, with girls scoring significantly higher than boys at both points in time (see table1) . no significant interaction was found for either measure, f (1,436) = 0.918, p=0.339 for gad-7 and f (1,433) =0.058, p=0.809. table 2 shows the frequency of positive screen for anxiety and depression, using cut off scores. the proportion of all respondents who screened positive for anxiety (gad-7 ≥ 11) increased from 28.3% before the pandemic to 49.5% for the time of home confinement (mcnemar test p <0.0001) and of those scoring within the severe anxiety range (gad-7≥17) increased from 3.8% to 20.5% (mcnemar test p <0.0001). the proportion of all respondents who scored above the phq-9 cut off 11 or greater indicating positive screen for depression increased from 48.5% before the pandemic to 63.8% for the time of home confinement (mcnemar test p <0.0001) and of those scoring within the severe depression range (phq-9 ≥20) increased from 10% to 27% (p<0.001). the proportion of respondents who reported having thoughts that they would be better off dead, or of hurting themselves in some way increased from 25.9% before the pandemic to 29.7% during the lockdown period (mcnemar test p <0.05). more specifically, the proportion of those who reported having these thoughts nearly every day increased from 6% before the pandemic to 11.1% during the lockdown. a two-stage multiple regression was conducted with depression symptoms severity score as the dependent variable. sex, baseline depression symptoms severity score and baseline anxiety symptoms severity score were entered at stage one of regression and lockdown experienced distress measure was entered at stage two (see table 3 ). in relation to anxiety symptoms, the multiple regression revealed that at stage one, female sex, baseline depression symptoms severity score and baseline anxiety symptoms severity score contributed significantly to the regression model, f (3,434) = 54.917, p<0.001, and accounted for 27.2% of the variation in gad-7 scores during the lockdown period. introducing the lockdown experienced distress score explained an additional 29.6% of variation in gad-7 scores and this change in to the best of our knowledge this is the first study reporting data concerning the impact of the nationwide lockdown on the mental health of last year senior high school students preparing for the national university entrance exams in greece. the results indicate a substantial and worrying increase in anxiety and depression symptoms from before the covid-19 outbreak to the period of nationwide lockdown. the high rate of 48.5% of positive screen for depression before the pandemic found in our sample is unprecedented in greece and may reflect heightened exam-related academic pressure linked to uncertainty about securing a university placement, and pressure relating to family and social expectations, as the exams approach. previous field studies in greece that have looked at the epidemiology of depressive symptoms among senior high school students reported prevalence rates of depression ranging from 17.4% (magklara et al, 2015) to 26.2% (lazaratou et al., 2010) ; however, neither study has provided prevalence figures for adolescents facing university entrance exams. research in countries with similar educational structures (e.g. korea, turkey) has reported comparable depression rates in this population group (yildirim et al., 2007; lee & larson, 2000) . the rates of positive screen for depression and anxiety during the lockdown increased significantly to 64% and 50% respectively and were similar to those reported for the senior grade three students during the covid-19 outbreak in china, i.e. 60% for depression and 53% for anxiety (zhou et al., 2020) . the prevalence rates of severe depressive symptoms (phq-9 score ≥ 20) and severe anxiety symptoms (gad-7 score ≥ 17) during nationwide lockdown are staggering, as these increased from 10% before the pandemic to 27% for depression and from 3.8% to 20.5% for anxiety. the alarming rates of self-harm/suicidal thoughts merit attention, considering that the proportion of the participants who reported during the past two weeks having these thoughts increased from 26% before the pandemic to almost 30% during the lockdown period; and from 6% to 11.1% respectively with regards to having these thoughts nearly every day. the latter findings may be comparable with those reported in a recent study of university students' mental health during the nationwide lockdown in greece, whereby 9.7% of the sample reported currently thinking of committing suicide and doing some specific plans on how to do it, amounting to an almost 8-fold increase in suicidal thoughts (kaparounaki et al., 2020) . as expected, higher levels of depressive and anxiety symptoms prior to pandemic were associated with a further increase in symptomatology, but the impact of distress experienced during the lockdown was found to be the most robust predictor for a poorer mental health. after taking sex, baseline (one-month prior to pandemic) levels of depression and anxiety into account, the level of distress experienced during the lockdown was predictive of depression and anxiety levels in time of home confinement, accounting for about 30% of variance in symptoms severity scores. the findings of this study should be interpreted considering several important limitations in its design. first, the data was collected via online survey with social media and websites of educational news as the primary avenue for promotion. as such, random sampling did not occur which may have introduced sampling bias into the survey. the number of young people who saw the survey and chose not to participate could not be determined; however, it is plausible that those with a pre-existing interest in mental health and/or experiencing psychological distress would be more likely to respond to the survey, which may have inflated the rates of depression and anxiety. however, this limitation applies to most of the mental health surveys of covid-19, which have mostly used web-based convenience samples (pierce et al., 2020) . second, the measures, i.e. gad-7 and phq-9, have not been validated in greek adolescents' population. however, both demonstrated high internal consistency in the entire sample and in both sexes. third, both self-reported and retrospective rating of symptoms, prior and during the lockdown, are subject to recall bias. fourth, the estimates of anxiety and depression based on cut-off values on the self-report questionnaires may not be consistent with those based on clinical structured interviews. fifth, the use of the lockdown experienced distress measure developed for the purpose of the present study, despite its good internal consistency, being a non-validated previously measure may compromise the quality of the results. however, the within subjects' comparison of prior to (baseline) and during (current) the lockdown gad-7 and phq-9 scores supports our conclusion that the observed significant increase in depressive and anxiety symptoms was lockdownrelated. moreover, 30% of the variability in depression and anxiety scores in time of home confinement was explained by distress experienced during the lockdown. sixth, important predisposing and mediating factors, such as stressful and traumatic experiences prior to the pandemic, family/parental stress (e.g. related to financial difficulty, living conditions, social and interpersonal changes, family functioning) considered important predictors of psychological and mental health well-being were regrettably not assessed. although our results may be subject to sampling and recall bias, the unexpectedly high rates of anxiety, depression and self/harm suicidal thoughts warrant an urgent call to action. identifying and monitoring young people with such vulnerabilities, especially in time of preparation for the exams, is of paramount importance. to mitigate the mental health impact of potential future restrictions measures in response to the ongoing pandemic, the ministry of education, the community primary health services, the educators, and parents need to be aware of the downsides of the home confinement situation in order to address more effectively the issues related to added stress experienced by already stressed young people preparing for highly competitive university entrance exams. the covid-19 outbreak in greece may be an opportunity to look for new models to manage more effective education and to tackle the many flaws underpinning the system of the panhellenic university entrance exams. future studies should employ qualitative mixed methods design to explore the lived lockdown-related experiences of the students, their parents and their educators in the context of preparation for highly competitive exams. this approach will provide an important insight into the interplay between risk factors and buffers affecting young peoples' mental health outcomes and will inform interventions aiming at mitigating and managing mental health risks in future waves of pandemic. authors have no conflicts of interest to declare. i.g., g.t. and p.k. designed the study. v.e. designed and built the online data base for data collection. i.g. and v.e. analyzed the data. i.g. and v.e. drafted the manuscript, and g.t. and a.d. provided critical revisions. all authors approved the final manuscript for submission. validity of the patient health questionnaire (phq-9) as a screening tool for depression amongst nigerian university students depressive symptomatology among norwegian adolescent boys and girls: the patient health questionnaire-9 (phq-9) psychometric properties and correlates prevalence and correlates of depressive symptoms among in-school adolescents in a rural district in southwest nigeria the school: α place of challenge and psychosocial adaptation for pupils covid-19 pandemic: challenges and opportunities for the greek health care system adolescent psychiatric disorders during the covid-19 pandemic and lockdown the patient health questionnaire for adolescents: validation of an instrument for the assessment of mental disorders among adolescent primary care patients university students' mental health amidst the covid-19 quarantine in greece depressive symptomatology in high school students: the role of age, gender and academic pressure the korean 'examination hell': long hours of studying, distress, and depression some effects of the economic crisis on shadow education in greece depression in late adolescence: a cross-sectional study in senior high schools in greece the generalized anxiety disorder 7-item scale in adolescents with generalized anxiety disorder: signal detection and validation education for a bright future in greece, reviews of national policies for education says who? the significance of sampling in mental health surveys during covid-19 evaluation of the patient health questionnaire-9 item for detecting major depression among adolescents validation and utility of a self-report version of prime-md: the phq primary care study a brief measure for assessing generalized anxiety disorder: the gad-7 psychometric properties of the 7-item generalized anxiety disorder scale (gad-7) in a large representative sample of finnish adolescents patient health questionnaire for school-based depression screening among chinese adolescents the patient health questionnaire (phq) screeners. greek version of phq-9 the patient health questionnaire (phq) screeners. greek version of general anxiety disorder-7 (gad-7) high rates of depressive symptoms among senior high school students preparing for national university entrance examination in turkey prevalence and socio-demographic correlates of psychological health problems in chinese adolescents during the outbreak of covid-19 key: cord-342622-jqmz0mkz authors: edomah, norbert; ndulue, gogo title: energy transition in a lockdown: an analysis of the impact of covid-19 on changes in electricity demand in lagos nigeria date: 2020-07-25 journal: glob transit doi: 10.1016/j.glt.2020.07.002 sha: doc_id: 342622 cord_uid: jqmz0mkz in this study, we analyse the role of forced lockdowns on electricity consumption behaviour and its effect on momentary transition in electricity use. electricity consumption data for residential, commercial and industrial consumers within the lagos metropolis representing 259 electrical feeder locations were collected and analysed under three scenarios: first, we analyse a business-as-usual scenario without a lockdown; secondly, we analyse the case of a partial lockdown; and finally, we analyse the case of a total lockdown. the study revealed that aside government announcement of the lockdown, certain social practices triggered changes in electricity consumption and use leading to momentary energy transition. within the residential sector, increased cooking, home laundry, showering, and some professional practices that moved to the homes impacted on higher electricity consumption. reduced manufacturing practices limited to those involved in food, personal care and pharmaceutical products led to a reduction in electricity use within the industrial sector, while reduced electricity use in the commercial sector was triggered mainly by a scaling down of trading services to essentials. the study concludes by highlighting the impact of changes in electricity demand and consumption under these scenarios and its implications for energy transition and electricity planning. the coronavirus disease 2019 (covid19) is an acute respiratory disease caused by another novel coronavirus (sars-cov-2, previously known as 2019-ncov) [1] . the virus has spread worldwide leading to a global pandemic [2] . as of april 7, 2020, over 1.3 million people have been infected with the virus, with over 75,000 deaths recorded [3, 4] . the virus has grounded economic activities globally. livelihoods have been disrupted, economies affected and health facilities stretched globally. indeed, this pandemic has revealed some salient lapses and non-resilience in the design of our economic and social systems [5] . arguably, the coronavirus pandemic has revealed weak socio-economic systems resilience to economic shocks in several countries and across several regions of the world. there are increasing cases of job loses globally [6] . consequently, the global pandemic has forced organizations to seek different ways of working and continuing various forms of economic activities which has changed the way people use energy in a considerable way. the pandemic have seen a sharp increase in residential energy consumption across several regions of the world owing to the global shift to remote working and learning from home. most educational institutions have shifted teaching and learning to online distance learning modes while some non-essential and critical services are now done remotely from home. practices play a vital role in energy use (in general) and in electricity use (in particular) [7, 8] . theories of practices (or social practice theory) help us to understand how the things we do, and the way we go about doing them, impact on energy and electricity use [9] . shove argues that the quest for increased levels of cleanliness, comfort and convenience are the foundations that underpin increased energy use in the fulfilment of different practices [10] . social practices such as commuting, showering, cooking, etc., have become energy intensive over time [11, 12] . professional practices that manifest itself in various diverse forms such as trade and different forms of professional practices have also evolved over time [13] . indeed, policy making, and all its associated services, is also a practice [14] . the international energy agency (iea) argues that the residential sector accounts for over 25% of global electricity consumption [15] . arguably, with the covid-19 pandemic, most commercial and professional services are shifting to the homes. while this shift impact on lower electricity consumption in commercial and public services sector, a shift in the location of such services to residences may have impacted on a surge in residential electricity use for various energy services. what impact does this shift have (on-theground) on electricity demand? what are the implications of this shift and consequent increase in residential electricity consumption for energy security and safety? in what ways are the effects of the lockdown impacting on energy transition and electricity use? in this paper, we examine and analyse the impact of covid-19 on changes in energy (electricity) demand using lagos nigeria as a case study [16] . in this paper, we provide some methodological considerations, materials and methods in section 2. in section 3, we present the important findings. the implications of the changes in electricity demand owing to the covid-19 pandemic and what it means for an eventual transition in energy systems and use, and a transition towards a more sustainable and a less energy intensive consumption lifestyle are presented in section 4. in this study, we adopted the use of exploratory research tools to examine various scenarios of electricity demand and use in the residential, commercial and industrial sectors of a large segment of the lagos metropolis. this study was conducted in lagos nigeria. lagos is a fast growing megacity located in south-west nigeria with a population of about 20 million people. it was once the capital city of nigeria until 1991 and it still remains the commercial capital of nigeria [17] . lagos, as a city, records the highest electricity demand and consumption in nigeria with over 800 mw of the 4000 mw average national electricity generation output [18, 19] . the city is served by two of the eleven electricity distribution companies in nigeria which are the eko distribution company (ekedc) and the ikeja distribution company (ikedc). fig. 1 shows the distribution companies in nigeria and their geographical regions of coverage. average hourly instantaneous electricity demand was obtained for a five week period (starting from 1st march to april 5, 2020) to observe the changing dynamics of electricity demand and consumption patterns within the residential, commercial and industrial sectors in a large segment of the lagos metropolis covered by one of the distributions companies. we choose not to reveal the particular distribution company where data collection was done as we do not have the permission to reveal their identity. indeed, data collection of electricity demand and consumption and subsequent analysis was done under three scenarios as stated below: business-as-usual scenario when there was no lockdown as a result of the covid-19 pandemic partial lockdown scenario resulting from a call for action to limit the spread of the covid-19 virus total lockdown scenario resulting from a need for more stringent measures to curb the spread of the covid-19 virus the detailed weekly average electricity demand (in megawatts) is shown in appendix 1 based on data collected from 259 feeder locations across lagos for each of the five weeks under study. the details in appendix 1 are a summary based on a detailed hourly data for the 259 locations over the five week period. in the next section, we present the salient findings based on data collected and subsequent analysis carried out. 3. findings: comparing the effect of covid-19 response on electricity demand and consumption in this section, we present the findings of the analysis of electricity consumption within the residential, commercial and industrial sectors of a large segment of the lagos metropolis under three different scenarios as defined in the preceding section. first, we define the characteristics of the three lockdown scenarios (in section 3.1) and then present the electricity consumption dynamics under the various lockdown scenarios (in section 3.2). in section 3.3, we present those services and practices that impacted on changes in electricity use under various lockdown scenarios. in various parts of the world, it can be observed that lockdown scenarios vary greatly depending on the services impacted. indeed, what does the various lockdown scenarios mean within the lagos context? what services were impacted that defined or characterized the different lockdown scenarios? table 1 provides a summary of the three lockdown scenarios and the services that impacted or characterized each lockdown scenario. changes in electricity consumption dynamics leading to the total lockdown were highly impacted by the stepwise announcement by the lagos state government (effecting a partial lockdown from march 23, 2020) and the federal government of nigeria (effecting a total lockdown from march 30, 2020). a five week dataset of electricity consumption by sector and within each scenario analysed is presented in table 2 . the summarized dataset upon which table 2 is based are presented in appendix 1. from table 2 , it can be observed that there was a gradual increase in electricity consumption in all three (residential, commercial and industrial) sectors in the three weeks preceding the lockdown as energy consumers envisaged that a lockdown was inevitable. the increase in electricity consumption was prompted by a certain fear of a possible lockdown based on the observations of many countries already implementing some form of lockdown to contain the spread of the covid-19 virus. businesses were ramping up services and industries were ramping up production in preparation for a possible inevitable lockdown. residences were stocking up gradually. more refrigerators and deep freezers were in use as more items needed safe storage. within the third week of the business as usual scenario, some businesses were already contemplating and implementing the policy of allowing some staff work from home, or work in the office on some predetermined days based on a shift arrangement. this is what explains the increase in absolute amount of electricity consumed across sectors within the first three weeks as observed in table 2 and in fig. 2 . at the end of the third week of the business as usual scenario, the lagos state government announced some steps leading to a partial lockdown as defined in table 1 . this impacted on electricity consumption in the fourth week under observation (under partial lockdown scenario). from table 2 and the corresponding fig. 2 , it is observable that average commercial sector electricity consumption dropped from 3.07 mw in week 3 (business as usual) to 2.81 mw in week 4 (partial lockdown). there was also a corresponding drop in electricity consumption in the industrial sector from 2.54 mw to 2.02 mw under the same conditions. however, for the residential sector, there was not much change in electricity consumption from week 3 (3.72 mw) to week 4 (3.72 mw) representing business as usual and partial lockdown scenarios respectively. energy consumers in the residential sector needed time to adjust to their new realities. this is the main reason for the almost constant electricity consumption in the sector in weeks 3 and 4 as observed in fig. 2 . transiting from week 4 (partial lockdown) to week 5 (total lockdown), we observe that within the commercial sector, electricity consumption dropped from 2.81 mw to 2.63 mw. the industrial sector experienced a drop in electricity consumption from 2.02 mw to 1.41 mw while the residential sector experienced an increase in electricity consumption from 3.72 mw to 3.87 mw. from the data, it was also observed that there was an increase in total electricity consumption in the three weeks preceding the lockdown (under the business as usual scenario) from 3.37 mw through to 3.50 mw at the end of the third week. changes in practices during the partial lockdown scenario led to a decline in total electricity consumption to 3.40 mw as shown in table 2 . the week of the partial lockdown was also a transitionary for many electricity users as they were trying out different things to know what professional practices can be moved to the homes. this explains why there was no increase in residential electricity consumption from week 3 (business as usual scenario) to week 4 (partial lockdown scenario) as shown in table 2 . fig. 3 shows the average hourly instantaneous electricity demand for the residential, commercial and industrial sectors under the business-as-usual, partial lockdown and total lockdown scenarios. it can b observed that under the business-as-usual scenario, electricity demand for the residential sector hovered between 250 mw and 380 mw all through a 24 h period as observed in fig. 3(a) . under the partial lockdown scenario, there was a table 1 different lockdown scenarios in lagos and what it means on the ground. what it means on the ground considerable variation in hourly average electricity demand for the residential sector as observed in fig. 3(b) . under the total lockdown scenario, the hourly average electricity demand for the residential sector surged by crossing the 400 mw mark for most part f the day as observed in fig. 3(c) . the graphs also show the hourly instantaneous electricity demand pattern for the commercial and industrial sectors. it should be noted that some commercial and industrial electricity consumers maintained a 24 h operation. this explains why there seem to be a flattening of the line across a 24 h period. fig. 4 shows the percentage change in electricity demand and under the various scenarios studied. under the business-as-usual scenario, it is observed that about 43% of total electricity demand comes from the residential sector, with corresponding demand from the commercial and industrial sectors accounting for 33% and 24% of total electricity demand respectively. however, no change was observed in the percentage electricity demand by sector under a partial lockdown scenario. under a total lockdown scenario, we observed a sharp increase in residential electricity demand which accounted for 49% of total electricity demand and consumption. there was also a corresponding decline in industrial electricity demand and consumption from 24% under a partial lockdown scenario to 18% under a total lockdown scenario. no change was observed in the percentage electricity demand for the commercial sector. the different lockdown scenarios impacted on changes in social practices in different ways which in turn had a corresponding impact on electricity use. changes in load profile in the residential sector were impacted by two main practices. -cooking as a practice, with increased use of hotplates/electric cookers, electric boilers and other kitchen equipment (including refrigerators) for storing food items for the lockdown. since entire households were in their homes, more cooking and eating occurred -professional practices, which spurred the increased use electricity for working from home. this impacted on greater use of electricity for air conditioning during the day and for powering basic office devices. other practices which contributed to increased electricity use within the residential sector, but not as significantly as the aforementioned practices, includes: laundry services (requiring greater use of washing machines in the homes rather than outsourcing to laundry services companies); showering (which increased with more people at home and considering that the lockdown occurred during the hot season) which means regular pumping of water from decentralized borehole systems; and entertainment (through increased use of television and video games in homes). within the industrial sector, manufacturing practices for nonessential and non-food products were ground to a halt during the total lockdown scenario which explains the reason for the decline in both absolute amount and percentage of total electricity consumed within that sector. in the commercial sector, there was a gradual shift of some professional services to the homes during the announcement of the partial lockdown measures which did not primarily affect them. some businesses envisaged that there would be an eventual announcement of a total lockdown. this impacted on increased electricity use in the residential sector and reduced electricity use in the commercial sector (as shown in table 2 ) as we transit from one lockdown scenario to another. however, there was a reduction in trading practices and the diverse forms of professional practices that occurs within the commercial sector during the shift from business as usual to partial lockdown and the eventual shift to a total lockdown scenario. table 3 shows a summary of social practices that impacted on electricity consumption in each sector under the various scenarios. the findings of this study shows that energy transition occurs in various ways with respect to speed of change, degree of permanence and temporality. indeed, we argue that energy transition can manifest itself under three contexts: 1. momentary energy transition that last for a short time owing to some imposed rules that leads to momentary behavioural change towards energy use. after some time, the energy consumers go back to business as usual scenario if there are no intrinsic motivations for change. this type of energy transition usually last for less than a year. in this study, we see how lockdowns can lead to momentary energy transition which is characterized by a change in the energy consumption patterns forced by the lockdown scenario. the results show that under lockdown scenarios, imposed rules can pave the way for momentary energy transitions which only last for a short time (mostly less than one year). indeed, a major challenge still exists. how do we effect a movement from momentary energy transition through to a permanent energy transition drawing from lessons from the lockdown? from the findings, we observe that certain professional practices can be carried out from home which can shape working patterns and an eventual change in energy consumption patterns. organizations are already seeing the possibilities and learnings provided by the lockdown and how this can shape future work and related professional practices. consumer behaviour plays a very vital role in electricity consumption patterns [8] . the quest for convenience, cleanliness and comfort has considerable impact on energy use [10] . indeed, behavioural change in energy and electricity use (as observed in this study) cannot occur without some external policy influence driven by some national or regional energy strategies [21, 22] . this is evident as observed in the changes that occurred from the partial lockdown to the total lockdown scenario. energy transition during the lockdown period leading to changes in electricity consumption patterns tells us that there might still be many unexplored scenarios that can help cities, countries and regions to achieve their energy transition goals. indeed, the partial and total lockdown scenarios necessitated by the global covid-19 pandemic have shown that there is a need to explore more scenarios in our electricity and energy systems design and planning. this study brings to mind the need for electricity systems planners to include resilience in the design and implementation of electricity systems [23, 24] . the task of effecting behavioural change in electricity consumption cannot be left to energy consumers alone. increasing electricity bills has not proven effective in effecting behavioural change. this is because electricity consumption pattern is something habitual and it has been argued that cost benefit analysis cannot change automatic behaviour [25] . indeed, behavioural change is important for a successful energy transition. in this study, we have shown how a forced lockdown due to the covid-19 global pandemic can lead to a momentary transition in energy use and changes in electricity demand patterns. we also see how a forced lockdown impacts on momentary behavioural change in energy use. indeed, it is important for governments and communities to re-evaluate the policy levers used in directing, managing and shaping changes in energy systems that can lead to a more permanent energy transition in energy consumption patterns. there is need to incorporate measures to improve systems resilience in responding to shocks caused by unforeseen adverse scenarios such as natural disasters, diseases and other forms of epidemic and pandemics while considering the learnings they present in effecting energy systems change. this research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. credit authorship contribution statement norbert edomah: writing -original draft. gogo ndulue: data curation, formal analysis, writing -original draft. the author declare no conflict of interest. the origin, transmission and clinical therapies on coronavirus disease 2019 (covid-19) outbreak -an update on the status covid-19) events as they happen, world heal organ updated understanding of the outbreak of 2019 novel coronavirus (2019-ncov) in wuhan quantifying economic resilience from inputeoutput susceptibility to improve predictions of economic growth and recovery monitor 2nd edition: covid-19 and the world of work updated estimates and analysis energy provision and housing development: re-thinking professional and technological relations peak electricity demand and social practice theories: reframing the role of change agents in the energy sector consumption and theories of practice converging conventions of comfort, cleanliness and convenience consumers, producers and practices: understanding the invention and reinvention of nordic walking conceptualizing connections: energy demand, infrastructures and social practices the dynamics of social practice: everyday life and how it changes influences on energy supply infrastructure: a comparison of different theoretical perspectives electricity and energy transition in nigeria federal government of nigeria federal government of nigeria policy making and energy infrastructure change : a nigerian case study of energy governance in the electricity sector provision of electricity to african households: the importance of democracy and institutional quality principles and criteria for assessing urban energy resilience: a literature review modelling future electricity: rethinking the organizational model of nigeria's electricity sector getting a grip on habitual practices: understanding the drivers of household energy consumption key: cord-306270-9bsr2jz0 authors: chowdhury, rajiv; luhar, shammi; khan, nusrat; choudhury, sohel reza; matin, imran; franco, oscar h. title: long-term strategies to control covid-19 in low and middle-income countries: an options overview of community-based, non-pharmacological interventions date: 2020-07-13 journal: eur j epidemiol doi: 10.1007/s10654-020-00660-1 sha: doc_id: 306270 cord_uid: 9bsr2jz0 in low and middle-income countries (lmics), strict social distancing measures (e.g., nationwide lockdown) in response to the covid-19 pandemic are unsustainable in the long-term due to knock-on socioeconomic and psychological effects. however, an optimal epidemiology-focused strategy for ‘safe-reopening’ (i.e., balancing between the economic and health consequences) remain unclear, particularly given the suboptimal disease surveillance and diagnostic infrastructure in these settings. as the lockdown is now being relaxed in many lmics, in this paper, we have (1) conducted an epidemiology-based “options appraisal” of various available non-pharmacological intervention options that can be employed to safely lift the lockdowns (namely, sustained mitigation, zonal lockdown and rolling lockdown strategies), and (2) propose suitable application, pre-requisites, and inherent limitations for each measure. among these, a sustained mitigation-only approach (adopted in many high-income countries) may not be feasible in most lmic settings given the absence of nationwide population surveillance, generalised testing, contact tracing and critical care infrastructure needed to tackle the likely resurgence of infections. by contrast, zonal or local lockdowns may be suitable for some countries where systematic identification of new outbreak clusters in real-time would be feasible. this requires a generalised testing and surveillance structure, and a well-thought out (and executed) zone management plan. finally, an intermittent, rolling lockdown strategy has recently been suggested by the world health organization as a potential strategy to get the epidemic under control in some lmi settings, where generalised mitigation and zonal containment is unfeasible. this strategy, however, needs to be carefully considered for economic costs and necessary supply chain reforms. in conclusion, while we propose three community-based, non-pharmacological options for lmics, a suitable measure should be context-specific and based on: (1) epidemiological considerations, (2) social and economic costs, (3) existing health systems capabilities and (4) future-proof plans to implement and sustain the strategy. the coronavirus disease-2019 (covid-19) pandemic has claimed more than 500,000 lives worldwide [1] and has been responsible for significant economic disruptions globally [2] . similar to the high-income nations, low and middle-income countries (lmics) also responded to covid-19 by implementing various population-level measures, including strict nationwide lockdowns and physical distancing [3] . worldwide, with no effective treatments for covid-19 and a vaccine at least a year away, these measures have been generally effective in preventing health systems from becoming overloaded, especially in the lmics where: (1) the risk of disease transmission is high (populations are often large and rajiv chowdhury, shammi luhar, and nusrat khan contributed equally. dense, with a high degree of interaction and physical contact), (2) awareness of how to prevent disease is often poor (eg, clean water and hygiene practices), (3) public health systems are often under-resourced (eg, safety equipment and intensive care units/icu), and (4) access to healthcare is limited and reliant on largely out-of-pocket payment. these strict social distancing interventions, however, come with a price: they are unsustainable in the long term given their social, economic and psychological impacts. for example, a recently completed survey in bangladesh showed that after its initial days of lockdown, a staggering 72% of urban and 54% of rural households had lost their main source of earnings [4] . therefore, many lmics are currently lifting the lockdowns, irrespective of the status of infection and the level of contagion. it remains, however, unclear what would be an optimal strategy for 'safe re-opening' (given the likelihood of disease resurgence), especially across low-income settings, where diagnostic capacities and surveillance infrastructure is poor [5] . in this regard, we have considered three communitybased non-pharmacological strategies for lmics (which aim to strike a balance between health protection and preventing economic collapse) and propose appropriate application, ideal pre-requisites, and inherent limitations for each. they include: (1) sustained mitigation, (2) zonal lockdown, (3) rolling lockdown (dynamic measures). these strategies (as summarized in fig. 1 ) should not be considered as mutually exclusive, and could be further adapted and combined depending on local disease epidemiology and socioeconomic circumstances. following the initial national lockdowns, staying on a 'mitigation-only' phase (a strategy adopted by developed countries such as france, switzerland and italy) has involved measures such as physical distancing, wearing masks, testtrace-isolation of suspected cases, shielding of the vulnerable and banning mass gatherings [6] . the successful implementation of this no-lockdown mitigation-only approach, however, is contingent on a number of key factors. first, the implementation of the earlier strict lockdown has resulted in a significant reduction of contact rates, new infections, and case fatality in the country [7, 8] . in this regard, somewhat worryingly, many lmics, where lockdown has recently been lifted, appear to have an upward trend of cases and deaths [9] . second, there is availability of nationwide surveillance, mass testing operations and rapid case isolation to tackle any resurgence and to facilitate containment [10, 11] . third, for contact-tracing, enough trained contact tracers (or scalable digital platforms) are available, with a relatively sparse target population (minimising the possibility of super-spreading events). in this regard, the effectiveness of contact-tracing might be importantly minimised in large, dense countries such as bangladesh (~ 1300 people/sqkm), compared to sparsely populated countries like spain (~ 90 people/sqkm) [12] . contact-tracing is also less effective at the height of community spread when the rates are on the rise. fourth, individual and population-level adherence to mitigation measures (eg, physical distancing, hygiene, home quarantine) will be ensured. for many lmics, however, this remains a challenge given large-scale social stigma and suboptimal risk communication strategies [13] . finally, healthcare services must be able to adequately cope with the resurgence in new cases, including availability of specialised care, hospitals and icu beds. in many lmics, there is however a chronic shortage of (1) critical care infrastructure (only 48,000 ventilators are available in india to serve its 1.3 billion people [14] ), (2) personal protective equipment (ppe), (3) training of health workforce, and (4) good working conditions-all of which reduce system efficiency and enhance likelihood of transmission among healthcare workers. despite being far less restrictive than a full lockdown, a mitigation-only strategy is also not immune to financial hardship as it can lead to some socioeconomic disruption (e.g., reduced production due to workplace social distancing) -somewhat compromising its sustainability over a prolonged period. for example, sweden adopted some of the most liberal mitigation measures in the world such as keeping restaurants, bars, and gyms open throughout the previous few months, whilst encouraging physical distancing rules. however, the country is still expected to suffer ~ 10% contraction in its economy in 2020 according to the swedish central bank [15] . the idea of fencing between infected and healthy communities, termed cordon sanitaire, has been deployed during a variety of outbreaks for centuries. in line with this principle, as an exit strategy, many countries have transitioned to a system of "zonal (or local) lockdown" [16] . this system entails identification of specific "hotspots" where a sudden outbreak cluster, with a high number of cases, have been identified in real time. such clustered social distancing works by dividing the population into "zones" according to the geospatial distribution of disease cluster contained within, so that interactions within a zone are significantly greater than interactions between zones [17] . transmission hotspots, or "red zones" are subject to strict lockdown measures than "green zones", where very few or no new cases have been identified for several days. such strategies were adopted in france [17] , fig. 1 a visual summary of the three proposed community-based non-pharamacological option for developing countries with green zones defined by areas where the virus transmission is relatively low and there is not as burdensome pressure placed on the healthcare system. the "zonal lockdown" approach has several important requirements. first, this categorisation of hotspots is typically a dynamic process, which requires an ability to reliably identify, in real time, areas that meet or fall short of the pre-specified lockdown criteria. this requires continuous data-driven feedbacks on: (1) regional daily confirmed cases (either by date of reporting or onset of symptoms), and (2) other time-series information needed to calculate the changes in region-specific effective reproduction number (r, the average number of secondary infections per infected individual), including daily numbers of hospitalized cases, daily numbers of deaths in different age groups, and transmission dynamics (eg, average time from infection to death) [17] . while such strategy has been successfully established in developed settings (such as france, where testing is widespread with 0.52 daily tests being done per 1000 population), this remains challenging in many lmics due to (1) absence of large-scale population surveillance system based on randomly-selected individuals (e.g., in bangladesh, the testing approach has focused on purposive, self-referred samples, with significant selection bias), and (2) poor testing laboratory facilities and reporting capacities (e.g., in pakistan, only 0.09 daily tests are being conducted per 1000 individuals) [18] . in this regard, india has adapted a scalable mass "pool testing" approach [19] . this cost-effective strategy involves collecting multiple samples in a tube and testing them with a single rt-pcr assay run. if the test is negative, all the people tested are negative. if it is positive, every person has to be tested individually for the virus. this approach reduces the time needed to test large swathes of the population [20] . second, the classification of the zones should also be multifactorial. this should not only take into consideration the incidence rate, but also the other epidemiological (e.g., doubling rate of new cases; number of deaths) and administrative aspects (e.g., available hospital and icu beds; testing and surveillance structure; residential versus industrial zone). third, managing the zones efficiently to reduce transmission both within and outside of the zones is a major undertaking. recent reports from india shows that infection size in many containment areas is 100-fold to 200-fold higher than the cases reported at those sitesindicating that containment efforts within zones may not have fully paid off [21] . therefore, detailed apriori standard operating procedures should be devised to include aspects on (1) within-zone public health measures (eg, risk communication, house-to-house surveillance, test booths, contacttracing, case referral systems, ambulance and medical facilities), (2) within-zone measures of emergency services (eg, food supply, law enforcement, isolation centres, and burial facilities), and (3) outside-zone measures such as creation of "buffer" zones (e.g., in india [19] ) that surround the main containment zone to minimise out-of-zone transmissions. such detailed protocols are crucial for efficiency. in iran, for example, suboptimal zone management has increased risk of a second wave [22] . finally, similar to sustained mitigation strategy, the zonal lockdown will be most effective when the overall rate of infection is in decline, accompanied by exhaustive vigilance. while zonal lockdown, if implemented properly, can help contain the spread of the virus, efficacy of this approach can be reduced by other concurrent transmission networks, such as those linked to economic and social interdependency between zones [17] . additionally, the impacts on the economy, particularly inside the zones, can be considerably more severe than under mitigation where the economy essentially opens with restrictions, exacerbating economic hardship in countries with already weak economic performance and social security nets. therefore, these aspects merit careful consideration during the planning phase of this strategy. intermittent or "rolling" lockdown measures take place when strict social distancing measures are applied and lifted periodically. this strategy has been described as a potentially effective measure to minimise uncertainty in both effective r values, and in the severity of the virus (i.e. the proportion of cases requiring icu admission) [23] . this approach may be particularly suitable for the lmics with large and dense populations, high patterns of contact, poor economic/ health systems resilience, and weak testing/contact tracing capacities. furthermore, this approach addresses both key elements of society that needs safe-keeping: life and livelihood, and aims to provide a balance between avoiding public health systems being overloaded and grinding economies completely to a halt [24] . a recent paper mathematically modelled the effects of either a strict 50-day suppression or a 50-day mitigation, followed by 30 days of relaxation (during which businesses are allowed to reopen, with basic hygiene measure kept in place), in 16 economically diverse countries. in these models, a strict 50-day lockdown, that reduces the effective r value to 0.5, prevented icu beds overload and led to considerably fewer deaths (130,000 during 18 months in the 16 countries they modelled) compared to a more relaxed 50-day mitigation/30-day relaxation cycle (~ 3.5 million predicted deaths globally) and under nointervention (counter-factual) scenario (8 million predicted deaths) [25] . to further contextualize the value of such concept, a subsequent paper estimated that (1) a single, one-off lockdown will be insufficient to bring the pandemic under control, and (2) secondary peaks would be larger than the first, without continued restrictions [26] . however, as with the other strategies, rolling lockdown approach is also contingent on several factors. first, before implementing a rolling lockdown, every developing country should carefully consider the economic and social costs to implement these measures. second, impacts on incidence and case-fatality will rely on local levels of adequate adherence to social distancing measures. third, this approach would also bring a new set of logistical challenges. therefore, countries will need to formulate bespoke plans for reorganising business supply chains, so that they align with the economy opening and closing. while such readjustments to complement a schedule of lockdown is not ideal, unprecedented challenges often require unusual and adaptive solutions, especially if other alternative exit strategies are not feasible. finally, by establishing a detailed surveillance system while the lockdown takes place, countries should adapt the duration of the lockdown and relaxation periods according to the local growth rate and pattern of the epidemic. a recent example of this has been in pakistan, where the world health organization has recommended a 14-day-on/14-day-off rolling lockdown to control the epidemic [27] . similarly, rolling lockdowns do not have to be generalised, these can also be adapted as regional or zonal rolling lockdowns within a country, i.e., to apply specifically in areas with high and sustained new-onset covid-19 cases per population. for example, zonal rolling lockdowns have been proposed in the gauteng province of south africa-one of the worst affected regions in the country-to control the rapid increase in infection rates [28] . while many lmi countries are currently lifting the lockdowns due to economic reasons, it is crucial for the policy makers to recognise that preserving health is equally important for reviving the economy. this is of important relevance to the lmics where large proportions of working-age population are vulnerable to adverse covid-19 outcomes, owing to high prevalence of comorbid conditions (such as diabetes, obesity and hypertension) [29] . furthermore, if a country has constant high incidence of a deadly disease, it may become rather challenging for the local economy to thrive in such environment [30] . therefore, equal priorities must be put on protecting lives as well as livelihood when adapting an exit plan. in this regard, we have proposed several non-pharmacological strategies that may enable the lmics to safely open the economy, while allowing for preservation of health. however, it is crucial that the selection of a suitable, "context-specific" strategy is based on some key considerations: (1) local epidemic growth rate, (2) existing health infrastructure (to survey, test, and treat, at scale), (3) social and economic costs, and (4) carefully-devised plans to implement and sustain the measures. author contributions rc conceptualised the paper. rc, sl and nk led the manuscript drafting. ohf, src and im provided scientific and technical comments to strategic options discussed in this manuscript. rc, sl and nk did the necessary background literature review. im and rc produced the visual abstract. rc leads several covid-19 projects and published covid-19 modelling studies related to nonpharmacological interventions. sl is an epidemiologist involved with various cambridge-led covid-19 projects. nk is a clinician and doctoral researcher in global public health involved with various cambridge-led covid-19 projects. ohf is the director of institute of social and preventive medicine at the university of bern, where he advises various national (switzerland) and regional (latin america) covid-19 technical response committees. src is a member of covid-19 technical response committee in bangladesh. im is the executive director of brac institute of global development, leading an ongoing brac-soas-cambridge project to formulate an adaptive and integrated framework for health responses to covid-19 in the developing countries. conflict of interest none to declare. open access this article is licensed under a creative commons attribution 4.0 international license, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the creative commons licence, and indicate if changes were made. the images or other third party material in this article are included in the article's creative commons licence, unless indicated otherwise in a credit line to the material. if material is not included in the article's creative commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. to view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. covid-19 coronavirus pandemic. covid-19 coronavirus pandemic global coronavirus watch: ems weigh health vs economy the great lockdown: worst economic downturn since the great depression. imf blog (global economy) understanding the impacts of covid-19 on livelihoods in bangladesh: findings from the pprc-bigd rapid response survey public health surveillance: a tool for targeting and monitoring interventions from mitigation to containment of the covid-19 pandemic: putting the sars-cov-2 genie back in the bottle isolation, quarantine, social distancing and community containment: pivotal role for old-style public health measures in the novel coronavirus (2019-ncov) outbreak the positive impact of lockdown in wuhan on containing the covid-19 outbreak in china infections are rising fast in bangladesh, india and pakistan. the economist coronavirus: a common approach for safe and efficient mobile tracing apps across the eu, brussels contact tracing, testing, and control of covid-19-learning from taiwan stigma around covid-19 hampers the fight against it covid-19 in india: state-wise estimates of current hospital beds, intensive care unit (icu) beds and ventilators. the center for disease dynamics the riksbank is supporting an economy in crisis. stockholm. 2020 covid-19: how does local lockdown work, and is it effective? exit strategy: from selfconfinement to green zones statistics and research coronavirus (covid-19) testing. our world in data guideline for rt-pcr based pooled sampling for migrants/returnees from abroad/green zones containment plan novel coronavirus disease 2019 (covid 19) version 2 15-30% people in containment areas exposed to covid-19: icmr's serosurvey. the new indian express iran fears second wave after surge in cases impact of non-pharmaceutical interventions (npis) to reduce covid-19 mortality and healthcare demand rolling lockdowns could protect both economies and health in low-income countries. the conversation dynamic interventions to control covid-19 pandemic: a multivariate prediction modelling study comparing 16 worldwide countries projecting the transmission dynamics of sars-cov-2 through the postpandemic period who recommends pakistan reimpose intermittent lockdowns as covid-19 cases rise sharply. the reuters gauteng looking at 'intermittent' lockdown as it prepares for worst in covid-19 cases. sunday times low-and middle-income countries face up to covid-19. the nature social distancing and supply disruptions in a pandemic. cambridge working papers in economics key: cord-117800-jzokod4q authors: umer, hamza; khan, muhammad salar title: evaluating the effectiveness of regional lockdown policies in the containment of covid-19: evidence from pakistan date: 2020-06-04 journal: nan doi: nan sha: doc_id: 117800 cord_uid: jzokod4q to slow down the spread of covid-19, administrative regions within pakistan imposed complete and partial lockdown restrictions on socio-economic activities, religious congregations, and human movement. here we examine the impact of regional lockdown strategies on covid-19 outcomes. after conducting econometric analyses (regression discontinuity and negative binomial regressions) on official data from the national institute of health (nih) pakistan, we find that the strategies did not lead to a similar level of covid-19 caseload (positive cases and deaths) in all regions. in terms of reduction in the overall caseload (positive cases and deaths), compared to no lockdown, complete and partial lockdown appeared to be effective in four regions: balochistan, gilgit baltistan (gt), islamabad capital territory (ict), and azad jammu and kashmir (ajk). contrarily, complete and partial lockdowns did not appear to be effective in containing the virus in the three largest provinces of punjab, sindh, and khyber pakhtunkhwa (kpk). the observed regional heterogeneity in the effectiveness of lockdowns advocates for a careful use of lockdown strategies based on the demographic, social, and economic factors. the world is struggling to combat the covid-19 pandemic, which has spread from wuhan china to over 190 countries. unless there is a viable treatment or vaccine to treat covid-19, the world is taking possible preventive measures to minimize the spread of the disease. lockdown (complete or partial) is one of the most evident and widely used preventive measures. the effectiveness of lockdown in controlling the spread of covid-19, however, is not a well-established outcome for a few reasons. first, so much is unknown about the novel coronavirus and that the situation is evolving, which gives policymakers little time to think through, implement, and properly investigate or foresee the effectiveness of any policy, such as a lockdown. second, while many countries-for instance, new zealand and germanyhave already implemented complete or partial lockdown (mcfall-johnsen et al., 2020) , it is just hard to conduct a long-term ex-ante analysis of the repercussions of policy concerning once-in-a-century pandemic. still, some preliminary studies tried to investigate the effect of lockdown strategies on the spread of covid-19. for instance, walker and colleagues (2020) predicted that interventions and lockdown strategies in almost all countries (precisely, 202 countries in their analysis) would reduce infections and deaths by nearly half. mushfiq mobarak and colleagues (2020) , contrarily argued that while lockdown strategies are viable umer & khan 3 options in high-income countries, low-income countries such as nigeria and pakistan cannot afford to have a fruitful lockdown. because of their weak capacity in enforcing lockdown strategies, these countries may witness counterproductive effects if such strategies make workers and migrants migrate back from heavily populated urban areas and spread the disease to remote rural areas, the researchers argue. similarly, we have a few studies that show different outcomes on the country-level. for example, research (dowd et al., 2020 ) on the italy outbreak shows the effectiveness of early lockdown. in italy, the covid-19 was first detected in the lodi province, which placed restrictions beginning february 22. as opposed to this, bergamo province, which started with fewer cases but did not impose restrictions until march 7, far surpassed the number of cases in lodi (stancati, 2020) . in a similar vein, a study (kumar and nataraj, 2020) on indian regions shows a regional differential in the spread of covid-19 in the face of regional policy variation. overall, the world is divided regarding the use and effectiveness of lockdown policies. on one side, we see countries like japan (du and huang, 2020; ian, 2020) and sweden (karlson et al., 2020) that have used mild lockdown or no lockdown and yet effectively contained the spread of the virus. on the other side, we see countries like new zealand and australia using strong lockdown policies to flatten the spread of the virus (fifield, 2020) . this article offers a systematic contribution to the aforementioned debate by examining the effectiveness of lockdown policies in the containment of covid-19 in the context of pakistan. specifically, the article explores how effective the lockdown strategy has been in combating covid-19 outcomes in the country. this evaluation of lockdown policies is based on the econometric analysis (such as regression discontinuity and negative binomial regressions) performed on official data from pakistan. pakistan is selected because it offers a valuable opportunity to analyze the effects of both complete lockdown and partial lockdown policies on the spread of the covid-19 virus. moreover, the regional use of lockdown policies in pakistan is heterogeneous and hence enables us to perform cross-regional analysis as well. we find that in comparison to no lockdown, complete and partial lockdowns have been ineffective in the containment of the virus in the three largest provinces punjab, sindh, and khyber pakhtunkhwa (kpk). on the other hand, complete and partial lockdowns have been umer & khan 4 very effective in the containment of the virus in the province of balochistan and the three administrative territories/regions of gilgit baltistan (gb), islamabad capital territory (ict), and azad jammu and kashmir (ajk). the observed regional heterogeneity in the effectiveness of lockdowns advocates for a careful use of lockdown based on the demographic, societal, and economic factors. "one size fits all" approach for lockdown could be counterproductive in some regions of the world and subsequently make the spread of virus more acute, as demonstrated by researchers in the context of africa as well (mehtar et al., 2020) . pakistan-home to about 220 million and wobbly health infrastructure that has close to 1.5 million hospital beds (khan & latif, 2020) -reported the first case of the covid-19 on february 26, a returning pilgrim from neighboring iran (hashim, 2020) . on the same day, the pakistan federal ministry of health confirmed another positive case in islamabad (ali, 2020) . since then, the virus has diffused quickly. by march 18, all the administrative regions of pakistan, including four provinces (punjab, sindh, kpk, and balochistan), the two autonomous territories (ajk and gb), and the federal territory of islamabad registered positive cases. the entire country reported over eighty-five thousand confirmed cases and 1,770 deaths, as of june 4, 2020. 1 in terms of the total number of cases and deaths, pakistan ranks 17 th and 21 th worldwide, respectively. 2 while now the virus has entered the community transmission stage, initially, all the confirmed cases in pakistan had recent travel history from iran, syria, london, and saudi arabia. the covid-19 pandemic has spread unevenly across regions within pakistan, with the four regions (provinces) making up more than 95 percent of the cases as of june 4, 2020. sindh registered the most cases at over 39,900, followed by punjab (31,104), kpk (11, 373), and balochistan (5, 224) . 3 the province of punjab reported 607 deaths, the most in the country, followed by sindh (555) and kkp (500) and balochistan (51). the situation in the three special regions or territories is not that bad, with islamabad, gb, and ajk reporting 3,544, 1 this data was obtained from worldometers: https://www.worldometers.info/coronavirus/ the data was cross-verified here: https://www.cdc.gov/coronavirus/2019-ncov/global-covid-19/world-map.html 2 ibid. 3 official government website for data on covid-19 was consulted: http://covid.gov.pk/stats/pakistan source: http://covid.gov.pk/stats/pakistan as a nascent federalist country, 5 when the interior ministry of pakistan announced a lockdown on march 23 to combat the spread of the virus, all the seven administrative regions also implemented their regional lockdown measures at or around march 23. army troops were deployed throughout the country to help the divisions in tackling the spread of the virus. 6 initially, the regions implemented a full or complete lockdown, with ict shutting down as early as on march 12, punjab on march 22, and all other divisions on march 24. 7 4 ibid. 5 pakistan is a federalist country, with provincial governments having the right to decide on important issues, according to the eighteenth amendment of the constitution of pakistan. 6 https://www.geo.tv/latest/278812-government-calls-in-pakistan-army-troops-amid-coronavirus-outbreak 7 complete lockdown and partial lockdown information is extracted from several newspapers and online sources. these include: business recorder, dawn news, radio pakistan, technology times. also please refer to the report (covid-19 legislation and measures in pakistan) published by zafar kalanuri (2020). complete lockdown refers to complete shutdown of socioeconomic, religious activities and mobility pathways while partial lockdown refers to controlled opening of the aforementioned. complete description of lockdown types is mentioned on page 6. the duration of this full lockdown also varied, with ict and kpk observing a short duration lockdown for less than a week, punjab, sindh, and balochistan observing a medium duration lockdown for about two weeks, and ajk and gb observing a long duration lockdown for almost a month. later, the divisions moved to a partial or controlled lockdown (please see table 3 for details). policy-wise, pakistan acted quickly and formulated a national action plan for covid-19 early in february 2020 (mukhtar, 2020) . the ministry of national health services, regulation & coordination pakistan presented the plan, that was supposed "to provide (a) policy framework for federal, provincial, and regional stakeholders for building capacity to prevent, detect and respond to any events due to covid-2019 in pakistan." 8 along with the health ministry, national and provincial disaster management authorities, national command and operation center (ncoc), and national coordination committee (ncc) have been formulating, coordinating, analyzing, and implementing policy efforts about covid-19. with federal directives, regions have been managing outbreaks according to their circumstances. table 1 provides details about the total number of positive cases, deaths caused by and total tests performed in the seven regions of pakistan. workplaces, parks and other public places, ban on social gatherings and social events, closing of land and air transport and people are restricted to stay home unless they need medical help or require grocery shopping. partial lockdown refers to the controlled opening of economic activities for a specific time every day, limited resumption of land and air transport, and maintaining social distance during outdoor activities. most of the educational institutes, however, remain closed. 11 the data covers a time period of 60 days (march 12 to may 11, 2020) and encompasses no lockdown, complete lockdown, and partial lockdown phases. the entire data variables are described in table 2 . it is a time variable that represents day. in table 3 , the summary of the variables is reported. punjab, sindh, and kpk are three provinces witnessing the highest number of average daily positive cases and average daily deaths. ict had the longest complete lockdown (34 days), while kpk had the shortest one (5 days). 3/12--5/11 3/12--5/11 3/12--5/11 3/12--5/11 3/12--5/11 3/12-lockdown 12 note: n = observations; m = mean. standard deviations are in parentheses. *ajk did not report any death during the data duration specified in this paper. all the econometric analyses are performed in stata 16. the outcome variables include the number of daily deaths due to ccovid-19 and the number of people testing positive for ccovid-19. the main explanatory variables are lockdown and partial lockdown dummies. the starting point of the analysis is a visual representation of the impact of lockdown policies on the outcome variables. this is achieved by using regression discontinuity (rd) with the date as running variable and the lockdown and partial lockdown dates as multiple cutoffs. in stata and kpk) cumulatively representative of 96% of covid-19 cases are in figure 2 . as these four provinces account for the majority of the positive cases, we discuss them at length here, while binned scatter plots for ict, ajk, and gb regions are reported in appendix a. the binned scatter plots reported above indicate the lockdown policies induced heterogeneous regional effects on covid-19 outcomes. we further analyze the impact of lockdown policies on covid-19 outcomes systematically by using different regression techniques and estimate the following two equations. in the above equations ∑ 7 15 lr test is used for the evaluation of alpha. null hypothesis alpha =0 is rejected at 1% for both equations. and indicates poisson regressions had overdispersion (conditional variance exceeds conditional mean). hence, we use the output from negative binomial regressions for analyzing the effects of lockdown policies in table 4 . robust standard errors are in parentheses. ***p<0.1; **p<0.5; *p<0.1 regressions 1 while holding all other variables constant, the effect of daily tests performed is significant and positive on daily positive cases, however, the magnitude is very small (difference in logs of expected counts increases by 0.0002 in regression 1). next, we turn to main explanatory variables. in comparison to no lockdown, complete lockdown had a significant and negative discussion section 7, we explore the possible reasons leading to regional heterogeneity in the outcomes of lockdown policies. in this section we turn to another proxy for lockdown-the daily stringency score estimated by hale et al. (2020) . the score varies from zero to 100; a higher value indicates more stringent controls to contain the virus spread. the stringency score is based on the cumulative value of restrictions imposed on schools, workplaces, public events, social gatherings, public transport, stay home orders, domestic and international travel, public information campaign, testing policy, and contact tracing (hale et al., 2020) . essentially, lockdown and stringency measures are two different ways of quantifying restrictions in an economy. the stringency score by hale et al. (2020) is for the whole country, and resultantly, we are unable to do regional analysis using this score. therefore, we pool the regional data to obtain country-level data and subsequently perform analysis using daily stringency score as the main explanatory variable. the stringency score has 55 observations; its value ranges from 34 to 97, with a mean value of 86.29 and a standard deviation of 18.51. using the stringency score following two equations are estimated. negative binomial regression is used; the value of alpha is significant 17 , indicating poisson regression had over -dispersion. the regression output is reported in table 5. in regression 2 and 4, the variable for daily tests is excluded to check the robustness of the effect of the stringency score on the two outcome variables. this exclusion, however, does not alter the significance of the stringency variable. the effect of daily tests on daily deaths is insignificant in the main regression (1) level results show that stringency measures appear to be ineffective in the control of damage caused by covid-19 and in its spread as well. the regional results show heterogeneity in the effectiveness of lockdown measures while country-level results point towards the ineffectiveness of stringency measures. to reconcile the regional results with country-level results, we need to focus on the three important regions of punjab, sindh, and kpk. together these three regions account for 90% of the total positive cases and 96% of the total deaths during the time period considered in this study. as these three regions represent a huge share of covid-19 outcomes, they are playing a significant role in driving the country-level statistics. resultantly, the ineffectiveness of lockdown policies in these regions is reflected in the ineffectiveness of stringency measures at the country-level. as discussed earlier, the effect of regional lockdown policies on covid-19 outcomes is heterogeneous, with very few regions effectively using the lockdown policies to contain the spread of the disease. from a policy perspective, we identify here the factors that have contributed jointly to the ineffectiveness of lockdowns, specifically in the three largest regions of punjab, sindh, and kpk. the central government in pakistan has never been unified over the imposition of lockdown measures. the country's prime minister (pm) adamantly opposed lockdown fearing economic impacts on daily wagers that comprise a significant proportion of the country's labor force. on the other hand, several ministers in the pm's cabinet proposed strict lockdown measures to contain the virus. as a result of this confusion prevalent in the government circles, the potential risks of covid-19 were downplayed, the public did not observe the lockdown restrictions seriously, and resultantly the lockdown proved to be ineffective in three large provinces 18 . a significant proportion of pakistan's population inhibits in punjab (110 million), sindh (47.9 million) and kpk (30.5 million) regions (wazir & goujon, 2019) . together these three regions account for approximately 89% of the country's population, and a major proportion of this population resides in rural areas or slums in large cities (for example sindh's capital city karachi has the world's largest slum population approximated to be 2 million 19 ). in rural and specifically slum areas, social distancing or keeping one restricted to home are almost nonexistent, issues of cleanliness are acute, and poverty rates are high. all these factors, when combined together, provided an ideal habitat for the sharp spread of covid-19, even in the presence of lockdown measures. the informal labor force makes up 75% of the total labor force (48.75 million). about 40% of this informal labor force (19.5 million) is employed by the agriculture sector, while 60% (29.25 million) works for the industrial and service sector. 20 the majority of the informal labors are paid on a daily basis and unfortunately forced to leave homes in an attempt to earn money for food and subsequently violate and undermine the effectiveness of lockdown restrictions by serving as a potential source of virus spread within their workplace and residences. pakistan is a religiously homogeneous country with muslims making up 96.28% of the population. 21 in islam, daily five congregational prayers in the mosque are a vital part of worship. congregations and gatherings could be favorable grounds for viral transmission. in pakistan, the religious, as well as political leaders, were divided on the closing down of mosques. resultantly, government in coordination with religious leaders set up protocols for congregational prayers in mosques, however in most cases these protocols were not strictly followed, possibly leading to rapid transmission of virus even during the lockdown phases. 22 as an entirely unprecedented situation for masses, the heterogeneous behavioral response of the public to lockdown was evident. in the case of pakistan, poor handling of the situation by the government over using force to implement lockdown strategies in the big provinces led to fear in the minds and created a social stigma around covid-19. subsequently, people with symptoms kept on living with family members, and whole families got infected. these numbers eventually started showing up post lockdown in the largest regions. a lack of literacy and misinformation about the disease and its treatment further complicated the outcomes. on the contrary, in ajk and gb, where people were willing to observe lockdown, showed better outcomes. one reason why they showed this willingness is likely their high literacy rate than the national average rate. from a policy perspective, it is clear that lockdown effectively used by rich countries, including germany, japan, and usa, is to a large extent, ineffective in controlling the spread of virus in a poor country like pakistan. in fact, several researchers have already discussed the possibility of the lockdown being ineffective in poor countries (for example barnett-howell & mobarak, 2020; cash & patel, 2020) . the country-level results from this article support the predictions of these studies. we think contextualized strategies would be way more effective in the control of the virus in poor countries. in the case of pakistan, the socio-economic and political conditions and religious norms are some of the important contextual elements that should be considered while making future lockdown strategies. the purpose of the paper was to examine the impact of lockdown strategies in different regions ▪ federal and local governments should involve local religious scholars and community elders in teaching the public the importance of social distancing, hygiene, and prevention. they should also work on dispelling and addressing the stigma around covid-19 so that people can test themselves without anxiety and fear from society. ▪ government should allow shorter working hours, limit number of people at religious and social gatherings, and implement universal masking using cloth masks for the community. (1) (2) (3) robust standard errors are in parentheses. ***p<0.1; **p<0.5; *p<0.1 pakistan confirms first two cases of coronavirus, govt says 'no need to panic should low-income countries impose the same social distancing guidelines as europe and north america to halt the spread of covid-19? what japan can teach the world about the pandemic has covid-19 subverted global health? a practical introduction to regression discontinuity designs how does your kindergarten classroom affect your earnings? evidence from project star demographic science aids in understanding the spread and fatality rates of covid-19 did japan just beat the virus without lockdowns or mass testing? umer & khan 24 new zealand isn't just flattening the curve. it's squashing it variation in government responses to covid-19. blavatnik school of government working paper the curious case of south asia's 'low' coronavirus deaths covid-19 legislation and measures in pakistan sweden's coronavirus strategy will soon be the world's the coronavirus and pakistan: why people must immediately begin social distancing?" the national interest the impact of covid-19 mobility restrictions in india: comparing state and central responses assessing the 2017 census of pakistan using demographic analysis: a sub-national perspective countries around the world are reopening -here's our constantly updated list of how they're doing it and who remains under lockdown limiting the spread of covid-19 in africa: one size mitigation strategies do not fit all countries preparedness and proactive infection control measures of pakistan during covid-19 pandemic outbreak national action plan for corona virus disease (covid-19) pakistan lockdown of recovering italian town shows effectiveness of early action the global impact of covid-19 and strategies for mitigation and suppression key: cord-034579-3s26tjrd authors: mcauley, hamish; hadley, kate; elneima, omer; brightling, christopher e; evans, rachael a; steiner, michael c; greening, neil j title: copd in the time of covid-19: an analysis of acute exacerbations and reported behavioural changes in patients with copd date: 2020-10-30 journal: erj open res doi: 10.1183/23120541.00718-2020 sha: doc_id: 34579 cord_uid: 3s26tjrd introduction: the impact of the sars-cov-2 pandemic, and lockdown measures, on acute exacerbations of copd (aecopd) is unknown. we aimed to evaluate the change in aecopd treatment frequency during the first 6 weeks of lockdown in the uk compared with 2019 and assess changes in self-reported behaviour and well-being. methods: in this observational study in leicestershire, uk, patients with copd under a secondary care clinic were recruited. exacerbation frequency in the first 6 weeks of covid lockdown was compared with the same period in 2019 using electronic health records. a telephone survey was used to assess changes in anxiety, inhaler adherence, physical activity, and behaviour during the pre-lockdown and lockdown periods compared to normal. results: 160 participants were recruited (mean [sd] age 67.3 [8.1] years, 88 [55%] male, fev1 34.3 [13] % predicted). 140 [88%] reported at least one aecopd in the previous year. significantly more community treated exacerbations were observed in 2020 compared with 2019 (126 versus 99, p=0.026). the increase was a result of multiple courses of treatment, with a similar proportion of patients receiving at least one course (34.4% versus 33.8%). during “lockdown” participants reported significantly increased anxiety, adherence to their preventative inhalers, and good adherence to shielding advice (all p<0.001). a significant reduction in self-reported physical activity and visitors was reported (both p<0.001). discussion: treatment for aecopd events increased during the first 6 weeks of the sars-cov-2 pandemic in the uk compared to 2019. this was associated with increased symptoms of anxiety and significant behavioural change. acute exacerbations of chronic obstructive pulmonary disease (aecopd) are a frequent problem for people with copd, adversely affecting morbidity and mortality and are an important cause of unscheduled healthcare contacts including admission to hospital 1 global initiative for copd (gold) report grades the severity of these events according to treatment requirement, defining moderate events as those needing community provision of oral antibiotics and corticosteroids and severe events as those requiring hospitalisation 2 . healthcare provision for people with copd has been impacted by the severe acute respiratory syndrome coronavirus 2 (sars-cov-2) pandemic through the requirement for more distant/remote contact with the healthcare team to reduce the risk of virus transmission 3 . patients with copd who present with sars-cov-2 infection have a poorer prognosis 4, 5 , highlighting the need for primary prevention and risk reduction 5 . additionally, because of the appreciation of a greater risk of morbidity from sars-cov-2 infection 6 more stringent social isolation has been recommended for people with copd during the period of societal lockdown that has been implemented in most countries affected by the pandemic. in the uk this has been termed "shielding" and includes advice against leaving home for any reason other than for essential work or shopping with very limited exceptions. during this time healthcare professionals providing care for people with copd have reported lower than expected presentation rates for aecopd in both community and acute hospital settings 7 8 . however, it is unclear whether this is due to a genuine reduction in aecopd rates (potentially due to lower respiratory viral transmission 9 and/or atmospheric pollution) or due to higher thresholds for patient reporting to healthcare services because of fearfulness about contracting sars-cov-2 in healthcare environments, particular hospitals. in addition there is limited understanding of the impact of enhanced shielding on the psychological wellbeing and physical activity in people with pre-existing respiratory disease such as copd 10 11 . having a chronic condition such as copd does not appear to increase the likelihood of sars-cov-2 infection 12 , it does convey increased risk of hospitalisation and death 6 . firstly, in this observational study we recorded the change in moderate and severe aecopd treatment frequency (assessed objectively through prescription records or requirement for hospitalisation) during the first six weeks of societal lockdown in the uk compared with the equivalent period 12 months previously. secondly, we assessed self-reported behaviour change during the pre-lockdown and lockdown period by telephone interview in order to explore potential reasons for any observed changes in aecopd treatment frequency we compared rates of treatment for exacerbations of copd managed in the community and hospital setting in a single centre between the first six-week period of the sars-cov-2 "lockdown" in england (15 th march 2020 to 30 th april 2020) with the same six-week period the previous year (15 th march 2019 to 30 th april 2019). participants were prospectively recruited between 2 nd june, 2020 and 8 th july, 2020 and provided informed consent. all recruitment and interview calls were made either by experienced respiratory research nurses or clinicians. ethics approval was granted by the electronic community prescription records were used to record community exacerbation events and electronic hospital records similarly for hospital exacerbations. the terms community and hospital exacerbation, rather than moderate or severe 2 , are used in in this study due to the known change in hospital admission criteria during the peak of the sars-cov-2 pandemic in england when hospital bed capacity was considered of critical importance. community managed exacerbations were defined as those resulting in a prescription for oral corticosteroids and/or antibiotics but without hospital admission. only prescriptions for antibiotics that would typically be used to treat aecopd events in our region were included. hospitalised exacerbations were defined as admissions to hospital with a recorded discharge diagnosis of acute exacerbation of copd. as an additional analysis a scripted telephone survey was conducted to explore potential reasons for differences in exacerbation risk. participants were asked to compare behavioural and emotional changes with their baseline "normal" state as a reference. participants were asked to compare two discrete periods; (1) pre-lockdown, defined as the two weeks prior to "lockdown" (1 st march 2020 to 14 th march 2020) when participants were likely to be more aware of the threat of sars-cov-2 but restrictions had not yet been placed and (2) the "lockdown" itself (15 th march 2020 to 30 th april 2020). self-reported behaviour included; medication adherence to their regular prescribed inhaled therapy, anxiety, selfreported change in activity levels, and social behaviour (self-isolation, shielding, visitors to the home, arrangements for shopping). answers were captured with either a binary response (yes/no) or on a five-point likert scale (see online supplement for full details of questionnaire used). participants were eligible if they had a confirmed diagnosis of copd, under a specialist copd clinic (complex copd clinic, leicester, uk 13 ), and able to provide informed verbal consent via english language telephone consultation. the specialist clinic accepts any patient with copd with ongoing symptoms (e.g. admission to hospital for aecopd, mmrc >2, continued smoking, low body mass index, potentially lung volume reduction candidate, or established respiratory failure). all patients had confirmed airflow obstruction by gold criteria 2 . patients were contacted sequentially from the research database held in our centre of 258 patients alive at the start of the study who have previously consented to be contacted for research until this list was exhausted with a total of 3 attempts to contact participants who did not initially. the telephone call was made by either a nurse or doctor and participants gave informed consent verbally with this documented by the investigator due to the remote nature of the consultation. electronic gp and hospital healthcare records were used to capture new prescriptions for oral antibiotics or corticosteroids during the periods of interest, hospital admissions, as well as baseline characteristics, including latest spirometry. all spirometry had been performed at their previous clinic visit to glenfield hospital, leicester to ers/ats standard 14 . baseline data were described as mean (standard deviation), or n (%) as appropriate. paired data were compared using a paired t-test or signed-rank test for parametric data and nonparametric data respectively. categorical data were compared using chi squared. statistical analysis was performed using stata 16 (statacorp, usa). from previous data from our copd clinic we anticipated 0.8 exacerbations per patient in the observation period with a sd of 0.9. to detect a 25% difference in exacerbations within patients between 2019 and 2020 then 160 participants would be required (alpha=0.05, power 80%). 160 patients were recruited with baseline characteristics outlined in table 1. 140 (88%) reported at least one exacerbation in the previous year, and the majority 103 (64%) reported at least two. 149 (93%) patients were prescribed triple inhaled therapy and 138 (86%) were classed as gold stage 3 or 4 airflow obstruction. in the two weeks prior to lockdown 131 (83%) participants reported using their maintenance inhalers with the same frequency as they would during their stable state. 23 participants (14%) reporting increased use and 4 (2.5%) using less frequently than normal. during the lockdown period 42 (26%) participants reported increased use, 113 (71%) participants reported the same frequency of use and 4 (2.5%) reported using their regular inhaler less frequently than baseline (p<0.001) (figure 2a). anxiety 45 (28%) participants reported having more anxiety about their copd than normal during the pre-lockdown period compared to baseline, of which 42/45 (93%) reported anxiety as a "little more" than baseline and 3/45 (7%) as "much more". during the lockdown period 92 (58%) reported increased anxiety compared to normal (p<0.001), of which 48/92 (52%) were "a little more" anxious and 42/92 (48%) "much more" anxious (figure 2b). participants were also asked if they would avoid coming to hospital as an emergency during the pre-lockdown and lockdown periods due to fear of covid-19. 64 (40%) reported they would have avoided doing so during the pre-lockdown and 90 (56%) reported they would avoid emergency hospital attendance during the lockdown period (p<0.001). 83 (54%) reported physical activity was unchanged compared to normal during the prelockdown period with 63 (40%) reporting reduced activity and 11 (7%) reporting increased activity levels. this contrasted sharply to the lockdown period where only 26 (16%) reported maintaining the same level of activity as normal while 52 (32.5%) reported slightly less and 78 (49%) reported doing a lot less physical activity than normal implying a significant decrease in activity levels (p<0.001). only 4 (2%) reported increased physical activity levels ( figure 2c) . when asked about participation in a home exercise program, 50 (31%) patients and 56 (35%) patients reported participating in a home exercise program during the prelockdown and lockdown periods respectively. [figure2] participants were asked about shopping behaviour during the pre-lockdown and lockdown periods with a significant change being noted; during the pre-lockdown 89 (55.6%) reported going shopping themselves, while 33 (20.6%) reported that this was performed by someone who lives in the house with them and 38 (23.8%) reported it being completed by someone who does not live with them or being delivered to them. in contrast to this, during the lockdown only 11 (6.9%) reported that they still did their own shopping, with 37 (23.1%) having this task completed by someone living in their home and 112 (70%) reporting that it was done by someone who does not live in their home or delivered (pre-lockdown to lockdown, p<0.001). in the pre-lockdown period 142 (88.8%) participants reported continuing normal behaviour with only 16 (10%) shielding. once lockdown started only 7 (4.4%) reported continuing normal behaviour while 127 (79.9%) reported that they were shielding (p<0.001). during the pre-lockdown period 146 (91.3%) reported that they had visitors to their home compared to 31 (19.4%) during the lockdown (p<0.001). in this observational study a 38% increase in community managed exacerbation events during the covid-19 lockdown in 2020 was seen compared to the same six-week period in 2019, as measured by primary care prescription records. the number of patients suffering an exacerbation was unchanged. self-reported anxiety and inhaler adherence increased whereas pa was lower initially during the pre-lockdown period, but most pronounced during lockdown. severe exacerbations, as measured by hospital admissions, were seen within the cohort and represented 6% of all exacerbations. we observed a 50% decrease in hospital managed aecopd events during the covid-19 lockdown compared the same dates in 2019, though our study was insufficiently powered. a recent larger study, comparing hospital events, rather than individual patients, confirm our observations with a similar reduction in aecopd admission rates 8 . this may represent an effect of the increased use of rescue medications in the community resulting in reduced need for hospital admissions, though other factors are also likely to have played a role. this is the first report of the impact of the sars-cov-2 pandemic (and consequent societal lockdown) on objectively measured aecopd rates. our findings contrasts to reports of reduced aecopd events during the lockdown with physicians and copd nursing teams 7 . interestingly we did not observe an increase in the proportion of patients requiring rescue medication, but an increase in the number of multiple courses. possible explanations for these findings may result from either biological or behavioural differences. patients who would normally have been admitted to hospital with an exacerbation may have been managed in the community during the pandemic because of a combination of fearfulness on the part of the patient about transmission risk in hospital and a desire on the part of healthcare teams to reserve hospital bed capacity to manage patients suffering with covid-19 pneumonia. this behavioural explanation appears plausible, particularly as there was increased access to healthcare services via telephone consultations and reduced physical access to clinicians 15 . national guidance, updated in 2018, recommended an "action plan" which includes oral corticosteroids and antibiotics to be self-administered in the event of an aecopd 16 . increased concerns on the part of clinicians about the risks of hospitalisation in a patient population perceived to be at greater risk from sars-cov-2 might have lowered thresholds for prescribing action plans. patient concern that access to primary or secondary healthcare teams and pharmacies might be restricted might also have resulted in stockpiling behaviour during the pandemic 17 with patients potentially requesting multiple "rescue packs" to store in case they were unable to obtain these later. during remote telephone consultations clinicians may have felt unable to assess the severity of patients' symptoms resulting in a reduced threshold for prescribing acute rescue therapies. these behavioural aspects may provide cautions beyond the current pandemic in how future digital health services and clinics are arranged and incentivised. it is possible that the biological triggers for exacerbation events reduced for some patients because of reduced respiratory virus transmission and air pollution during lockdown. our study did not explore the underlying mechanisms of each exacerbation, which may be altered during the sars-cov-2 pandemic. it is possible that events could be driven by noninflammatory causes, termed "pauci-inflammatory" 18 , which are less responsive to oral antibiotics or corticosteroids 19 . we observed that the majority of participants reported increasing anxiety about their copd, particularly during the lockdown period. this would support a view that the observed increase in exacerbations may be underpinned by behavioural change and concerns around copd and emergency healthcare. it also highlights the need for potential psychological support in a vulnerable population, where anxiety and depression are common 20 . in addition, it is clear from our data that adherence to shielding advice was widespread, likely reflecting a shared concern among patients about risks from covid-19. likewise, we observed an increase in self-reported inhaler compliance suggesting greater health concern and vigilance. we also observed a greater dependence on others for day to day activities such as shopping and an overall reduction in physical activity among this cohort of patients with copd that contrasts to that reported amongst the general population during the lockdown 21 . previous studies have demonstrated the association of reduced physical activity levels and aecopd readmission risk 22 and while this study did not directly assess this effect it raises the additional possibility that exacerbation events increased because of increased breathlessness and reduced resilience due to deconditioning and sarcopenia 23, 24 . the longer term consequences of such altered activity behaviours remains to be seen but is of significant concern given the difficulty in providing timely and effective pulmonary rehabilitation in the context of the pandemic 25 . at the time of writing, we are approaching winter in the northern hemisphere and no sars-cov-2 vaccine has yet been demonstrated to be safe and effective 26 , exacerbations of copd are likely to increase with this season and result in increased hospitalisation. this study is a timely reminder that increased understanding of community prescribing practice and patient behaviour are important and may reveal effective tools in reducing morbidity and mortality in this group. firstly, patients with copd are going to require ongoing support and treatment, even if they are less likely to present to specialist or hospital services. previous evidence has shown that pandemic influenza poses a significant risk to patients with copd 27 with the consequence that viral pandemics such as sars-cov-2 are likely to pose a similar risk. developing robust and accessible systems to acutely review patients with copd remotely to guide them in their use of rescue and preventer medication may reduce symptom burden, hospital admissions and unnecessary courses of potentially harmful oral corticosteroids and antibiotics. it is less likely that the increased number of moderate exacerbations recorded from our prescription data represent an increase in airway inflammation but rather a composite of increased anxiety and caution with the aim of preventing hospital admissions and the consequence that other, non-pharmacological, interventions may have been effective in managing these events 27 . the conclusions drawn from this study are limited by both the relatively small sample size and the severity of the copd seen in the cohort recruited. though 160 patients has provided adequate power for statistically significant differences in community treated exacerbation and behavioural changes it has not been sufficiently large to detect changes in hospitalised events which would be better evaluated using larger datasets. in addition to this the cohort had established copd, under a specialist secondary care clinic, so results may not be applicable to those with milder disease, and less frequent exacerbations. adding further selection bias, patients recruited needed to be alive during the period of recruitment in may and june 2020, meaning that there may be survivor bias compared to those that died in 2019 and during the peak of the pandemic. our use of a survey to assess associated factors and explore possible causes for patterns seen was notably limited by recollection bias with questionnaire calls taking place up to seven weeks after the end of the period of interest and by our use of non-validated questionnaires. in summary, this study revealed an increase in treatment for community treated aecopd events among patients with severe copd during the sar-cov-2 lockdown. this finding was unexpected but may be explained by factors such as anxiety, which was increased in our patient cohort. significant behaviour changes including reduced physical activity, adherence to shielding advice and increased inhaler compliance. 957-63. 2. gold. global strategy for the diagnosis, management and prevention of chronic obstructive pulmonary disease covid-19 rapid guideline: community-based care of patients with chronic obstructive pulmonary disease (copd) nice guideline factors associated with covid-19-related death using opensafely covid-19 and copd severity and mortality associated with copd and smoking in patients with covid-19: a rapid systematic review and meta-analysis we continue to see very few #copd exacerbations, and not just because people aren't coming to hospital. i'd often wondered what 'intrinsic' exacerbation frequency might look like with cleaner air and much reduced transmission of (regular) viruses. the art of the possible disproportionate decline in admissions for exacerbated copd during the covid-19 pandemic acute exacerbations of chronic obstructive pulmonary disease: in search of diagnostic biomarkers and treatable traits caring for patients with copd and covid-19: a viewpoint to spark discussion mental health and the covid-19 pandemic comorbidity and its impact on 1590 patients with covid-19 in china: a nationwide analysis comprehensive respiratory assessment in advanced copd: a 'campus to clinic' translational framework standardisation of spirometry digital technologies in the public-health response to covid-19 chronic obstructive pulmonary disease: diagnosis and management: summary of updated nice guidance beyond panic buying: consumption displacement and covid-19 acute exacerbations of chronic obstructive pulmonary disease: identification of biologic clusters and their biomarkers examining the relationship between anxiety and depression and exacerbations of copd which result in hospital admission: a systematic review comprehensive pulmonary rehabilitation for anxiety and depression in adults with chronic obstructive pulmonary disease: systematic review and meta-analysis is the covid-19 lockdown nudging people to be more active: a big data analysis risk factors of readmission to hospital for a copd exacerbation: a prospective study bedside assessment of quadriceps muscle using ultrasound following admission for acute exacerbations of chronic respiratory disease. american journal of respiratory and critical care medicine gait speed and readmission following hospitalisation for acute exacerbations of copd: a prospective study pulmonary rehabilitation at a time of social distancing: prime time for tele-rehabilitation? the role of viral infections in exacerbations of chronic obstructive pulmonary disease and asthma have you had any periods of worsening of your breathing symptoms since the 1 st of march this year? if no -go to section b. if yes -go to question 2 have you had to use additional medication for these episodes? if yes, what was this medication? a. salbutamol (blue) inhaler? b. steroids? if so how many five day courses and on what dates? c. antibiotics? if so how many courses and on what dates? have you attended hospital for any of these episodes? in no -go to section b. if yes -go to question 5 if it weren't for covid do you think you have gone to hospital? were you diagnosed with an exacerbation of copd? 6. were you diagnosed with covid-19? if current or ex-smoker: estimate pack years 4. how many times did your copd symptoms worsen in the past 12 months sufficiently that you needed extra treatment (steroids or antibiotics) at home or at hospital? (never / once / two or more times) all questions need two answers, one relating to the pre-lockdown and one to the lockdown periods. 1. during the pre-lockdown and lockdown periods did you? a. normal behaviour b. shielding (not leaving your home or interacting face to face avoiding close contact (2 meters/8 feet) with anyone who does not live in your home) during the pre-lockdown and lockdown periods who was living in your household? a. lived alone b. lived with a partner/spouse c. lived with working age children d. lived with (pre)school age children e. lived in an institutionalised setting during the pre-lockdown and lockdown periods did you have any visitors to your home? y/n if yes were they: a. adults b. (pre)school age children during the pre-lockdown and lockdown periods how did you obtain essential items and groceries? a. self b. another person who lives in your home c. deliveries/ someone who does not live in your home during the pre-lockdown and lockdown periods did you start any new medications? y/n if yes during the pre-lockdown and lockdown periods did you use your regular (non-salbutamol) inhalers more or less regularly? (less) during the pre-lockdown and lockdown periods did during the pre-lockdown and lockdown periods were you more or less active than normal? (less)1 -2 -3 -4 -5 (more) during the pre-lockdown and lockdown periods did you feel more or less anxious about your copd? (less) during the pre-lockdown and lockdown periods did you avoid coming to hospital as an emergency due to fear of covid 19 we wish to acknowledge the work of the following members of the nihr leicester brc key: cord-268524-lr51ubz5 authors: droit-volet, sylvie; gil, sandrine; martinelli, natalia; andant, nicolas; clinchamps, maélys; parreira, lénise; rouffiac, karine; dambrun, michael; huguet, pascal; dubuis, benoît; pereira, bruno; bouillon, jean-baptiste; dutheil, frédéric title: time and covid-19 stress in the lockdown situation: time free, «dying» of boredom and sadness date: 2020-08-10 journal: plos one doi: 10.1371/journal.pone.0236465 sha: doc_id: 268524 cord_uid: lr51ubz5 a lockdown of people has been used as an efficient public health measure to fight against the exponential spread of the coronavirus disease (covid-19) and allows the health system to manage the number of patients. the aim of this study (clinicaltrials.gov nct 0430818) was to evaluate the impact of both perceived stress aroused by covid-19 and of emotions triggered by the lockdown situation on the individual experience of time. a large sample of the french population responded to a survey on their experience of the passage of time during the lockdown compared to before the lockdown. the perceived stress resulting from covid-19 and stress at work and home were also assessed, as were the emotions felt. the results showed that people have experienced a slowing down of time during the lockdown. this time experience was not explained by the levels of perceived stress or anxiety, although these were considerable, but rather by the increase in boredom and sadness felt in the lockdown situation. the increased anger and fear of death only explained a small part of variance in the time judgment. the conscious experience of time therefore reflected the psychological difficulties experienced during lockdown and was not related to their perceived level of stress or anxiety. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 in 2020, faced with a virus that is uncontrollable because of its unknown [1] and virulent nature (sars-cov-2), the governments of different countries of the european union, as well as of the whole world, found themselves obliged to impose a lockdown on their citizens. this unprecedented public measure is thought to allow the health system to manage the number of patients in hospital and ensure that they receive proper care in the context of the covid-19 outbreak. in france, confinement was officially imposed in the month of march (on march 17 th at 12:00 noon). this lockdown, which requires a large number of people to stay at home, thus depriving them of their liberty, is a situation never previously encountered and its psychological consequences in the short and medium term are not yet known. researchers into time perception can nevertheless easily imagine that this life in lockdown completely changes individuals' relationship to time, i.e. their experience of time. however, to our knowledge, no studies have as yet investigated this question. very recent scale surveys or survey projects on covid-19 conducted all around the word (e.g., china, korea, iran and united kingdom) suggest that the lockdown situation generates new or heightened emotional states in the form of an increase in psychological distress [2] [3] [4] [5] [6] . nonetheless, in the different distress scales used, the different dimensions of emotion (valence and arousal) were not dissociated, and no survey has examined their relationships to time experience, even though emotion and the experience of time are known to be intrinsically linked. the aim of the present study was thus to conduct a scale survey on a large sample of an as yet untested population-french people-in order to assess not only the perceived stress related to covid-19 but also the emotions (happiness, boredom, arousal) felt during as compared to before the lockdown and their links to the subjective experience of time. the experience of time corresponds to one's feeling about time, i.e., the conscious judgment of the speed of the passage of time [7, 8] . this has received relatively little attention by researchers in the field when compared to research into individuals' abilities to perceive short durations (< 1 minute). this is probably due to the challenge of objectively examining just what makes up the experience of each individual, and therefore the role of higher-level cognitive mechanisms (e.g., consciousness, memory, self-awareness) [9] [10] [11] . indeed, the judgment of the passage of time can be seen as a mirror of the subjective experience of one's internal state [12] [13] [14] . for example, contrary to the generally held belief that time seems to pass faster as we get older, some studies have demonstrated that the feeling of the passage of time in the immediate moment is not directly related to age (young adult vs. older adult), but to people's subjective emotional experience and lived activities [10, 15, 16] . the passage of time is in fact a sensitive index of emotional experience felt in the present moment and of its variations as a function of life conditions. it is thus important to investigate individuals' judgments about how fast time seems to pass in the exceptional situation of lockdown and the factors explaining these. from a general standpoint, the literature provides evidence of the role of emotional experience as a critical factor in the experience of time. nevertheless, the famous expression "time flies when you feel good; time drags when you feel bad" is not straightforward to explain, as negative feelings are diverse and may involve varying mechanisms. more precisely, the emotional experience can be divided into two fundamental dimensions, valence (pleasure vs. displeasure) and activation (calmness vs. excitement/alertness) [17, 18] . these two dimensions interact in the characterization of any given emotion. for example, while the emotions of sadness and fear are both negative, the former is weakly activating (or even deactivating) while the latter is strongly activating. accordingly, the level of felt arousal has been shown to be a prominent factor in temporal mechanisms: the more individuals report being in a state of arousal, the faster time is reported to pass. several studies have shown a lengthening of estimates of short temporal intervals in situations of acute stress, for example when participants are faced with unpleasant stimuli [19] [20] [21] or when they imminently expect a very unpleasant event, e.g., electric shock [22, 23] . however, few studies have examined the effect of chronic stress on time judgments, such as that experienced by people with the covid-19 virus or subjected to lockdown. in the context of chronic stress, i.e. when stress is extended over several days or weeks as in the case of hospital nurses, cocenas-silva et al. [24] showed that duration judgments were no longer altered by physiological stress as measured by physiological markers, but rather by subjective psychological stress as assessed by a self-reported scale. in addition, one can assume that different mechanisms are at work in the case of an emotion, such as fear (an immediate and ephemeral negative state directed towards a specific event), compared to a more diffuse affective state, like anxiety or perceived stress (a prolonged negative state whose origin is not necessarily identified) [25] . the covid-19 pandemic, i.e., the risk that you or your loved ones will be affected by the disease as well as uncertainty about this disease, could produce chronic stress that has consequences for mental and physical health. it is well known that chronic stress affects the immune system, suppressing protective and increasing pathological immune responses [26] . there is thus a risk in this period of pandemic that the chronic stress related to covid-19 and its corollaries (anxiety, fear of death) are particularly high and therefore impact the subjective experience of time by speeding up the perceived passage of time. consequently, we hypothesized a significant relationship between stress and time experience during the lockdown imposed by the covid-19 pandemic. furthermore, in this covid-19 period, it is critical to consider not only the disease-related perceived stress but also the consequences for life of being locked down at home, as well as the direct and indirect effects on daily psychological and social functioning. as a recent survey highlighted, confining people increases their sense of boredom [2] . boredom corresponds to "the aversive state of wanting, but being unable, to engage in satisfying activity" and involves, in particular, low arousal, negative affects [27, p 483 ]. in particular, some studies have shown that boredom produces a feeling of the slowing down of time rather than a speeding up [14, 28] . an alternative hypothesis was thus that boredom would prevail over stress in the experience of time. since boredom is associated with negative emotion of low level of arousal, we thus expected participants to experience of slowing down of time with the boredom experienced during the lockdown. it was not possible a priori to identify which hypothesis would be valid, i.e., which are the factors related to and influencing the experience of time in a lockdown situation, the perceived stress in the stressful situation of covid-19 and/or-by contrast-other affective states characterized by a decrease in arousal such as boredom. indeed, on one hand, the fear and distress generated by the morbid nature of the crisis and its repercussions (fear for one's health and for that of one's family and friends) or by inappropriate housing quality (stress at home) or working conditions (job stress) could increase people's sense of alertness, and therefore lead to a speeding of the passage of time. on the other, confinement at home and social distancing could result in an increased sense of sadness (i.e., less happiness) and boredom, and thus in the feeling that the passage of time slows down. here, a large sample of french people were asked to answer a scale survey during the lockdown period. this consisted of a series of questions, i.e., demographic questions but also questions on the stress perceived (covid-19 stress, home stress, job stress, anxiety), the emotions (happiness, arousal, boredom) felt during compared to before the lockdown and the experience of time. the participants were asked to assess their experience of the passage of time according to three periods of the lockdown: in the immediate moment, during the day, during the last week, as well as before the lockdown for comparison purposes. the sample consisted of 4364 french participants, 3436 women and 928 men (mean age = 41.5, sd = 12.81, min = 16, maxi = 89, n 16-17 years = 11). the participants completed the questionnaire at home (72.5%) or at work (27.5%). the study was reviewed and approved by the human ethics committees sud est vi, france (clinicaltrials.gov nct 04308187). all participants were volunteers and were informed of the objective of the survey and that their data would be processed anonymously and be used for research purposes. the ethics committee waived the need for written consent considering that if people respond to the questionnaires by going to the website, they are giving their consent. furthermore, they can withdraw it at any time. the few minors who completed the questionnaire did so with the consent of their parents who sent them the survey. the responses to the demographic questions allowed us to characterize the surveyed population. 71.8% of participants were married or equivalent (civil partner, etc.) and 27.2% were single (1% other). their distribution as a function of education level was: 1.5% certificate of general education, 21.9% high school vocational certificate, 0% high school diploma, 40.6% bachelor's degree, 24.5% master's degree and 11% doctoral degree. the percentage of participants per professional category was: jobseekers: 4.4%; students: 6.2%; farmers: 0.3%; craftsmen/shopkeepers/business executives: 5.7%; white-collar workers: 30%; manual workers: 8.9%; intermediate professions, 35.7%; retired: 6.3% (2.5% no response). we implemented an open epidemiological, observational, descriptive study by administering a self-reported questionnaire proposed to volunteers using redcap 1 software available through the covistress.org website. the redcap 1 questionnaire was hosted by the university hospital of clermont-ferrand. the questions analyzed in this manuscript were therefore specific questions included in a large questionnaire composed of different thematic sections of questions (s1 questions). the thematic sections were presented in random order after the demographic questions. the online questionnaire was distributed several times through mailing lists held by institutions and french social groups. there were no exclusion criteria. the data that we analyzed were obtained for the period of lockdown from march 31 th to april 12 th , 2020, whereas the french lockdown was ordered on march 17 th at 12:00 noon. the time taken to complete the survey lasted between 5 and 20 minutes on average, depending on sub-items. for the main outcomes, we used a visual analog scale (vas), i.e., a non-calibrated line of 100 mm, ranging from 0 to 100 [29, 30] . the subjective experience of time was thus assessed using this vas, which went from very slowly (0) to very fast (100). the question was "what are your feelings about the speed of the passage of time". there were four time questions, one for the passage of time before the lockdown, and three for during the lockdown: now, for the day, and for the week. the stress resulting from covid-19 as well as job stress and home stress, health-related and financial concerns and anxiety were assessed using the same vas. the emotional dimensions tested were also assessed with the vas for the period before the lockdown and during the lockdown (now): fear of death (not at all vs. at lot), arousal (calm vs. excited), happiness (sad vs. happy), anger (peaceful vs. angry), boredom (occupied vs. bored). the quality of sleep and level of fatigue were also examined in the survey using the vas. as explained above, these different questions were presented in different thematic sections presented in a random order (s1 questions). we performed analyses of variance on the subjective experience of time. we also examined correlations and ran a linear regression model on all the measures of interest by using the standardized data. we used the variance inflation factor (vif) to examine the multicollinearity in the regression analysis [31] . finally, to examine the results of the linear regression model in more detail, we also performed an analysis of mediation. the analyses were performed with spss and the bonferroni correction was systematically applied when necessary. a preliminary analysis of variance performed on the subjective experience of time showed a marked difference between the experience of time before and during the lockdown (fig 1) . that time passed faster when a longer period of time was considered, i.e., a week compared to a day or the present moment (bonferroni comparisons, p < 0.01). to simplify the results, the subsequent statistical analyses are based on the difference in time ratings for the question on the period before the lockdown and that for the present moment (during the lockdown). indeed, the meaning of temporal judgment during the lockdown is relative to that before the lockdown. in addition, the results were similar when the analyses were only performed on the ratings for the present moment. a positive value of our temporal difference index therefore indicates that the individuals experience a slowing down of time during the lockdown, a negative value a speeding up of time and a null value no difference. the anova performed on this temporal difference index, with level of education, professional category and whether the individuals were at work or home as factors, did not show any significant effect (all f < 1). there was indeed no significant difference in time experience before the lockdown situation as a function of these factors. only a small effect of professional category was observed in the present time judgment during the lockdown, f the anova on the temporal index with sex and marital status (single vs. not single) as factors showed a significant main effect of sex, f(1, 4084) = 14.77, p < 0.001, η 2 p = .004, and status, f(1, 4084) = 11.74, p < 0.001, η 2 p = .003, with no sex x status interaction (p > 0.10). this suggests that the single people in our sample tended to experience a greater difference in the flow of time during the lockdown when compared to before (29.19 vs. 24.19) . indeed, in the lockdown situation, time in the present was judged to pass slower by the single people (m = 47.12, sd = 30.12) than by the others (m = 53.93, sd = 29.15). the women also tended to feel a greater slowing down of time than the men (29.41 vs. 23.89) during as compared to before the lockdown, but time passed faster for the women than for the men before the lockdown (80.51 vs. 74.37), f(1, 4084) = 71.11, p < 0.001, η 2 p = .02. nevertheless, their responses to the stress questions indicated that they tended to be more stressed than the men, even though the sex difference only explained a very small proportion of variance ( table 1 shows the correlation matrix (s1 table) between the subjective experience of time (difference in the judgment of the passage of time between before the lockdown and the present moment, i.e., during the lockdown) and the different tested factors. an examination of table 1 reveals that several dimensions were associated with the slowing down of time during as compared to before the lockdown. with regard to stress, the participants experienced that time passed slower-rather than faster-with an increase in the level of perceived stress, i.e., the perceived stress related to covid-19 (r = .18) as well as the stress at home (r = .23) and at work (r = .08). a slowing down of time was therefore observed as the stress level increased. this deceleration of subjective time was observed even if the stress value reported on the vas was high, and higher for covid-19-related stress than for home and job stress (covid-19 stress, m = 61.50, sd = 28.87; job stress, m = 57.94, sd = 32.65; home stress, m = 46.97, sd = 32.65, f(2, 7466) = 342.78, p < 0.001, η 2 p = .08 (all bonferroni tests, p < 0.001). the rating for each type of stress was indeed significantly different from zero (t(4196) = 138.18, t(4184) = 93.06, t(3892) = 10.13, respectively, all p < 0.001). finally, the stress resulting from covid-19 was more closely associated with anxiety (r = .75, p < 0.001), the fear of death (r = -.42, p < 0.001) than it was with the experienced time per se. inconsistently with our first hypothesis, the level of correlation between the experience of time and covid-19-related stress was therefore very low, and this was also the case for stress in the other contexts (home, work). as suggests table 1 , the experience of time was more correlated with boredom (r = -.48, p < 0.001) and decreased happiness (r = .39, p < .0001) than with the level of perceived stress. therefore, the participants experienced a slowing down of time as boredom increased and happiness decreased during the lockdown. as the time judgment was significantly correlated with several dimensions, to identify the best predictor of the subjective experience of time we performed a regression analysis on the time judgments with the different significant dimensions entered into the same model ( table 2 ). the examination of multicollinearity in the regression analysis using the vif indicated no problematic presence of multicollinearity (all vif < 3) [31] . the results of this regression analysis indicated that the perceived stress resulting from covid-19 and its spread was not a table 1 . correlations between the passage of time (difference between before the lockdown and for the present, i.e., during the lockdown) and the different tested factors (z-scores). participants were in the lockdown situation, the more they experienced a slowing down of time. indeed, time was experienced as passing increasingly slowly in the present moment compared to before the lockdown as the level of boredom rose (fig 2) . it also seemed to slow down as happiness decreased, i.e., as sadness increased (fig 3) . increasing boredom and decreasing happiness were therefore the two main predictors of the experience of the passage of time during the lockdown. since these two dimensions are related, we conducted statistical analyses to estimate whether the boredom mediated the effect of emotion on the experience of time and, conversely, whether emotion mediated the effect of the boredom of the experience of time. the mediation analyses indicated that boredom contributes to explaining the effect of emotion on the experience of the passage of time, with a significant indirect effect of 0.159 (β), se = .01, 95% ci (.138; .1812), z = 14.7, p < 0.001, 34.4% of mediation) (fig 4) . however, the direct effect of emotion (sadness) on the time experience remained significant (β = . the results of our survey showed that the stress felt by a broad cross-section of the french population during the lockdown was high, in particular with regard to stress relating to the covid-19 pandemic, as is indicated by the rating of 61.50 (+/-28.87) on a 100-mm vas. the level of perceived stress linked to covid-19 was even higher than the stress at work and at home. covid-19 stress was, in fact, related to the participants' anxiety and their fear of death. the more anxious and frightened they were about death, the more stressed they were in the face of this disease. these results are entirely consistent with the initial results of surveys on covid-19 conducted, in particular, in china [5, 6] and iran [4] , which have shown an increase in psychological distress as a result of the covid-19 pandemic. however, as reported by qui et al. [5] , it is noteworthy that people's distress does not reach a pathological level (m = 23.65), with only 5% of the population suffering from severe distress and 29% from mild or moderate distress. in addition, the proportion of individuals presenting psychological distress disorders before the covid-19 is unknown. however, the chinese suffer less psychological distress and have greater life satisfaction when working in the office than at home, whereas the opposite seems to be the case in the french population, as suggested by the significantly lower level of stress at home than at work. this suggests that there are some differences in culture or living conditions between people in different countries with regard to stress management in similar social isolation situations. the originality of our results is to show that, although the level of stress was quite high, it had little impact on the current subjective experience of time. indeed, the participants did not feel a speeding up of time related to the increase in their stress level. this is contrary to the results of studies on timing which have described a lengthening of duration estimates and the experience of a faster passage of time when the levels of stress and anxiety are high [21, 32, 33] . however, these findings were obtained in intense and concisely emotional situations, when the subjects were faced or expecting a forthcoming threatening event, or in individuals with high-anxiety traits. in the situation of lockdown at home, the current level of stress was therefore not high enough to affect the sense of time. indeed, the level of arousal remained low, although it increased slightly between the period before and during the lockdown. to conclude, one might nevertheless think that it would have been more convincing to record the physiological markers of stress. however, this was not possible in the lockdown situation which was rapidly decided on by the public authorities [34, 35] . in addition, cocenas et al. [24] recently showed that perceived stress was a better predictor of changes in time estimates than physiological stress per se in the case of prolonged stressful situations, for example in the case of hospital nurses at work. in addition, the likelihood of encountering a series of intensely stressful events may be reduced in the present isolation situation. family life involving the care of children can obviously be a source of stress. our study did indeed indicate that women were more stressed at home than men, but were even more so when they were single than part of a family, and that the number of children only slightly increased the stress level at home (r = .08, p < 0.001). rather than covid-19-related stress or home and job stress, our study showed that it was the emotional experience of everyday life during the lockdown that influenced the sense of time. indeed, the participants clearly reported experiencing a slowing down of the passage of time during in comparison to before the lockdown. and the most reliable predictors of this slowing down were the feelings of boredom and sadness. our results are consistent with those of recent studies on time judgments that have pointed out the critical role of emotion in human beings' sense of time [for a review 35] and of boredom [14, 28, 36] . these studies have indeed found a slowing down of time as both sadness and boredom increase. in line with theoretical models of boredom [27] , the present study found that the degree of boredom experienced was related not only to arousal but mostly to negative emotional experience: the more bored people were in lockdown, the sadder they were. the boredom is known to be linked to depression [37, 38] , and depressed people feel a slowing down of time [39] . consequently, the experience of boredom in the lockdown and the judgment of a slower passage of time have increased sadness and could lead to pathological depression. however, in the lockdown situation, the level of boredom explained a proportion, but not all, of the effect of sadness on the experience of the passage of time. other factors that we need to examine in a future study could also help to explain sadness and time experience in the lockdown, such as social withdrawal. the changes in the sense of time in lockdown were therefore due to the significant increase in both boredom and sadness. the literature on boredom suggests that it is involved in a multitude of behaviors and psychological dimensions and that it has a negative side, as in the sadness observed in our study, as well as a positive side. indeed, trait boredom is associated with psychological difficulties (e.g., drug abuse, depression, anxiety, binge eating) [40, 41] . however, some recent functional approaches have also suggested that boredom constitutes a key signal to change behavior by orientating humans to try to find a more satisfying situation [42] . in the context of lockdown, one may therefore wonder what influence this feeling of boredom has on the development of pro-social behaviors or on compliance with the containment situation in the short or longer term (does it only result in bad things or also in good things?). in the lockdown situation, people may have more time. however, they "die" of boredom and sadness and time slows down, drags on. the sense of the passage of time is, ultimately, a phenomenological time that is closely related to the self and the sense of existence [13] . as stated by jean-paul sartre, human beings are defined by their acts and their effects on others. however, when they have more time but are isolated and cannot act-they have nothing to do-they are overwhelmed by sadness and boredom. it would seem important for future surveys to examine whether this feeling is valid in all cultures and for all people. it also seems to be important to identify whether other factors specific to individual characteristics or living conditions, to representations/beliefs toward covid-19 or government policies contribute to changes in the sense of time in the lockdown situation. some authors nevertheless defend the benefits of boredom. however, this raises the question of individual abilities to cope with the feeling of boredom in industrial societies. individual differences in coping with boredom can potentially predict psychological difficulties, health problems and increased vulnerability to psychopathologies such as depression [43] . it is thus a serious problem and one which has to be taken into account. in conclusion, the changes in the sense of time in the lockdown situation, imposed as an efficient solution to the covid-19 pandemic, reflect the major psychological difficulties that people are experiencing during the lockdown. (docx) s1 table. table of members of the research group are nicolas andant, maélys clinchamps china; peter dieckmann -copenhagen academy for medical education and simulation (cames), denmark how will country-based mitigation measures influence the course of the covid-19 epidemic? the psychological impact of quarantine and how to reduce it: rapid review of the evidence multidisciplinary research priorities for the covid-19 pandemic: a call for action for mental health science the distress of iranian adults during the covid-19 pandemic-more distressed than the chinese and with different predictors. medrxiv a nationwide survey of psychological distress among chinese people in the covid-19 epidemic: implications and policy recommendations unprecedented disruption of lives and work: health, distress and life satisfaction of working adults in china one month into the covid-19 outbreak passage of time judgements intertwined facets of subjective time passage of time judgments in everyday life are not related to duration judgments except for long durations of several minutes passage of time judgments are not duration judgments: evidence from a study using experience sampling methodology what day is today? a social-psychological investigation into the process of time orientation mindfulness meditation, time judgment and time experience: importance of the time scale considered (seconds or minutes) awareness of the passage of time and self-consciousness: what do meditators report? psych journal individual differences in self-rated impulsivity modulate the estimation of time in a real waiting situation time does not fly but slow down in old age experience sampling methodology reveals similarities in the experience of passage of time in young and elderly adults a circumplex model of affect core affect, prototypical emotional episodes, and other things called emotion: dissecting the elephant the effect of expectancy of a threatening event on time perception in human adults time estimation of fear cues in human observers negative emotionality influences the effects of emotion on time perception fear and time: fear speeds up the internal clock emotional modulation of interval timing and time perception chronic stress impairs temporal memory. timing time percept anxiety makes time pass quicker while fear has no effect effects of stress on immune function: the good, the bad, and the beautiful the unengaged mind: defining boredom in terms of attention what happens while waiting? how self-regulation affects boredom and subjective time during a real waiting situation clinical stress assessment using a visual analogue scale validity of occupational stress assessment using a visual analogue scale extracting the variance inflation factor and other multicollinearity diagnostics from typical regression results when time slows down: the influence of threat on time perception in anxiety the effects of valence and arousal on time perception in individuals with social anxiety jobstress study: comparison of heart rate variability33. in emergency physicians working a 24-hour shift or a 14-hour night shift-a randomized trial urinary interleukin-8 is a biomarker of stress in emergency physicians, especially with advancing age-the jobstress* randomized trial the temporal dynamic of emotional effect on judgments of durations proneness to boredom mediates relationships between problematic smartphone use with depression and anxiety severity relationships between boredom proneness, mindfulness, anxiety, depression, and substance use time perception in depression: a meta-analysis time flies when you're having fun: temporal estimation and the experience of boredom psychometric measures of boredom: a review of the literature high boredom proneness and low trait self-control impair adherence to social distancing guidelines during the covid-19 pandemic intrinsic enjoyment and boredom coping scale: validation with personality, evoked potential and attention measures the covistress network is headed by pr. frédéric dutheil (frederic.dutheil@uca.fr) chu key: cord-253910-pmurx4jh authors: miles, david; stedman, mike; heald, adrian h title: “stay at home, protect the national health service, save lives”: a cost benefit analysis of the lockdown in the united kingdom date: 2020-08-13 journal: int j clin pract doi: 10.1111/ijcp.13674 sha: doc_id: 253910 cord_uid: pmurx4jh introduction: the covid‐19 pandemic has transformed lives across the world. in the uk, a public health driven policy of population ‘lockdown’ has had enormous personal and economic impact. methods: we compare uk response and outcomes with european countries of similar income and healthcare resources. we calibrate estimates of the economic costs as different % loss in gross domestic product (gdp) against possible benefits of avoiding life years lost, for different scenarios where current covid‐19 mortality and comorbidity rates were used to calculate the loss in life expectancy and adjusted for their levels of poor health and quality of life. we then apply a quality‐adjusted life years (qaly) value of £30,000 (maximum under national guidelines). results: there was a rapid spread of cases and significant variation both in severity and timing of both implementation and subsequent reductions in social restrictions. there was less variation in the trajectory of mortality rates and excess deaths, which have fallen across all countries during may/june 2020. the average age at death and life expectancy loss for non‐covid‐19 was 79.1 and 11.4 years respectively while covid‐19 were 80.4 and 10.1 years; including adjustments for life‐shortening comorbidities and quality of life plausibly reduces this to around 5 qaly lost for each covid‐19 death. the lowest estimate for lockdown costs incurred was 40% higher than highest benefits from avoiding the worst mortality case scenario at full life expectancy tariff and in more realistic estimations they were over 5 times higher. future scenarios showed in the best case a qaly value of £220k (7xnice guideline) and in the worst‐case £3.7m (125xnice guideline) was needed to justify the continuation of lockdown. conclusion: this suggests that the costs of continuing severe restrictions are so great relative to likely benefits in lives saved that a rapid easing in restrictions is now warranted. severe acute respiratory syndrome coronavirus-2 (sars-cov-2),) also known as covid-19 virus, was spreading alarmingly in early march 2020. the matter of whether the national health service (nhs) would be able to deal with rapidly rising numbers of seriously ill people was unclear. a group at imperial college put the likely level of united kingdom (uk) deaths if there was no change in behaviour at 500,000 (1). the uk government followed the example of several other european countries in with the prime minister announcing severe restrictions on individual movement. (2) the key message was to stay at home to protect the nhs to save lives. this was a lockdown. the precise extent to which the lockdown contributed to a subsequent slowing in the rate of new infections and deaths is not clear, though that it did so to some extent seems very likely. as a highly infectious disease, the government response was to "save lives" through "stay at home" to reduce transmission outside the home, curtailing non-essential recreation, travel and suggesting people work from home. "protect the nhs" reflected concern that the key services such as hospitals, especially intensive care units (icu), would have been overloaded and unable to treat the potentially large number of very ill patients and so there would have been increased mortality. the evidence from the first 3 months suggests that most of these measures worked. however, some quickly became less relevant. the nhs rapidly put in place sufficient surge hospital & icu capacity to cope with the very high level of expected demand, so service overload was no longer an immediate concern. mateen et al (3) showed that nhs converted 2,000 normal acute beds into icu and further 11,000 beds were coming on stream within the 5 additional nightingale units (there were further 1200 beds booked in private icus (4) ). this gave a total potential capacity of over 14,000 icu beds. they also showed at the peak in early april 3,000 icu this article is protected by copyright. all rights reserved beds were being used by covid-19 patients in the context of what turned out to be 60,000 excess deaths, i.e. 1 icu bed to 20 excess deaths. this suggests that the 10,000 surge icu beds gave technically sufficient icu capacity to support a pandemic size resulting in up to 200,000 excess deaths. in the general wards, (3) showed around 12 ,000 beds were occupied by covid-19 patients, which gave a ratio of 1 general and acute (g&a) bed to 5 excess deaths. there were still 30,000 beds free and a further 8,000 beds in private hospitals had been booked (4); these 38,000 g&a beds gave sufficient headroom for a pandemic up to 200,000 excess deaths. the nhs through its own extreme efforts was far from being overwhelmed. while it is clear that the cost of the lockdown has been large, the uk finance minister covered some of these in his summer statement (5) , just how great it is will not be known for many years. this cost -as well as the benefits of lockdowns -should be measured in terms of human welfare in the form of length and quality of lives. such measurement is profoundly difficult. yet measurement of the costs of restrictions needs to be weighed against the benefits of different levels of restrictions to assess what is the best policy now. we use rules that the national institute for clinical excellence use to guide decisions on public health expenditure by the nhs (6) and which implicitly value years of lives saved to assess benefits of the lockdown relative to its likely cost -costs both in narrow economic terms but also in health terms. the cost of severe restrictions plausibly rises more than in proportion to the length of a lockdown. there is some evidence that the benefits of maintaining a lockdown may be diminishing as described in bongaerts et al (2020) (4) . in this article, we aimed to calibrate what the costs and benefits of severe restrictions might be and what that implies about the policy that should now be followed in the uk. bringing together costs and benefits is necessary if good policy decision are to be made. there is no simple way to do this that is clearly ethically justifiable, empirically reliable and widely accepted. however to make no assessment is just to make policy in a vacuum. this article is protected by copyright. all rights reserved we selected 10 european countries with populations over five million and average income/person over £15,000/year and examine the relation between the outcomes of the excess death as reported in human mortality database of the university of california (5) and the different levels of lockdown as calculated and reported as a composite measure based on nine response indicators including school closures, workplace closures, hospitality venue closures and travel bans, in the government response stringency from blavatnik school of government, university of oxford (6) to see if there were any clear relationships. a detailed breakdown by country is described in the results section. another approach is to focus on quality-adjusted life years (qalys) that may have been saved as a result of restrictions that have been in place in the uk up to early june and to convert that to a metric that can be compared with estimates of the cost of the restrictions. to that, we add estimates of the value of health care resources saved (both now and into the future) because lockdown reduced the numbers of sick people. that is the strategy we follow. we then go on to make estimates of costs and benefits of alternative ways forward with restrictions eased to different extents. we make use of the guidelines established in the uk by the national institute for health and care excellence (nice) for the use of resources in the uk health system (see nice (2013) (7)). these are guidelines applied to resource decisions that have a direct impact on lives saved. it is hard to see how you could run a public health care system without such rules. the benefits are measured as the number of quality life-years gained compared to the potential with different levels of lockdown. to establish these total quality life years, we examine the number of deaths directly reported and those excess deaths which were defined as the number of deaths above the average over comparable periods for recent years (10) . we examine the international situation to see if there was any relationship between level of lockdown and levels of excess deaths. we look at the ages of those that died to based on the normal ons evaluate their life expectancy at those and then their comorbidities to establish a quality of life we add to these estimates of the saving of nhs resources from a much-reduced demand upon its resources as lockdown slowed the numbers who became seriously sick. this article is protected by copyright. all rights reserved to implement this, we need to assess how many likely extra years of good life might be enjoyed by the people who would have died but for a lockdown. we assume that the age and health of those who would have died are similar to that of those who have died with the virus. the ons has been publishing each week the number of deaths where covid-19 has been recorded as a possible cause by quinary age and gender. by applying the average life expectancy (11) to the actual recorded covid-19 deaths by age and gender a total life expectancy years loss can be calculated. the average figure for years of life lost does not account for the fact that those who have died with covid-19 have often been in poor health, conditional on their age. we, therefore, examine reports on their actual levels of serious comorbidities and assessed their life-shortening impact. we evaluate the quality of life they might have expected in those remaining years. we report benefits finding against both the full life expectancy loss and the quality of life adjusted loss. the lockdown has reduced the amount of intensive care unit (icu) and general and acute (g&a) ward costs required by the potential covid-19 patients. we evaluate reports on the current levels of use and against the number of deaths that occurred and extrapolated to the higher numbers seen in the scenarios. to estimate the future costs of the lockdown we consider the wider shorter and longerterm economic, social and health effects but feel they would all be reflected within the impact on gross domestic product, and so take the latest measured values from the ons and the latest estimates of future values from the bank of england, office of budget responsibility (obr) and other experts. it seems plausible that a large fraction of these estimates of lost output is due to the lockdown. however, even without a governmentmandated lockdown there would have been some reduction in incomes so only a part of the lockdown effect is incorporated into our model finally we apply the same methodology to evaluate the future policy over the timing and rate of easing of the lockdown. this article is protected by copyright. all rights reserved section 1 -preamble: recorded cases, deaths and excess deaths infections rose dramatically in many european countries between february and march of 2020 and, with some lag, so did deaths attributed to the virus. excess deaths are a more reliable measure of the overall cost in lives of the virus given the policies that were adopted for dealing with it. figure 1 shows how a measure of excess deaths for a group of european countries with similar levels of income and health care provision to the uk. this is total deaths above the average of such deaths over the comparable months in previous years. figure 2 shows a measure of the stringency of government restrictions introduced in european countries to counter the spread of the virus. in late february or early march 2020, many european countries brought in severe restrictions on movement meaning that the majority of populations stayed home and numbers able to work fell dramatically. new cases of the infection and of deaths ascribed to the virus were significantly lower within a few weeks of restrictions being introduced. evaluation of the economic impact of the lockdowns by sector using the number of staff furloughed reported by her majesty's revenue and customs (hmrc) (12) as metric for the reduction in gdp (13) also shows that 60% of the loss in gdp comes from those areas of the economy (real estate, manufacturing, administration, construction, professional services, it, energy and water, financial) that have lower interaction with vulnerable groups while only 40% come from areas that could be seen to have a higher impact with vulnerable groups (retail, accommodation & food services, arts, entertainment, transport, health and social work) suggesting broad-based lockdown may not have been the best use of resources. for the uk the office for budget responsibility (obr) and the bank of england estimate that gdp is likely to have fallen by between 25% and 35% in q2 2020 and by 10-15% in 2020 relative to 2019; unemployment may rise to around 10%. the obr central estimate, and the illustrative scenario for the bank of england made in may 2020, is that in 2020 the uk gdp will be around 13-14% lower than in 2019. the national institute for economic and social research (in its may quarterly report) put the cumulative loss of output in the uk over 10 years at over 30% of annual gdp. this article is protected by copyright. all rights reserved the estimates from the bank of england and the obr assume that easing of restrictions after june 2020 will mean that the lockdown is then soon over; it seems plausible that their estimates of economic cost are therefore estimates of the impact of the lockdown that had been in place in the uk from march to june and not of a continuation of the lockdown into the second half of 2020 and beyond. the obr is explicit about this; in describing their forecasts they note: "the table below summarises the results of our three-month lockdown scenario where economic activity would gradually return to normal over the subsequent three months." the bank of england in its may economic assessment takes a similar line: "underlying the illustrative scenario for both the uk and the rest of the world is an assumption that enforced social distancing measures remain in place until early june and that they are then lifted gradually over the following four months, until the end of q3". in that illustrative scenario, gdp in 2020 is 14% below the 2019 level (table 1a, care between mid-march 2020 and the beginning of june 2020; outpatients seen were 64% down and elective admissions were 75% down (16); attended appointments in general practice were down 35% (17). the impact of the stress of the 'lockdown' on anyone with a pre-existing mental health condition, let alone the population as a whole, is yet to be determined. this was eloquently addressed by kilgore et al in their recent paper (18) which described greater loneliness and elevated depression and higher suicidal ideation in those socially isolating on a standard clinical screening instrument. the observed effect sizes were large, suggesting that social isolation is likely to have a tangible and meaningful impact when considered at the population level. furthermore, the cost from disrupted education of children and students will be felt over a horizon of many years, even decades. this article is protected by copyright. all rights reserved the guidelines in the uk set out by nice are that treatments that are expected to increase life expectancy for a patient by one year (in quality of life adjusted years, qalys) should cost no more than £30,000 (7). we apply that figure to possible total numbers of qalys saved by restrictions to estimate their benefit. table 1 shows the calculation of ons life expectancy lost by age and gender. average life expectancy loss comes out at 10.1 years per covid-19 death. (the average life expectancy years lost for a non-covid-19 death is higher at 11.4 years confirming that the age for covid-19 mortality is slightly older than normal mortality). the median covid-19 age at death is around 80 and the average life years lost for the older 50% is 5 years and for younger 50% is 15 years. in their detailed study of 23,804 hospital deaths in england from covid-19 from 1st march 2020 to 11th may 2020, valabhji et al (2020) (19) found that various life-shortening risk factors were significantly more prevalent in those patients who died of covid-19 than in the general population. this included diabetes (33% vs 5%), and previous hospital admission for significant cardiovascular comorbidities including coronary heart disease (31% vs 3.5%), cerebrovascular disease (19.8% vs 1.5%) and heart failure (17.7% vs 1%). other comorbidities such as dementia in its various forms, chronic obstructive pulmonary disease (copd), vitamin d deficiency, and hyperlipidaemia were not collected and compared, but it is plausible that these would also show similar levels of differences. each of these comorbidities has been shown to significantly increases the risk of early death. the national diabetes audit in their mortality study (20) found that the presence of diabetes increases a person standard mortality risk by a factor of 1.6. this article is protected by copyright. all rights reserved it is, therefore, plausible that those patients who died of covid-19 were, on average, already in relatively poor health for their age and this poor health would give them a life expectancy, on average, significantly below that of the age-equivalent general population. these comorbidities and conditions also reduce the person's quality of life, as well as its quantity (21) . the impact of poor health through long-term conditions and comorbidities are usually incorporated into modelling through a quality of life utility factor which ranges from 1 (healthy) to 0 (death); this is used to adjust the total life years. beaudet in the current situation, the following existing costs are associated with the current 60,000 excess deaths. the following associated hospital activity was reported  icu: the intensive care national audit and research centre (icnarc report) (25) showed that 10,130 patients with covid-19 were treated (including 72% advanced respiratory support, 30% advanced cardiovascular support, 26% renal support) for a median of 11 days, which based on reference costs of £1,503/day for a mix of 1/2/3 organ support (26) gives a total £16,500/icu admission. from the current ratio, 1 icu this article is protected by copyright. all rights reserved admission is associated with 6 excess deaths, this is equivalent to £2,600 icu costs incurred/ excess death. (19) . if 20% of icu survivors experience these effects, there would be a total loss of 18,000 life years in the future relative to the current 60,000 excess deaths or 0.3 qaly/excess death, which if valued at @£30,000/qaly gives £9,000 qaly value lost /excess death. together these would bring approximately £20,000 healthcare benefits for each excess death avoided through the lockdown. the benefit in terms of estimated lives saved suppose that a group of people who each had expected quality-adjusted remaining years of life of 5 years, and who might have died with the virus, has been spared that because of government restrictions ("the lockdown"). we will assume that the benefit of the restrictions that prevented such deaths are the value of 5 quality-adjusted years of life multiplied by the number of lives saved. the nice £30,000 threshold is an assessment of the (maximum) resource cost that would be justified for the uk health service to make an this article is protected by copyright. all rights reserved expected saving of one quality-adjusted year of life. to save 5 qaly would be worth up to £150,000. we apply this figure of £150,000 (or a figure of £300,000 if we make no adjustment for co-morbidities and take 10 life years lost per death) to estimates of the possible number of lives saved as a result of lockdowns to give an overall benefit number. to that, we add an estimate of £20,000 other health costs saved (per life saved) based on the evidence summarised above. we compare that aggregate number with an estimate of the lost resources from the lockdown. the hospitalised fatality rate (30) has fallen from 6%/day at the start of april to 1.5% in mid-june. so, the estimate of 500,000 deaths made back in march and based upon fatality rates then may have been particularly pessimistic. this article is protected by copyright. all rights reserved at the other end of the spectrum would be estimates of net saved lives that are effectively zero. we set the lowest estimated net set lives well above that and use (rather arbitrarily) a "lowest" estimate of 20,000. for each life saved we apply a factor of either 5 or 10 quality-adjusted extra years of life, each valued with the nice guideline figure of £30,000 (7). we also factor in that alongside fewer deaths there would have been far fewer demands made upon the resources of the public health system and we have placed a value of that per potential life saved of £20,000. on the cost side the lowest resources cost is just to count the gdp that would have been produced in 2020 but for the lockdown established in march and assuming the lockdown to be eased from the end of june. this assumes a rapid bounce back by the end of the year so there is no effect on incomes and output from the start of 2021 onwards. that was the scenario envisaged by the bank of england in their may 2020 assessment of the economic outlook when they put the gdp loss in 2020 at around 14%. the obr estimate for lost output in 2020, also based on an assumed rapid recovery in the second half of the year, is close to 13%. it seems plausible that a large fraction of these estimates of lost output is due to the lockdown. but even absent a government-mandated lockdown there would have been some reduction in incomes. if the lockdown effect was only twothirds of the total, then the bank of england and obr estimates might imply around a 9% fall in gdp as a result of it. at the high end of the spectrum would be an estimate of 15% of gdp lost in 2020 and lower output for the next few years on top of that as economic activity does not return to normal for several years with some firms permanently damaged by the lockdown and the large rise in unemployment slow to be reversed, even if restrictions are quickly removed from mid-2020. a shortfall of gdp of 15% in 2020; 7.5% in 2021 and 2.5% in 2022 (so that the cumulative lost output would then be 25% of gdp). that would be at the more this article is protected by copyright. all rights reserved as noted the national institute of economic and social research put their best guess of the narrowly defined economic cost higher again. tables 2 and 3 show the cost-benefit calculations of the lockdown based on such ranges: in each cell, we report three numbers: benefits (+), costs (-) and (in red) the balance of the two -all measured as £ billion. tables 2 and 3 (with benefits raised by a factor of 3) this would still generate costs of the lockdown in excess of benefits in nearly all the cases considered. that judgement is, however, made with the benefit of hindsight: we now know more about the scale of the economic costs of the lockdown than was known in march, and also know about how deaths and new infections have evolved across europe. the more interesting policy issue is what it is best to do now: how quickly should the lockdown be eased given what we know now? that issue we consider in the next section. we apply a similar cost-benefit methodology to consider policy options for the level of restrictions applied in the uk over the next 3 months (july-september 2020). the options we consider fall under two broad headings: this article is protected by copyright. all rights reserved 2. move quickly to minimal lockdown (easing restrictions rapidly and relying on existing tracking of the cases/deaths to prevent re-emergence of the virus) we consider the following scenarios for the consequences of each policy for the evolution of covid-19 deaths: 1. very limited easing of restrictions results in a continuing steady fall in the death rate over 13 weeks down to single figures per week at the end of three months. each week deaths are assumed to be 0.7 x deaths of the previous week. each week they are 15% higher than the week before. the assumed paths of deaths under the 4 scenarios are shown in table 4 . in each case, we set the initial level of deaths in the week prior to each scenario at the last ons our low-end estimate of the (narrowly defined) cost of the march to early june lockdown was 9% of gdp -a figure of a little over £200 billion. one might assume that a continuation of the lockdown over the next three months with only a very limited easing of restrictions generates a further cost of the same size. but the rapid easing of restrictions is unlikely to generate zero costs. such costs may still be substantial, though likely far lower than a continuation of lockdown policies. a conservative estimate of the benefits of easing the lockdown is that the £200 billion costs under lockdown might be half that size. this article is protected by copyright. all rights reserved this would generate a benefit from easing of £100 billion over three months to be set against any extra lives lost. under all scenarios the cost of easing is a small fraction of the benefits -the maximum cost of £14 billion should be set against a conservative estimate of benefits of £100 billion. one would need to value qalys at £220,000 -over 7 times the nice guideline value of £30,000 to make a continuation of the lockdown warranted in the scenario of the greatest number of live years not lost with costs / qaly much higher for less live years saved. that runs counter to agreed uk policy on the economic viability of health interventions (7). it is not straightforward to assess exactly how much of the slowing in new infections and deaths is directly attributable to the lockdown. people were altering their behaviour before severe restrictions were introduced -mobile phone data reveals sharp declines in the movement of people some weeks ahead of lockdown. some changes in behaviour (washing hands, avoiding crowds) may have been effective in reducing infections but at a low economic cost. it is also possible that a significant degree of immunity may have built up by the time severe restrictions were introduced because the infection may have spread quite widely and largely unnoticed with the asymptomatic a very large fraction of the infected. a substantial proportion of the population may have been effectively immune from the virus when lockdowns started not just because of recovery from past infections that conferred a degree of immunity but also because a significant proportion of the this article is protected by copyright. all rights reserved population may never have been susceptible. in this regard at least two studies reported possible immunity against sars-cov-2 due to previous infections with harmless coronaviruses. braun et al (33) showed the presence of sars-cov-2 reactive t cells in covid-19 healthy donors and grifoni et al (34) reported the detection of sars-cov-2reactive cd4+ t cells in ∼40%-60% of unexposed individuals, suggesting cross-reactive t cell recognition between circulating "common cold" coronaviruses and sars-cov-2 all three factors described above may have played a role, and all would mean that wieland (2020) (38) modelled the spread of the infection across germany and concluded that infections were past their peak and starting to decline ahead of the introduction of government restrictions there. the results were summarised thus: "in a large majority of german counties, the epidemic curve has flattened before the social ban was established (march 23). in a minority of counties, the peak was already exceeded before school closures." friston (39) and levitt (40) both conclude that the numbers of people not susceptible to the covid-19 virus were already very substantial before lockdowns were introduced and that the virus was burning itself out. this article is protected by copyright. all rights reserved testing based on the presence of antibodies, however, put the level of those who have had the infection in european countries where the virus has spread most rapidly at only 5-10%, though in some areas within countries it is still high enough to have had a significant impact on the r-value, which is the number of people subsequently infected by each infected person. antibody testing itself is not a perfect tool and additional research is needed to determine if and to what extent a positive antibody test means a person may be protected from reinfection with sars-cov-2. serology tests should not be used as a stand-alone tool to make decisions about personal safety related to sars-cov-2 exposure, lerner et al (41). in contrast to many other european countries, the swedish strategy has been one of adopting much less restrictive measures that is far short of a lockdown (see figure 2 ). infections and deaths have been far higher than in neighbouring denmark and norway, but excess deaths are lower than in many european countries and only one third the level (relative to population) in the uk. in terms of overall impacts, there is contradictory evidence. born et al (2020) (42) and krueger, uhlig and xie (2020) (43) argue that the swedish strategy has been successful. but health outcomes in countries most similar in terms of climate, the density of population and standard of living (that is denmark and norway) appear to be much better. the uk data show a significantly higher cumulative death rate than sweden; figure 1 and (44) show excess deaths relative to expected in the uk at more than twice the swedish level by early june 2020. on this measure, sweden sits near the middle of the pack for european countries. death rates in several countries with harder "lockdowns" have been significantly higher than in sweden. a great deal of evidence is already emerging on the (narrow) economic impacts of restrictions. estimates made by deb et al (2020) (45) to identify the particular effect of restrictive policies (lockdown) suggest that they reduced economic activity by 15% in the 30 days after they were adopted. they find that stay-at-home requirements and workplace closures are the costliest in economic terms. preliminary estimates from the this article is protected by copyright. all rights reserved uk office for national statistics showed a slightly more than 20% fall in gdp in april 2020, the first full month after the lockdown. bonadio et al (2020) (46) put the impact on output and incomes (i.e. gdp) of policies to counter the spread of the infection on gdp averaged across 64 countries even higher, at around 30%. costs which will come further down the road because of disruption to healthcare (47) and education are harder again (48) to measure relative to the more immediate effects on economic production and employment (49) . the lockdown can be seen as having 2 elements: a social lockdown (distancing, no social gathering, recreation or sports events etc) and an employment lockdown (not travelling or going to non-essential work). the latter is not likely to be the most effective response as infection amongst the employed was not a major source of mortality. linking the total population by age group with employment (50) (52) showed that the infection rate amongst employed age group (20-49) was 70% higher than the stay at home age groups, and combining this with the population by age group shows 51% of all infections and 1% of deaths were recorded within this age group. in the uk it is hard to be sure of the scale of benefits of the lockdown in terms of lives saved and the avoidance of the resources of the health service being exhausted. in terms of lives saved estimates range from very few lives saved to a high of perhaps 450,000 lives saved (that is the difference between the 500,000 or so deaths projected by there are reasons to be sceptical of figures at the high end of that scale which puts the saving of lives from the lockdown at several hundreds of thousands:  the low cost of effective forms of behavioural change (washing hands, avoiding crowds) adopted by individuals makes it unlikely that in the uk there would have seen this article is protected by copyright. all rights reserved 500,000 deaths even with no government restrictions; the 500,000 figure from the benefits of a lockdown are, however, not just in terms of lives saved -though that is of immense value. valuing the health care resources saved because the lockdown reduced the numbers of sick people is also an important element of its benefits. this article is protected by copyright. all rights reserved there is a need to normalise how we view covid-19 because its costs and risks are comparable to other health problems (such as cancer, heart problems, diabetes) where governments have made resource decisions for decades. treating possible future covid-19 deaths as if nothing else matters is going to lead to bad outcomes. good decision making does not mean paying little attention to the collateral damage that comes from responding to a worst-case covid-19 scenario. the lockdown is a public health policy and we have valued its impact using the tools that guide health care decision in the uk public health system. on that basis and taking a wide range of scenarios of costs and benefits of severe restrictions, we find the lockdown has consistently generated costs that are greater -and often dramatically greater -than possible benefits. movement away from blanket restrictions that bring large, lasting and widespread costs and towards measures targeted specifically at groups most at risk offers is now prudent. such a policy has now been implemented. ethics approval and consent to participate: the analysis used nationally available general practice-level data with no patient identifiable data. therefore, we felt that ethics permission was not required. this article is protected by copyright. all rights reserved funding: no external funding was accessed to fund the work impact of non-pharmaceutical interventions (npis) to reduce covid-19 mortality and healthcare demand prime minister's statement on coronavirus a geo-temporal survey of hospital bed saturation across england during the first wave of the covid-19 pandemic hsj-nhs block books almost all private hospital sector capacity to fight covid-19 closed for business how nice measures value for money in relation to public health interventions center on the economics and development of aging (ceda) www.mortality.org covid-19 government response stringency index blavatnik school of government ons deaths registered weekly in england and wales, provisional: week ending 26 hmrc coronavirus (covid-19) statistics ons monthly gdp and main sectors to four decimal places 14 letter to the editor. loneliness : a signature mental health concern in the era of covid-19 type 1 and type 2 diabetes accepted article this article is protected by copyright. all rights reserved and covid-19 related mortality in england: a cohort study in people with diabetes estimating utility values for health states of type 2 diabetic patients using the eq-5d (ukpds 62) review of utility values for economic modeling in type 2 diabetes the fatality and morbidity components of the value of a statistical life the value of health and longevity the intensive care national audit and research centre (icnarc) report on covid-19 in critical care reference cost collection: national schedule of reference costs -year 2017-18 -nhs trust and nhs foundation trusts) cabinet briefing 26/6/2020: accessed 2 features of 16,749 hospitalised uk patients with covid-19 using the isaric who clinical characterisation protocol life expectancy and years of life lost in chronic obstructive pulmonary disease: findings from the nhanes iii follow-up study declining death rate from covid-19 in hospitals in england when to release the lockdown: a wellbeing framework for analysing costs and benefits demographic perspectives on mortality of covid-19 and other pandemics presence of sars-cov-2 reactive t cells in covid-19 patients and healthy donors. medrxiv targets of t cell responses to sars-cov-2 coronavirus in humans with covid-19 disease and unexposed individuals fundamental principles of epidemic spread highlight the immediate need for large-scale serological surveys to assess the stage of the sars-cov-2 epidemic a phased approach to unlocking during the covid-19 pandemic -lessons from trend analysis assessing the spread of the novel coronavirus in the absence of mass testing flatten the curve! modelling sars-cov-2/covid-19 growth in germany on the county level the corona chronologies" available at do lockdowns work? a counterfactual for sweden macroeconomic dynamics and reallocation in an epidemic financial times coronavirus tracked: the latest figures as countries start to reopen the economic effects of covid-19 containment measures global supply chains in the pandemic doesn't need lockdowns to destroy jobs: the effect of local outbreaks in korea labor markets during the covid-19 crisis: a preliminary view the cost of the covid-19 crisis: lockdowns, macroeconomic expectations, and consumer spending ons a05 sa: employment, unemployment and economic inactivity by age group ons deaths involving covid-19, england and wales: deaths occurring in covid-19) infection survey pilot: age specific mortality rate 53. mortality associated with covid-19 outbreaks in care homes: early international evidence adelina comas-herrera this article is protected by copyright. all rights reserved key: cord-337516-hrcf2udq authors: dickens, borame l.; koo, joel r.; lim, jue tao; park, minah; quaye, sharon; sun, haoyang; sun, yinxiaohe; pung, rachael; wilder-smith, annelies; chai, louis yi ann; lee, vernon j.; cook, alex r. title: modelling lockdown and exit strategies for covid-19 in singapore date: 2020-08-01 journal: nan doi: 10.1016/j.lanwpc.2020.100004 sha: doc_id: 337516 cord_uid: hrcf2udq abstract background with at least 94 countries undergoing or exiting lockdowns for contact suppression to control the covid-19 outbreak, sustainable and public health-driven exit strategies are required. here we explore the impact of lockdown and exit strategies in singapore for immediate planning. methods we use an agent-based model to examine the impacts of epidemic control over 480 days. a limited control baseline of case isolation and household member quarantining is used. we measure the impact of lockdown duration and start date on final infection attack sizes. we then apply a 3-month gradual exit strategy, immediately re-opening schools and easing workplace distancing measures, and compare this to long-term social distancing measures. findings at baseline, we estimated 815 400 total infections (21.6% of the population). early lockdown at 5 weeks with no exit strategy averted 18 500 (2.27% of baseline averted), 21 300 (2.61%) and 22 400 (2.75%) infections for 6, 8 and 9-week lockdown durations. using the exit strategy averted a corresponding 114 700, 121 700 and 126 000 total cases, representing 12.07–13.06% of the total epidemic size under baseline. this diminishes to 9 900–11 300 for a late 8-week start time. long-term social distancing at 6 and 8-week durations are viable but less effective. interpretation gradual release exit strategies are critical to maintain epidemic suppression under a new normal. we present final infection attack sizes assuming the ongoing importation of cases, which require preparation for a potential second epidemic wave due to ongoing epidemics elsewhere. funding singapore ministry of health, singapore population health improvement centre. case importation and the non-elimination of local cases, which parallel economic and social issues from lockdown implementation. sustainable and public-health orientated exit strategies are therefore required. we searched pubmed from database inception to may 20, 2020, for articles using the search terms "covid-19", "sars-cov-2", "exit strategies" "lockdown" and "control measures". our search yielded three modelling papers of relevance. two articles examined control strategies within wuhan with one focusing on the effects of social mixing and the other on case isolation. another estimated the efficacy of long-term social distancing in the us. we found no articles which modelled the impacts of national lockdown and exit strategies for countries within the western pacific outside of china. in response to the rising number of local coronavirus disease 2019 (covid-19) cases, many countries have implemented lockdowns to restrict movement and community interactions and thereby suppress infection. details vary [1] but measures commonly include the closure of premises such as non-essential workplaces, schools, recreational facilities and places of worship. essential services in healthcare, transport, cleaning, food services and supply chains may remain open to sustain the economy and welfare of the population. most lockdowns have successfully reduced the reproduction number and curtailed epidemic growth, but emerging from a lockdown prematurely without sufficient planning may result in epidemic rebounding [2] , as the previous suppression of cases may leave a substantial proportion of the population susceptible to infection. intermittent lockdowns have been proposed [3] through the cycling of initiation and cessation of these interventions-possibly until 2022-to avoid exceeding hospital capacities. concerns exist, however, on the long-term economic disruption, negative consequences on social and mental well-being, and costly administration required to ensure compliance. here, in the context of lockdown and social distancing measures, we explore the impact of a gradual release exit strategy (gres) with a rollback of social distancing interventions operating outside of trigger mechanisms from emergency intermittent lockdown measures. health and isolation facility expansion and preparation will be required during the lockdown phase regardless of control strategy to maintain a high standard of healthcare, which can substantially reduce case fatality rates [4] . gres can provide timelines to government agencies, services and businesses, and allow society and the economy to return to a new-normal state with continuing extensive education on the importance of safe distancing, hygiene practices and precautionary measures such as mask wearing. we assess gres through an agent-based model of a city, modelled upon the city-state of singapore. the implementation of social distancing measures in singapore has been progressive from 10 march with the rollout of gradually stricter recommendations and regulations [5] , leading to the implementation of a 'circuit breaker' [6] , or lockdown, on 7 april 2020 in response to rising autochthonous cases. although singapore's lockdown seemingly had no effect on several large outbreaks among foreign worker dormitories that started in early april, it appears to have arrested the epidemic growth in the general population, using school closures and workplace distancing as recognized and effective attack rate reduction measures [ 7 , 8 ] . substantial education and productivity losses are expected with all children tele-learning and an estimated 80% non-essential employees working from home or unable to work [9] . in response, a 3-tier financial support package for businesses and individuals has been released to provide financial assistance for families, support networks, workers and students amounting to ∼$38.8 billion usd [10] . on 21 april 2020, the singaporean government announced an extension of social distancing measures to 8 weeks in total with plans to end on 1 june 2020 [11] . we utilise the geographical, demographic and epidemiological model of singapore for respiratory diseases (geodemos-r), an agent-based epidemic simulation model comprising of a synthetic and calibrated population where the impact of interventions can be measured. the model has been previously used to estimate the effects of early epidemic control [12] and of home versus institutional isolation of cases [13] , and has been updated and expanded to assess exit strategies. in summary, we investigate the effects of early social distancing, lockdown and gres in 480 days by: (1) establishing a limited control baseline with case isolation and quarantining of family members only; (2) measuring the impact of 6-week (early cessation), 8-week (planned) and 9-week (extended) lockdown in duration, at different start dates of 5, 6, 7 and 8 weeks, on the final infection attack size, when compared to a no-exit strategy where lockdown is immediately lifted; (3) estimating the effects of this lockdown with gres which includes the immediate re-opening of schools post lockdown, due to their limited epidemic suppression impact [14] , with a 3-month readjustment period. in the first 2 months, 50% of the workforce returns physically to work, followed by 1 month at 75%, before full re-opening to pre-epidemic levels; and (4) comparing lockdown to a no-lockdown strategy with longterm social distancing of differing start times at 5, 6, 7 and 8 weeks from the epidemic start date and durations of 2, 4, 6 and 8 weeks. during this period, 50% of the adult population is actively working, schools are closed and active social distancing is being done within the community. the model used is geodemos-r, an agent-based epidemiological model with a synthetic resident population that is heavily calibrated to be representative. full details are explained elsewhere; [12] here we provide a summary. households were constructed using 178 census tables based on a sample of 20 0,0 0 0 households. an heuristic search algorithm was used to create a total of 3.77 million singaporeans with the attributes of age, ethnicity and gender. remaining attributes were drawn randomly from the specified attribute's marginal distribution and summary tables of the population comparable to the census tables formed whose fit were assessed using pearson's chi-squared statistic. zero count cells were avoided by setting them to 0.1. each attribute was fitted using a monte carlo swapping algorithm, which swaps two random individuals' data until the improvement in fit becomes negligible ( < 0.001% improvement for 10,0 0 0 runs). the same process was used to generate partners, allocate children and create multi-generational families for 1.14 million households. households were geolocated within discrete areas named subzones according to spatial characteristics outlined in the census, and individuals allocated workplaces and schools appropriately based on distributions of commuting time data from singapore's household interview travel study and ez-link data which is a 1-month record of the majority of the population's public transport activities. in the transmission model, day and night steps exist to differentiate movement behaviour and infection likelihood between individuals within the household, school, workplace and community. during the day, workers interacted with individuals in the same workgroup, and students within their classes. these two groups also interacted with those in the wider area around their workplace and school at community rates. individuals who were not working or studying were modelled to interact with people in the same residential community. in the night step, individuals interacted primarily within their households where children had the highest probability of contact with his or her family members. suppose i and j are two individuals in the synthetic population with j becoming infected, we denote the probability of j infecting i on day t in location type g as (1) here β g is constant for location type g , defined as the home, workplace or school which both individuals belong to; β g is obtained from a contact rate study (supplementary table 1 , 2) for different social group settings in singapore where the contact rates serve as the likelihood for individuals to infect one another at specific group locations and the wider spatial subzone area. overall, the probability of individual i getting infected from location type g on day t is therefore given as, here g is a set of individuals of location type g, g t is the subset of all individuals who belongs to set g and are infectious on day t . we use | . | to denote the size of the set of individuals. hence, | g | is the total number of people in set g . the number of people in set g that would be infected on day t can be denoted by a random variable x g ( t ) and where g r is the subset of g consisting of individuals that have been removed through hospitalisation and subsequent recovery. the total number of people α infected on day t can then be expressed by summing over all the different sets of individuals in the population, for each simulation, the model was run for 480 daysapproximately 15 months-to estimate the number of cases by the end of april 2021; cases began entering singapore in january 2020. with more epidemiological information available, model parameters that were previously based on the severe acute respiratory syndrome coronavirus (sars-cov) have been updated (a summary of parameters and interventions is provided in supplementary table 1 ). we assumed that the basic reproductive number ( r 0 ) for sars-cov-2 was 2.0 [15] , the asymptomatic rate was 17.9% [16] and the incubation cumulative distribution function was modified to have a median incubation period of 4 days [17] . the r 0 parameter was built with a multiplier γ , which modified the infectiousness parameter of each individual in the simulation [18] . we first simulate the initial 4 weeks for a selected γ ∈ [0, 1]. this γ has a correspondence with r ∈ r , the value of the exponential model exp( rt + b 0 ) that best fits the simulated 4-week outbreak. this value of r was used in the linearized form of the susceptible-exposed-infectious-removed (seir) model [19] to compute the corresponding r 0 . through this process we obtained a corresponding value of r 0 for γ . a grid search was performed on γ , with γ that corresponded closest to r 0 = 2.0 being selected. for each scenario, we had case importation based on a poisson model with λ = 2.0. to calculate λ, case importation data was used to fit a model on the expected number of daily case importations, where λ was the model's average case importation over time. for this study, we ran 100 simulations for each intervention. there was no limit to isolation capacity as we assumed that the majority of the symptomatic cases will be transferred to hospital or community isolation facilities. our baseline scenario included the isolation of ascertained cases and home quarantine of their household members. all ascertained cases are assumed to have a 24-h delay before they are no longer infectious to the wider population to accommodate for healthcare facility visitation and testing. we additionally assumed perfect compliance of those under household quarantine as strict punitive measures are in place. for lockdown, a harsher penalisation on the contact rate was implemented with an initial 2-week period of social distancing, followed by a 6, 8 or 9-week period where schools remained closed, and work and community contact rates are further reduced to 20%. this further reduction is to simulate essential work and economic activity still being carried out by the population during the lockdown period. the starting points were at week 5, 6, 7 and 8 of the epidemic. for the post lockdown strategy, two scenarios were modelled. the first assumed conditions went back to a pre-epidemic state with all schools, workplaces and the community at 100% transmission. the second, labelled as the gradual release exit strategy (gres), assumed a successive restoration of contact rates over a period of three months. in the two months directly after lockdown, schools reopen and contact levels are restored at 50%, followed by 1 month of 75% restoration of workplace and community contact levels, and then pre-epidemic contact rates at 100%. this represents a cautious and planned approach to avoid heavily abnormal contact disruption from the initial lockdown period whilst maintaining a level of epidemic suppression. for a long-term social distancing strategy, school closure occurs and we assume 50% suppression in contact rates at workplaces and within the wider community. this measure can begin at week 5, 6, 7 and 8 where at each starting point, social distancing was implemented for 2, 4, 6 and 8 weeks in duration. after the end of the social distancing period, all schools were reopened with contacts restored to pre-intervention levels. the two weeks of social distancing pre lock-down are also assumed to follow the same contact rate reduction levels with school closure. we present the main results using a 17.9% asymptomatic proportion, and the same analysis for 44% in the supplementary information ( supplementary figs. 1-4) . for the limited control baseline, the total number of infections by 480 days was 815 40 0 (iqr: 814 60 0-816 50 0), which represents 21.62% (21.60-21.65%) of the total population ( table 1 ) a delayed post lockdown secondary peak size of 6060 (5260-6740) daily cases occurs for a 6-week lockdown duration starting at week 7, which is reduced further to 5890 (5040-6590) for an 8week lockdown in duration ( fig. 1 (e) ). whilst substantially lower than the baseline peak, the initial pre-lockdown peak is relatively high. this peak is suppressed however for a lockdown starting at week 6 ( fig. 1 (c) ). although the secondary peak is greater at 14 100 (13 200-15 000) and 13 900 (12 800-14 700) daily cases respectively, it represents peak suppression throughout the epidemic in comparison to the baseline. lockdown start time has a larger impact than duration as relatively small ranges in the cases averted exist; for an asymptomatic rate of 44%, the total number of infections by 480 days rose to 1 085 0 0 0 (1 084 0 0 0-1 086 0 0 0), representing a 24.9% increase in the infected population. for lockdown and gres, early lockdown at 5-weeks had a final infection attack size of 700 700 (699 000-702 800) total cases for a 6-week lockdown, 693 700 (691 000-695 700) for 8-week lockdown, and 689 400 (686 80 0-692 20 0) for a 9-week lockdown ( table 1 ) lockdown start time has a large impact on gres ( fig. 1 and table 1 ). with early lockdown implementation at 5 weeks, when compared to no-exit strategy, gres averts an average of 96 200 cases (12.07%) in total for a 6-week early cessation lockdown, 100 400 (12.64%) for the planned 8-week and 103 600 (13.06%) for the prolonged 9-week. at this 5-week start time, gres also reduces the secondary peak size by 9 900 (46.92%), 10 100 (49.03%) and 10 200 (50.25%) at a ∼28-day delay respectively ( fig. 1 (a) ). if implemented later and close to the epidemic peak at week 8, the efficacy of gres diminishes with 11 300 (1.91%), 10 600 (1.81%) and 9900 (1.69%) cases averted in total, with no determinable secondary peak ( fig. 1 ) . gres' lessening utility with later lockdown start times is due to a large proportion of cases having already occurred in the first half of the epidemic. post lockdown peak sizes for a 6 week start time lockdown strategy are substantially reduced with 8740 (8580-8980), 8760 (8590-8930) and 8850 (8650-8980) daily cases for a 6, 8 and 9week duration lockdown with gres and 14 100 (13 20 0-15 0 0 0), 13 900 (12 800-14 700) and 13 800 (12 800-14 700), respectively, without ( table 1 ) . this represents a 38.01, 36.98% and 35.87% reduction. for an earlier implementation lockdown time at 5 weeks, a 47.92%, 49.03% and 50.25% corresponding post lockdown peak reduction is observed. lockdown duration had a limited impact on gres, acting as a suppressive, not a preventative, measure. when utilising the 6week early cessation, planned 8-week lockdown and prolonged 9week lockdown, a difference of 89 10 0, 94 0 0 0 and 98 300 cases can be averted across the different start dates of 5, 6, 7 and 8 weeks using gres in comparison to lockdown alone; differing only by 9200 cases between these duration times. at a greater asymptomatic proportion of 44% we observed similar outcomes in terms of cumulative infection numbers for lockdown implementation on week 5 and 6 (supplementary figs. 1 and 2) although it represented an overall accelerated epidemic. the largest difference was observed for a 6-week lockdown with gres at week 5 where 833 600 (827 700-837 400) infections were observed in comparison to 700 700 (699 000-702 800) at an 17.9% asymptomatic rate. long term social distancing at durations of 2, 4, 6 and 8-weeks cause resulting final infection attack sizes of 820 100 (818 500-821 40 0), 804 90 0 (801 30 0-808 0 0 0), 783 60 0 (779 0 0 0-786 30 0) and 759 0 0 0 (753 80 0-763 0 0 0) total cases across implementation start times of 5, 6, 7 and 8 weeks ( table 2 ) . post social distancing peak sizes show suppression, particularly for implementation times of week 7, although if implemented at week 6, the initial peak is also reduced ( fig. 2 ) . the secondary peaks caused by early lockdown and gres are comparable to long-term social distancing with 50% activity in the community and the workplace, wherein 6-week interventions starting at week 7 cause a maximum of 5390 (4750-6070) cases for the former and 5490 (4710-6040) cases in the latter ( fig. 2 (e) ). long term social distancing is less effective at durations of 4 weeks or lower, averting up to 1.29% of cases relative to the baseline for an implementation time at 5 weeks ( table 2 ) . for a duration of two weeks in social distancing, a maximum of 11.22% of cases relative to the baseline can be averted if implemented close to the epidemic peak at 8 weeks. for longer durations of 6 to 8 weeks, 3.90-6.92%, 9.59-12.84%, 17.27-20.25% and 25.59-28.48% of cases can be averted relative to the baseline for start dates of 5, 6, 7 and 8 weeks respectively. for a week 6 implementation time, 12.84% of cases can be averted critical uncertainties lie ahead on moving forward during lockdown periods in when and how to release the population back to a new normal state. our findings suggest that earlier lockdown requires a gres to prevent a large secondary peak, which becomes jid: lanwpc [m5g; august 1, 2020; 13:30 ] less important the later the lockdown is implemented. as singapore entered lockdown relatively early where ∼100 cases were being reported daily, a lockdown with no exit strategy could result in up to 796 900 (791 800-799 300) total cases across 480 days, which averts 2.27% of cases in comparison to the baseline. gres can avert up to 14% of cases for the same scenario. for lockdowns implemented at later weeks, substantial proportions of cases are averted, as observed for other epidemics [20] , with a final infection attack size of 591 500 (577 80 0-60 0 10 0) at week 8 as an example, although this will not be viable for most countries where their healthcare capacity cannot cope with large, early epidemic peaks. when gres is compared with no exit strategy for a 5 and 6week implementation start time, it results in a reduction of 8.01-13.06% in final infection attack size, and 0.78-1.91% reduction for a 7 and 8 week start time as a large proportion of cases have already occurred in the initial epidemic ramp up. gres therefore shows good utility in populations where substantial proportions are susceptible to infection, reducing the overall outbreak size and slowing down infection spread with a greater number of transmission events prevented. this may provide opportunity to introduce prophylaxis or vaccination measures in the future, or allow the implementation of mandatory testing of all incoming travellers, which can prevent these infection events from occurring altogether as the epidemic dies out. this is especially paramount as the number of cases beyond 365 days, although considerably lower than the initial peaks, still ranges from 6.51-513.21 cases per day for gres strategies (supplementary table 3a ) and 4.94-90.03 for long term social distancing strategies (supplementary table 3b ) with the ongoing risk of secondary case spread from imported cases. at the start of the epidemic however, the duration of lockdown has a limited impact on the efficacy of gres in terms of final infection attack size, which suggests that the period should be used for epidemic preparation rather than an ongoing control method, although an extended lockdown of 8 or 9 weeks will avert more cases. should medical capacity permit, long term social distancing remains a viable strategy provided it is carried out for a duration of at least 6 weeks with 50% of the population working and schools closed. the total infection attack size is less for social distancing relative to lockdown with no exit strategy, but remains greater than gres which is the most effective strategy. for overall peak size minimization, both a social distancing and gres implementation time of 6 weeks is optimal, although if earlier lockdowns or social distancing measures have been implemented, gres is essential for peak reduction. with a greater asymptomatic rate of 44%, the epidemic is considerably accelerated requiring earlier intervention at 4 or 5 weeks to slow infection spread. gres remains effective at a 23.1% reduction in the final infection attack size in comparison to the baseline although these implementation dates are notably much later into the epidemic in comparison to the 17.9% asymptomatic proportion findings. multiple complexities exist in the framing of exit strategies to the public, including the attribution of accountability among policymakers and individuals, the acceptance of uncertain economic burdens, and need for flexibility as technologies emerge and other countries respond differently as the epidemic progresses. lockdown and gres can maintain public preparation, awareness and adherence which may abate if complete cessation and re-initiation of measures are introduced for extensively long periods, as observed generally for long-term therapies and lifestyle changes [21] . the strategy also partially mitigates isolation fatigue and social disruption, and should be conducted along with case finding, contacttracing and quarantining, mass-testing in key groups such as those working with vulnerable populations, and serological surveys. with gres in place, which will still require healthcare system ramp-up, real-time forecasting efforts can also continue to estimate appro-priate intervention trigger times should another lockdown be required. investment into the healthcare system to accommodate for the caseloads estimated in the study could avoid recurrent lockdowns, if the healthcare system can cope with the number of severe cases. the prolonged flattening, not elimination, of the epidemic across scenarios is due to the constant influx of a conservative 2 estimated imported cases per day at the global travel hub, an estimated 17.9% (or 44%) prevalence of asymptomatic infections and inevitable future relaxation of travel restrictions to allow for influxes of short and long-term worker immigration. additionally, containment measures and lockdowns vary widely in duration and severity between countries, [1] causing differing and delayed epidemic peaks among source countries, making lockdown a temporary protection measure against highly uncertain external epidemiological forces where delays in reactive control measures abroad or within singapore will result in inevitable national case spread. although seroprevalence data has yet to be released, due to the ongoing lockdown measures, a substantial proportion of the population is suspected of being susceptible to infection, which exacerbates the effects of case importation. ongoing concerns regarding the effects of school closure remain at the forefront of covid-19 control policy. gres prioritises school re-opening as clinical manifestations of children's covid-19 appear generally less severe [22] . the societal impacts of school closure among young children, working adult parents and older children in terms of productivity and education loss are likely to be extensive and unsustainable. with continued school closure, socioeconomic inequalities will be further exacerbated, despite governmental intervention to provide intensive distance learning for all school children [23] . overall, outside of education, schools operate as a safety net for at-risk children, providing nutritional, emotional and social support as well as vaccinations and development opportunities. limitations of this study include uncertainties on the current number of infections, which could not be used directly to validate the initial phase of the epidemic in the agent-based model. parameter estimations from larger global studies were therefore used to simulate epidemics although wide ranges in observations continue to be reported for parameters such as asymptomatic rates. the model can be calibrated in the future with real time forecasting effort s and result s of serology surveys to better reflect covid-19 prevalence and incidence within singapore. intra workplace and school contact patterns were additionally not accounted for, which may cause greater or fewer localized contact events depending on internal grouping structures. further uncertainties include the current and future adherence to social distancing measures, future importation rates and spatial heterogeneities in infection rates. ongoing epidemics in high density accommodation (supplementary figure 5) and their effects in the wider community were also not accounted for, which requires further investigation. bld and jrk performed the modelling and wrote the manuscript. bld, jrk, vjl and arc designed the intervention strategies. mp and sq performed data collection. jtl, hs, ys, rp, aws, lyac, vjl and arc critically revised the manuscript. the study sponsors had no role in the study design, analysis, interpretation of the data or writing of the report. lanwpc [m5g first-wave covid-19 transmissibility and severity in china outside hubei after control measures, and second-wave scenario planning: a modelling impact assessment projecting the transmission dynamics of sars-cov-2 through the postpandemic period how will country-based mitigation measures influence the course of the covid-19 epidemic past updates on covid-19 local situation covid-19 (temporary measures) act 2020. singapore: republic of singapore, thursday the effect of school closure on hand, foot, and mouth disease transmission in singapore: a modeling approach effectiveness of workplace social distancing measures in reducing influenza transmission: a systematic review strong national push to stem spread of covid-19 interventions to mitigate early spread of sars-cov-2 in singapore: a modelling study institutional, not home-based, isolation could contain the covid-19 outbreak school closure and management practices during coronavirus outbreaks including covid-19: a rapid systematic review early transmission dynamics in wuhan, china, of novel coronavirus-infected pneumonia estimating the asymptomatic proportion of coronavirus disease 2019 (covid-19) cases on board the diamond princess cruise ship investigation of three clusters of covid-19 in singapore: implications for surveillance and response measures flute, a publicly available stochastic influenza epidemic simulation model comparative estimation of the reproduction number for pandemic influenza from daily case notification data strategies for mitigating an influenza pandemic long-term adherence to health behavior change epidemiology of covid-19 among children in china mitigate the effects of home confinement on children during the covid-19 outbreak the authors declare no competing interests. supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.lanwpc.2020.10 0 0 04 . all data used within the study is publicly available. geodemos-r data is available on request. key: cord-264821-68us87xb authors: labrague, l.; ballad, c. a. title: lockdown fatigue among college students during the covid-19 pandemic: predictive role of personal resilience, coping behaviours, and health date: 2020-10-20 journal: nan doi: 10.1101/2020.10.18.20213942 sha: doc_id: 264821 cord_uid: 68us87xb background: the lockdown measures imposed by many countries since the onset of the covid-19 pandemic have been useful in slowing the transmission of the disease; however, there is growing concern regarding their adverse consequences on overall health and well-being, particularly among young people. to date, most studies have focused on the mental health consequences of the lockdown measures, while studies assessing how this disease control measure influences the occurrence of fatigue are largely absent. aim: the aims of this study are two-fold: (a) to examine the levels of lockdown fatigue, and (2) to determine the role of coping behaviours, personal resilience, psychological well-being and perceived health in fatigue associated to the lockdown measure. methods: this is an online cross-sectional study involving 243 college students in the central philippines during the sixth month of the lockdown measure implemented due to the covid-19 pandemic. five standardised scales were used to collect the data. results: overall, college students reported moderate levels of lockdown fatigue, with a mean score of 31.54 (out of 50). physical exhaustion or tiredness, headaches and body pain, decreased motivation and increased worry were the most pronounced manifestations of fatigue reported. gender and college year were identified as important predictors of fatigue. increased personal resilience and coping skills were associated with lower levels of lockdown fatigue. conclusion: college students experience moderate levels of fatigue during the mandatory lockdown or home confinement period. resilient students and those who perceive higher social support experience lower levels of fatigue during the lockdown period compared to students with low resilience and social support. lockdown fatigue may be addressed by formulating and implementing interventions to enhance personal resilience and social support among college students. the covid-19 pandemic is a global health issue that has significant health and economic implications. since its emergence in china in november 2019, the disease has infected over 34.8 million people worldwide, claimed at least 1 million lives, and been reported in 215 countries or territories (world health organization [who] , 2020). among the nations around the world, the united states of america, brazil, india, and russia remain the most affected, together comprising approximately 30% of the overall confirmed cases of coronavirus. in the western pacific region, the philippines recorded the highest number of confirmed cases and deaths, with more than 50% of the cumulative cases and 40% of the cumulative deaths (who, 2020) . in an effort to mitigate the transmission of the coronavirus, many countries around the world have adopted various disease control measures, including strict social distancing and mandatory lockdown or stay-at-home orders (ren, 2020; singh et al., 2020) . in the philippines, the government imposed a nationwide mandatory lockdown, also referred to as 'community quarantine', starting in march 2020, forcing people to stay home and restricting all forms of physical and social activities outside the home, with exceptions made for frontline and essential workers. in addition, schools were physically closed in mid-march 2020 and remain closed as of this writing, with remote teaching and learning environments being implemented as a temporary solution. these measures, along with other disease control strategies, were found to effectively reduce the number of confirmed cases and deaths associated with covid-19 in the country (department of health, 2020) , as well as in other countries (fowler et al., 2020; chen et al., 2020) . though the lockdown policies effectively mitigated or slowed the transmission of the coronavirus disease, they have adversely affected people's way of life, with serious consequences for mental and psychological health and well-being, particularly among young people (volkan & volkan, 2020; marroquín et al., 2020) . according to the australian psychological society (2020), prolonged lockdown may cause fatigue or exhaustion in an individual as a result of the overwhelming disruptions on their routines and activities, social isolation, lack of security, imminent threat to health and unpredictability of what is ahead, and may manifest as a mix of physical, mental and/or emotional signs. though fatigue is subjective, it is generally an undesirable experience in which an individual is engulfed with an overpowering sense of tiredness that is not relieved by rest or food intake, intense yearning to rest, lack of physical and mental energy and decreased motivation and sense of enjoyment (trendall, 2001) . it diminishes an individual's ability to function normally on a daily basis and may consequently lead to a decreased quality of life (ream & richardson, 1996) . previous research has provided compelling evidence of lockdown-related fatigue among australian citizens after a few months of the nationwide lockdown mandate (nitschke et al., 2020) which appears to worsen as time passes (meo et al., 2020) . manifestations of lockdown-related fatigue included sadness, physical exhaustion, reduced interest in previously enjoyed activities, emotional outbursts and anxiety and fear (australian psychological society, 2020). other signs indicating increasing fatigue during the lockdown period included tiredness (jiao et al., 2020) , sleep disturbance (majumdar et al., 2020 ), uncertainty, loneliness (singh et al., 2020 ), irritability (jiao et al., 2020 , fear and increased worry (dangi et al., 2020) , lack of motivation (kapasia et al., 2020) and loss of interest in previously enjoyed activities (margaritis et al., 2020) . young people such as college students are particularly vulnerable to the adverse mental and psychological health consequences of the stay-at-home orders or lockdown measures, as they pose a potential threat to their physical, mental and emotional health as well as their educational and developmental progress (singh et al., 2020) . evidence has shown significant increases in the prevalence of mental issues such as anxiety, depression, and psychological distress (husky et al., 2020; al omari et al., 2020) and symptoms of physical exhaustion, including tiredness, headaches, insomnia, fatigue and muscle pain (branquinho et al., 2020; majumdar et al., 2020) , in young people during the mandatory lockdown period. hence, measures should be implemented to better support young people during the pandemic in order to reduce the ill effects of the lockdown on their mental, psychological and physiological wellbeing. positive coping skills and personal resilience are key factors that may protect an individual from lockdown-induced fatigue and other mental and psychological health consequences of the pandemic and the measures implemented to control the disease. personal resilience is important for successful recovery from difficult or stressful circumstances (hart, brannan, & de chesnay, 2014) , while coping skills are helpful to resolve or hasten the resolution of a problem (piergiovanni, & depaula, 2018) . in the context of a pandemic, adequate personal resilience and coping skills are vital to help an individual cope with the negative effects of the . cc-by-nd 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) pandemic and support their mental health (labrague & de los santos, 2020) . studies have shown that individuals with poor coping skills (liang et al., 2020) and a negative mind set characterised by excessive worrying, hopelessness and pessimism (moore et al., 2020) are at higher risk for developing mental and psychological issues related to the pandemic, possibly including lockdown-induced fatigue. conversely, prior reports associated adequate coping skills and personal resilience with improved mental health and reductions in psychological issues such as loneliness, anxiety, depression and stress across populations during the height of the coronavirus pandemic (ye et al., 2020; elmer et al., 2020) . strengthening resilience and enforcing healthier coping skills may therefore help an individual combat fatigue related to the lockdown or home confinement measures and other stressors associated with the inevitable changes brought about by the pandemic. despite evidence showing the increased tendency of young people to develop fatigue related to lockdown measures, no studies examining how individual resilience and coping skills reduce fatigue in college students have yet been conducted. therefore, this study was conducted to examine the levels of lockdown-induced fatigue and its association with personal resilience and coping skills in college students. a cross-sectional study utilising an online data collection approach was conducted during the sixth month of the mandatory lockdown implemented in the philippines due to the coronavirus pandemic. . cc-by-nd 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 october 20, 2020. . https://doi.org/10. 1101 /2020 this study included college students enrolled in different colleges and universities in western samar, philippines. using the g*power program software, an estimation of required sample size was performed. a sample size of 222 was found to be required for five predictors to attain an 80% power, with an effect size of 0.05 and alpha set at 0.05 (soper, 2020) . three hundred students were initially invited; however, only 243 responded to our online survey. to qualify for the study, students had to: a) be currently enrolled in a college or university, b) be a full-time student, c) be either male or female, and d) consent to participate in the study. five standardized scales were used to gather data including the lockdown fatigue scale (lfs), brief resilience scale (brs; smith et al., 2008) , coping behaviors questionnaire (cbq; carver et al., 1997) , and a single-item measure of general health. lockdown fatigue scale. this scale was used to evaluate signs of exhaustion associated with the lockdown or home confinement measures to slow the spread of coronavirus. the lfs was designed by the researcher based on an extensive review of the literature and structured interviews of 15 individuals who were affected by the mandatory lockdown during the pandemic. the 10-item scale was answered by the participants on a 5-point likert-type scale that ranged from 1 (never) to 5 (always). the highest possible score was 50, and the scores were categorised as indicating low (1-12), mild, (13-25), moderate (25-37), and high or severe (38-50) fatigue. the internal consistency value of the scale in the present study was 0.84. the content validity of the scale was 0.934 and the test-retest reliability was 0.913. . cc-by-nd 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 october 20, 2020. . https://doi.org/10. 1101 /2020 brief resilience scale. this scale determined students' ability to bounce back from traumatic or unpleasant events associated with the pandemic and the imposed lockdown measure. nurses answered the scale by responding to a 5-point likert-type scale ranging from 0 (does not describe me at all) to 5 (describes me very well). previous research found optimal validity and reliability of the scale (labrague & de los santos, 2020; smith et al., 2008) , and in the current study, the internal consistency value of scale was 0.90. coping behaviours questionnaire. this scale examined the ways college students coped during the mandatory lockdown period. the scale comprised 8 items that were categorised into four dimensions: seeking information and consultation, use of humour, mental disengagement and spirituality/sources of support. participants answered the items using a 5-point likert-type scale that ranged from 1 (strongly disagree) to 5 (strongly agree). previous research (savitsky et al., 2020) established optimal criterion validity and excellent reliability of the scale, reporting an internal consistency value of 0.85. the internal consistency value for this scale obtained in the present study was 0.89. health. this single-item measure of general health was used to assess the overall personal health of the college students. participants were asked to rate their overall health using a 5-point likert-type scale (1 = poor, 5 = excellent). the test-retest reliability value of the item in the present study was 0.91, which was higher than the value previously reported (α = .89; labrague et al., 2020). the institutional research ethics committee of samar state university, philippines (irerc ei 0123 i) granted the ethical clearance for this study. since the schools were closed . cc-by-nd 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 october 20, 2020. . https://doi.org/10. 1101 /2020 during the data collection period, an online survey was created using google forms and sent to email addresses of the students within the province. basic information about the study, along with the letter seeking their consent, were contained in the introductory page of the online form. to ensure the anonymity of the participants, names were not requested during submission. the online survey was conducted for a period of one month from august to september 2020, which corresponds to the sixth month of the mandatory lockdown measure in the philippines. followup emails were sent to students on a weekly basis to remind them to complete the survey. data completeness was checked before entering data into spss version 25. to quantify the data, we calculated frequencies, standard deviations, and weighted means. bivariate analysis was facilitated using the independent t-test, pearson's correlation coefficient (r) and analysis of variance to examine correlations between key study variables. bonferroni's test was used for post hoc analysis. variables that yielded significant correlations with the outcome variable were entered into the multiple linear regression. the level of statistical significance was set as p < 0.05. two hundred forty-three college students from different schools and colleges in the region were invited to be part of the study. the average age was 20.77 years, with a standard deviation of 2.66 years. the majority of the participants were female, and more than half were in their first and second years of college education. more than half of the participants were enrolled in public schools in urban areas (table 1) . the mean scale scores for the personal resilience and . cc-by-nd 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 october 20, 2020. . https://doi.org/10.1101/2020.10.18.20213942 doi: medrxiv preprint psychological well-being measures were 3.949 and 5.377, respectively. for the perceived general health and coping skills measures, the mean scale scores were 3.843 and 3.818, respectively. table 2 shows the responses of the participants on the lfs. the mean scale score of the lfs was 31.54 (sd: 6.930) out of a maximum possible score of 50. among the different items on the scale, the items 'i worry a lot about my personal and family's safety during this pandemic', 'i frequently felt weak or tired as a result of this lockdown', and 'i have been nervous or anxious' obtained the highest mean values. the items that obtained the lowest mean values were 'i have been feeling irritable', 'i have been experiencing a general sense of emptiness', and 'i have difficulty falling or staying asleep over thinking about this pandemic' ( table 2) . table 3 , several of the variables correlated significantly with the lockdown fatigue. an independent t-test showed a significantly higher mean score on the lfs in female compared to male students. further, analysis of variance showed a significant difference in the lfs mean score in participants grouped according to level of education, and post hoc analysis using the bonferroni test showed significantly higher mean scores in the lfs in firstyear and third-year students compared to fourth-year students. pearson's correlation coefficient showed a significant negative relationship between personal resilience and lockdown fatigue. a similar pattern was observed between coping skills and lockdown fatigue. variables that were significantly correlated with the outcome variable were entered into the multiple linear regression model (table 4 ). the model explained 15.7% in the variance of the lfs, which was statistically significant (f = 4.130, p < 0.001). among the different variables, gender and level of education predicted lockdown fatigue, with female students (β = -0.122, p = 0.047) and those in the lower levels reporting an increased lockdown fatigue. further, increased scores on the personal resilience (β = -2.295, p = 0.023) and coping skills (β = -2.045, . cc-by-nd 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 october 20, 2020. . https://doi.org/10.1101/2020.10.18.20213942 doi: medrxiv preprint p = 0.042) measures were associated with a significant decrease in scores on the lockdown fatigue measure. the current study examined the extent of fatigue experienced by college students during the mandatory covid-19 lockdown period and the influence of students' demographic variables, personal resilience, coping skills, psychological well-being and perceived general health in the development of lockdown fatigue. the mean scale score of the lockdown fatigue measure was 31.54 (sd: 6.930) out of a possible score of 50, suggesting a moderate level of lockdown fatigue in the sample studied. due to the absence of a similar tool to measure fatigue during the mandatory lockdown period, comparing and contrasting our study findings with previous studies is not possible. however, this result was in line with that of a previous study by nitschke et al., (2020) , who, using the chalder fatigue questionnaire observed a significant level of fatigue in australian citizens a few months after the mandatory lockdown was enacted. using google trends to examine the effects of the home confinement measures implemented in europe and america, brodeur et al., (2020) found compelling evidence of substantial increases in sadness, boredom, worry, loneliness and fatigue in the general population from the initial weeks until the fourth months of the implementation of the measures. reports from india, the usa and saudi arabia also showed substantial evidence that individuals become increasingly tired and fatigued as time lapses, suggesting that efforts should be made to effectively support this group of individuals and to prevent the adverse consequences of prolonged lockdown or home confinement (meo et al., 2020; majumdar et al., 2020) . as higher levels of fatigue may adversely affect the physical, . cc-by-nd 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 october 20, 2020. . https://doi.org/10.1101/2020.10.18.20213942 doi: medrxiv preprint mental, behavioural and cognitive functions of an individual (trendall, 2001) , it is critically important to develop strategies to address this issue through evidence-based approaches. government planners should periodically review the effectiveness of the lockdown measures being implemented and consider ways to ease the measure without compromising the health of the population. among the different manifestations of fatigue, the participants in this study reported tiredness or physical exhaustion, lack of motivation, worry or fear and anxiety as the most pronounced symptoms. the reported symptoms of lockdown fatigue in this study were similar to those previously identified by the australian psychological society (2020), which included sadness, physical exhaustion, reduced interest in previously enjoyed activities, emotional outbursts and anxiety and fear. this result is similar to that of a study by majumdar et al., (2020) in which indian professionals and students exhibited various indicators of fatigue, including tiredness, higher stress and anxiety levels and increased worry for their personal security and the safety of their families, after a few months of the home confinement measure. regression analysis identified gender as an important predictor of lockdown fatigue, with female students experiencing an increased level of fatigue compared to male students. this result should be interpreted with caution due to the disparity in the proportion of male and female participants in this study. nevertheless, this result may indeed be due to gender disparity with regards to expression of feelings and emotions, including worry, fear, sadness and anxiety, and even in their expression of pain and bodily discomfort. mounting evidence has shown that men tend to suppress their emotions and feelings, while women are more vocal when expressing their emotions (chaplin et al., 2008; tolin & foa, 2008) . this result is a corroboration of the longstanding gender stereotype within the philippine culture in which the expression of thoughts, . cc-by-nd 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 october 20, 2020. . https://doi.org/10.1101/2020.10.18.20213942 doi: medrxiv preprint feelings and emotions is more acceptable for women than for men. in addition to fatigue, previous evidence has also shown that women had a higher inclination to develop other mental and psychological issues such as stress disorders, major depression and anxiety and panic disorders than men during the height of the covid-19 pandemic (elmer et al., 2020; pouralizadeh et al., 2020) . this result suggests the dire need for the implementation of gendertailored strategies to effectively manage the adverse impact of the lockdown measure and reduce fatigue. results of this study differ from those of the study by nitschke et al., (2020) , in which gender did not contribute to the development of fatigue in australian citizens during the pandemic. another important finding was the direct influence of college students' year level on the development of lockdown fatigue. in particular, graduating students reported decreased levels of fatigue compared to students in the lower levels. this result was expected, as during the course of their education, students acquire adaptive behaviours, positive coping abilities and higher resilience (benner, 2004) that are vital when confronted with stress-inducing situations such as the coronavirus pandemic. previous studies have demonstrated a significant decline in stress levels and marked improvement in coping abilities in college students as they progress to the higher levels of education (kumar & nancy, 2011; fornés vives et al., 2016) . this finding calls for a greater need to support college students, particularly those earlier in their college careers, through relevant interventions to improve their coping skills and personal resilience so that they can effectively handle the mental, physical and psychological consequences associated with home confinement measures or lockdown. regression analysis also revealed a significant negative association between personal resilience and lockdown fatigue, suggesting the protective role of individual resilience against . cc-by-nd 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 october 20, 2020 . . https://doi.org/10.1101 /2020 the consequences of the mandatory home confinement measure. in other words, resilient students are less likely than non-resilient students to experience fatigue during the lockdown period. to our knowledge, this study is the first to report a causal relationship between personal resilience and fatigue associated with the lockdown measure, hence adding new knowledge to nursing science. increasing individual resilience has been shown to be an important strategy to help an individual bounce back from adversity when faced with various stressors and stress-inducing events and traumatic situations (cooper et al., 2020) . our result is in accordance with earlier studies that linked personal resilience with positive psychological and mental health outcomes across populations during the height of the coronavirus pandemic (ran et al., 2020; ye et al., 2020; labrague & de los santos, 2020) . this highlights the relevance of instituting interventions to foster personal resilience in students in order to reduce the occurrence of lockdown-related fatigue and other negative mental and psychological consequences associated with the pandemic. in the current study, students who reported higher coping skills reported having significantly lower levels of lockdown fatigue. adequate coping skills have been identified in the literature as a vital defence for an individual, offering long-term stress reduction effects during stressful or traumatic situations (labrague & mcenroe-petitte, 2018) . previous studies involving college students have also identified problem-focused coping behaviours, including seeking social support, and problem-solving behaviours as equally vital to increase their adaptability and hardiness against stressful events (labrague et al., 2017; farrell & langrehr, 2017) . further, adequate coping skills have been found to minimise the mental and psychological consequences of a traumatic events, emergency and disaster events and disease outbreaks (labrague & de los santos, 2020; hou et al., 2020) . higher levels of coping skills were found to contribute to a significant reduction in psychological issues (e.g. stress, anxiety, depression) related to the . cc-by-nd 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 october 20, 2020. . https://doi.org/10.1101/2020.10.18.20213942 doi: medrxiv preprint covid-19 pandemic among college students in china (cao et al., 2020) , the usa (tull et al., 2020) and switzerland (elmer et al., 2020) . in a recent study involving college students, high levels of fatigue due to social distancing measures were attributed to lower social connectedness with peers and friends and lower coping skills (nitschke et al., 2020) . it is therefore vital that measures towards reducing lockdown fatigue among college students be focused on strengthening their coping skills, thus improving their mental and psychological well-being and overall health. increasing communication and connections with friends and family is essential to reduce the negative impact of home confinement, and this can be accomplished with the aid of technology or social media. this study has several limitations that should be considered when interpreting the findings. first, due to the design of the study, it is impossible to establish causality between students' variables and lockdown fatigue. second, since most of the study variables are dynamic (e.g. fatigue, resilience, health) and may therefore change over time, it is important to use longitudinal research designs in future studies. third, to improve the generalisability and representativeness of the study, future studies should include more samples from other areas of the country. finally, the use of self-reported scales is a possible limitation of the study, as it may cause response bias. mandatory lockdown or home confinement measures to slow the transmission of covid-19 may cause considerable levels of fatigue in college students. female students, as well as those in the lower levels of education, were found to experience more fatigue than male and graduating students. further, this study provided empirical evidence linking higher personal . cc-by-nd 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 october 20, 2020. . https://doi.org/10. 1101 /2020 resilience and social support with decreased levels of lockdown-induced fatigue in students. strategies to manage or reduce lockdown fatigue among college students should consider the factors identified in order to effectively address this growing problem among this group of the population during the coronavirus pandemic. future studies testing the efficacy and effectiveness of interventions to reduce fatigue in college students should be undertaken. . cc-by-nd 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 october 20, 2020 . . https://doi.org/10.1101 /2020 singh, s., roy, d., sinha, k., parveen, s., sharma, g., & joshi, g. (2020) . impact of covid-19 and lockdown on mental health of children and adolescents: a narrative review with . cc-by-nd 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 october 20, 2020 . . https://doi.org/10.1101 /2020 . cc-by-nd 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 october 20, 2020. . https://doi.org/10. 1101 /2020 . cc-by-nd 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 october 20, 2020. . cc-by-nd 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 october 20, 2020. . https://doi.org/10. 1101 /2020 causal estimation of stay-at-home orders on sars-cov-2 transmission you want to measure coping but your protocol'too long: consider the brief cope nurse resilience: a concept analysis gender differences in response to emotional stress: an assessment across subjective, behavioral, and physiological domains and relations to alcohol craving stress level and coping strategies among youth during coronavirus disease lockdown in india students under lockdown: assessing change in students' social networks and mental health during the covid-19 crisis coping, stress, and personality in spanish nursing students: a longitudinal study stress, social support, and psychosocial functioning of ethnically diverse students the effect of stay-at-home orders on covid-19 infections in the united states resilience in nurses: an integrative review stress and anxiety among university students in france during covid-19 mandatory confinement how to deal with covid-19 epidemic-related lockdown physical inactivity and sedentary increase in youth? adaptation of anses' benchmarks impact of the covid-19 virus outbreak on movement and play behaviours of canadian children and youth: a national survey lockdown is associated with reduced distress and fatigue anxiety and depression and the related factors in nurses of guilan university of medical sciences hospitals during covid-19: a web-based crosssectional study self-efficacy and stress coping styles in university students. recommendations the brief resilience scale: assessing the ability to bounce back a-priori sample size calculator for multiple regression concept analysis: chronic fatigue sex differences in trauma and posttraumatic stress disorder: a quantitative review of 25 years of research psychological outcomes associated with stay-at-home orders and the perceived impact of covid-19 on daily life under the covid-19 lockdown: rapid review about the unique case of north cyprus coronavirus disease (covid-19) weekly epidemiological update resilience, social support, and coping as mediators between covid 19 related stressful experiences and acute stress disorder among college students in china. applied psychology: health and well being key: cord-219681-83p1ujl4 authors: gathergood, john; guttman-kenney, benedict title: the english patient: evaluating local lockdowns using real-time covid-19&consumption data date: 2020-10-08 journal: nan doi: nan sha: doc_id: 219681 cord_uid: 83p1ujl4 we find uk 'local lockdowns' of cities and small regions, focused on limiting how many people a household can interact with and in what settings, are effective in turning the tide on rising positive covid-19 cases. yet, by focusing on household mixing within the home, these local lockdowns have not inflicted the large declines in consumption accompanying the march 2020 national lockdown, which limited all social contact. our study harnesses a new source of real-time, transaction-level consumption data that we show to be highly correlated with official statistics. the effectiveness of local lockdowns are evaluated applying a difference-in-difference approach which exploits nearby localities not subject to local lockdowns as comparison groups. our findings indicate that policymakers may be able to contain virus outbreaks $textit{without}$ killing local economies. however, the ultimate effectiveness of local lockdowns is expected to be highly dependent on co-ordination between regions and an effective system of testing. how can covid-19 cases be contained without causing damage to the economy? this question dominates the thinking of policymakers, who face a seemingly uncomfortable trade-off between limiting virus transmission in the economy via reducing social contact, and maintaining economic activity which relies on social contact for the production and consumption of goods and services. the first wave of covid-19 in early 2020 saw most nations adopt stringent restrictions on social contact in almost all settings in order to contain the spread of the virus. these restrictions severely hindered the means of production and consumption in the economy, leading to large drops in output. however, recent improvements in testing and tracing leading to identification of clusters of cases in high-infection areas have facilitated a more targeted, localised approach to applying restrictions to social contact, known as "local lockdowns". this approach to limiting the spread of covid-19 was identified early-on in the pandemic as a beneficial strategy. 1 in some countries -such as in the ukgovernments have legal powers to implement such local measures, but this has been a source of political tension between local and national authorities. 2 the centralized uk approach offers a particularly interesting contrast to the us, where the policy response has limited national co-ordination -and none in regards to lockdowns. in this paper, we are the first academics to use a new source of real-time and highly granular european consumption data. we combine these with data on coronavirus cases to analyse the impact of local lockdowns on both covid-19 cases and local consumption. using a difference-in-difference methodology, we estimate the impact of local lockdowns imposed in the late summer of 2020 on a number of uk cities, examining the effects on both containment of cases and consumer spending. our study makes two main contributions. first, we introduce a new source of real-time, transaction-level consumption data -fable data -that can be used for economic research and to inform policymaking. these data contain transaction-by-transaction spending data, updated daily, for large representative samples of uk bank accounts and credit cards, with individual-level identifiers and geocode identifiers. we show these data are a highly correlated, leading indicator of official bank of england statistics -data that are only available in aggregated form and with many months lag -in contrast to fable data which are available in real-time and disaggregated (correlation coefficient of 0.91 january 2018 june 2020) . these data are applicable to a broad variety of questions in the analysis of individual consumption behavior. they present a new opportunity for researchers to measure consumption in arguably more reliable ways than using data from consumption surveys, which has become less reliable in recent decades and has prompted a variety of initiatives aimed at improving the measurement of consumption (see browning et al., 2014; landais and spinnewijn, 2020) . these data show the uk's economic recovery in spending april to august 2020 had stalled in september and october varies across regions and urban geographies. second, we advance understanding of the economic costs of mitigation strategies to contain the second wave of covid-19. this use-case is economically important and policy-relevant to the time of writing, as policymakers around the world are grappling with how to both contain second or third waves of virus outbreaks and also to keep economic recoveries going. local lockdowns are also a source of tensions between national and local governments and thus our research may help to inform such disputes. the uk local lockdowns we study apply to cities or small regions. they restrict how many individuals people can mix with, in what settings (e.g. restaurants) and under what requirements (e.g. outside, wearing masks). a key policy design was trying to enable people to keep consuming in covid-19 secure settings while limiting their interactions in less secure settings (e.g. households visiting each others' homes). we show that uk local lockdowns to contain virus outbreaks -covering one in four people by september 2020 -3 typically turn the tide on rising covid-19 cases. 3 they do so without the large declines in consumer credit card spending that we observed accompanying the first virus wave and national lockdown in early 2020. we reach this conclusion using a difference-in-difference design that compares the evolution of daily consumption in an area subject to a local lockdown compared to a similar, nearby locality not subject to such restrictions. our difference-in-difference approach is designed to be interpreted as descriptive not causal. 4 these data show -both for treatment and control localities -large drops in consumption when the march 2020 first virus wave and national lockdown occurred. we conclude there are little, if any, declines in spending from the local lockdowns. estimates for the time-path of cases, in contrast, show that while covid-19 cases typically continue to rise following the onset of a local lockdown (as measures take time to have effect) they then start to stabilize: indicating the local lockdowns had some short-term success. while our evidence indicates some initial successes from local lockdowns, in late september and october 2020 the uk (along with many european countries) experienced a rapid, nationwide rise in covid-19 cases (the cause of which is not yet clear). this has led to more restrictive local measures being introduced in mid-october including shutting down local businesses and active discussions on a return to another national lockdown. whether this new range of more restrictive national and local measures being imposed and considered result in changes to consumption more like the first national lockdown or the local lockdowns we study here will have profoundly different implications for the uk economy's prospects and other countries facing similar secondary virus waves. our study contributes to a burgeoning literature understanding the economic effects of covid-19. a variety of early studies showed how the onset of covid-19 dramatically changed consumption behavior. the first study to do so was baker et al. (2020) using us fintech data and following this opportunity insights (chetty et al., 2020a,b) produced 3 https://www.bbc.com/news/uk-england-52934822 4 we typically observe common pre-trends between control and treatment groups, however, we do observe noticeable increases in the number of positive covid-19 cases for the treatment groups just before and after local lockdowns. the nature of local lockdowns explain this behavior -a key component is to increase testing capacity and thus the number of positive cases will be expected to rise. however, some areas subject to local lockdowns had rises that appear too large and sharp to be driven by differential testing and in such cases the control localities are less suitable counterfactuals. a dashboard using multiple data sources to track regional us consumption behavior alongside other economic indicators. 5 beyond the us similar exercises have been carried out to understand household consumption in the early stages of the pandemic -showing remarkably consistent results (andersen et al., 2020; bounie et al., 2020; bourquin et al., 2020; campos-vazquez and esquivel, 2020; carvalho et al., 2020; chen et al., 2020; chronopoulos et al., 2020; horvath et al., 2020; surico et al., 2020; watanabe et al., 2020) . analysis of jp morgan chase data (cox et al., 2020; farrell et al., 2020) (aum et al., 2020; beach et al., 2020; barro et al., 2020; coibion et al., 2020; correia et al., 2020; cui et al., 2020; dave et al., 2020; friedson et al., 2020; hacioglu et al., 2020; glover et al., 2020; guerrieri et al., 2020; hall et al., 2020; lilley et al., 2020; miles et al., 2020; jones et al., 2020; toxvaerd, 2020; wang, 2020) . we combine data on cases of covid-19 identified by the uk's testing framework with consumption data provided by fable data limited. 6 on current (checking) accounts. data is at the account-level and hence we can follow spending behavior on an individual account over time. 8 fable data is similar to recently-available data sets from financial aggregators and service providers, but does not have some of the limitations of other datasets and uses anonymised customer data. 9 for each spending transaction we observe a standard classification merchant category code for the spending type. fable also produces its own categorizations of spending, utilizing the more granular information it has available from transaction strings. these data also differentiate between online and store-based transactions. for each uk account we observe the postcode sector of the cardholder's address. in the uk, postcode sectors are very granular geographies: there are over 11,000 postcode sectors in the uk with each sector containing approximately 3,000 addresses. where a transaction can be linked to a particular store, the full address of that store is available. where a transaction is of a listed firm, fable tags merchants to their parent groups and stock market tickers. for this study we focus on transactions denominated in british pounds sterling on uk-based credit card accounts held by consumers. 10 the median and mean transaction values are £15 and £39 respectively. transaction-level spending data is highly volatile -even with such large volumes of transactions -and we observe strong movements at high frequency due to seasonality and day of week effects. we therefore follow an approach to smooth the transaction volumes over time as used by opportunity insights on similar us data (chetty et al., 2020b,a) : aggregating spending by day at the level of geography of interest, taking a seven day moving average and dividing by the previous year's value. 11 finally, we normalize the series setting an index to 1 using the mean value 8 -28 january 2020. we also construct 8 in cases where one individual has multiple accounts, we cannot link multiple accounts in the data to the individual but can aggregate to a geographic region. 9 baker (2018) provides validation and application of us financial aggregator data. financial aggregator data for the uk is widely shared for research purposes by moneydashboard, uk-based a fintech (chronopoulos et al., 2020; bourquin et al., 2020; surico et al., 2020) . bourquin et al. (2020) analyse the characteristics of moneydashboard users. 10 we drop 113 individual credit card transactions over £50k as such outliers are unlikely to be consumer transactions and may distort results for very small geographic regions. 11 for 29 february 2020 we divide by an average of 28 february and 1 march 2019. 6 daily series using a 14 and 28 day moving averages in a analogous fashion. fable data have many useful features, such as timeliness (it is available the next working day whereas official statistics are typically available only with a lag of several months), geographic granularity (being available at a lower level than official statistics) and, transaction-level (enabling more flexible analysis than aggregated official statistics). these data can therefore potentially be use to construct leading indicators for policymakers and enable researchers to answer a broader set of research questions than was previously possible using prior data sources. however, while these features are potentially valuable, their usefulness depends in part on how this data series relates to comprehensive, official data. to explore this, figure a1 , panel a compares the time series of fable data uk annual changes in monthly credit card spending to the bank of england series and shows they are highly correlated: correlations 0.91 (january 2018 to july 2020), 0.90 (january 2019 to july 2020) and 0.98 (january 2020 to july 2020). bank of england data is only published in aggregated form monthly and with a lag (e.g. july's data was published at the start of september). figure a1 , panel b shows fable data measures for 7, 14, 28 day moving averages -which can be calculated daily in real-time -compared to the monthly series (which requires waiting until month end). these daily moving averages show the sharp drop in consumption in march 2020 far earlier than the monthly series. we thus conclude that we can use these data as a reliable real-time predictor of official data and as a reasonable proxy for measuring credit card spending. on aggregate we observe the sharp fall in uk credit card spending near the time of the spike in covid-19 cases and national lockdown announcement on 23 march 2020 and then a fairly steady recovery may -august. components of the national lockdown were ended in june and july but we do not observe rapid boost after these was lifted -indicating spending may have been suppressed by fear of the virus during this early phase of the pandemic. in september and october 2020, we observe these data to show 7 that the recovery in spending since april 2020 has stalled and has started to decline. to understand the potential of these data further, and the heterogeneous impacts of the covid-19 crisis, we disaggregate the national series by urban geographies. figure a2 disaggregates by eight urban-rural categories created by the uk national statistics agency -the office for national statistics (ons). 12 the figures illustrate that recovery has been fastest and strongest in 'business, education and heritage centres' -such areas are popular domestic tourist destinations and thus this is in line with consumers substituting foreign for domestic holidays. recovery was less strong in 'countryside living' -predominately rural areas but still noticeably stronger than other, more urban areas. for more urban areas, london has had a steady recovery whereas 'affluent england', 'services & industrial legacy', and 'urban settlements' are showing weaker recoveries. we use public data released by the uk government which details the areas affected by local lockdowns, including the dates of introduction and cessation of lockdown measures. each week public health england publishes a covid-19 surveillance report that includes 'the watchlist' showing the incidence (and trend) of covid-19 in local government areas (lower tier local authorities), whether household mixing is prohibited and lists areas on the watchlist. 13 scotland, northern ireland and wales have comparable data that we gather. 14 areas are added to the watchlist considering a variety of metrics using professional judgment of uk public health officials according to the uk government's covid-19 contain framework'. 15 areas on the watchlist fit into one of three categories: • concern areas -local area is taking targeted actions to reduce covid-19's prevalence (e.g. additional testing in care homes and increased community engagement with high risk groups). • enhanced support areas -more detailed plan agreed with the national team and with additional resources being provided to support the local team to control covid-19 (e.g. epidemiological expertise, additional mobile testing capacity). • intervention areas ('local lockdowns') -divergence from the measures in place in the rest of england because of the significance of covid-19's spread, with a detailed action plan in place, and local resources augmented with a national support to control covid-19. for this research we focus on local lockdowns. by september 2020, approximately one in four people in the uk were subject to a local lockdown. a key feature of such lockdowns is imposing restrictions on household mixing (e.g. preventing a tea party in someone's house) but permitting visits to more covid-19 secure settings (e.g. having tea outdoor at a restaurant with strict hygiene and social distancing standards) in order to encourage consumers to keep spending while also trying to contain the virus. across the local lockdowns there was variation in the how much household mixing was restricted (e.g. in caerphilly residents were not supposed to leave nor new people come in). 16 there were also nationwide (including in areas subject to local lockdowns) government financial incentives to encourage consumers to spend for much of the period of time we study. the most notable of these were cuts to sales taxes (value added tax, vat) on food, accommodation and attractions from 20% to 5% from july 2020 until january 2021 and 'eat out to help out' scheme offering a 50% discount (up to £10 per person on food and non-alcoholic drink) for eating out mondays, tuesdays and wednesdays during august 2020. we use a difference-in-difference methodology to estimate the relationship between local lockdowns and daily consumption. our estimates provide a description of the evolution of covid-19 cases and consumer spending pre-and post-lockdown on lockdown affected cities and comparison areas. we do not interpret our estimates as showing causal relationships -our comparison cities are not perfect counterfactuals for the evolution of covid-19 cases or consumer spending in the absence of a lockdown. table a1 lists the timing of local lockdown announcements, the local authorities affected and the control group localities to compare against. where there are multiple localities in the same area subject to a lockdown announcement on the same day we aggregate data (e.g. south and north lankarkshire to lanarkshire). for some areas subject to local lockdowns (e.g. belfast) no suitable control group city exists. we display descriptive results for thirteen pairs of treatment and control localities -but primarily focus on the manchester lockdown as that has a large sample, liverpool offers a good control and the case is particularly informative for considering a potential london lockdown. 17 16 https://www.bbc.com/news/uk-england-52934822 17 other areas studied that were subject to local lockdowns are: aberdeen, birmingham, bolton, caerphilly, glasgow, greater glasgow, lanarkshire, leeds, leicester, newcastle, preston and wolverhampton. we first describe the time series for spending in these regions. a standard differencesin-differences approach uses a parsimonious regression estimation approach, such as that presented in (equation 1). the outcome of interest y g,t is offline credit card spending measured as changes in an index of yearly changes in 7 day moving average of spending. the unit of observation is a day (t) for each local authority group (g) either subject to the local lockdown or the control group to compare it to. to explain our methodology we draw upon the example of manchester. manchester is the treated group (t reat g = 1) subject to the local lockdown and we use a 'similar area' (one of comparable size and in the same part of the country) -liverpool -as a control group (t reat g = 0). af ter t is an indicator equal to one if the time period is after the local lockdown announcement. (t reat g * af ter t ) is the interaction of the above two terms: it is an indicator equal to one if the time period is after the local lockdown announcement and the area is in the treated group. the difference-in-difference estimation approach allows for these areas to have different time-invariant relationships with consumption (α for liverpool and α + β for manchester) and γ t is a series of daily dummies (with t = −1 omitted) to control for any common time-varying factors (e.g. national changes in covid-19 cases, economic policies). δ t provides an estimate for the relationship between local lockdown and consumption. to better understand the local lockdowns we modify equation 1 to estimate a dynamic specification creating weekly dummies (af ter w,g ) for the weeks preceding (w ∈ {−3, −2, −1, 0}) and following (w ∈ {1, 2, 3, 4}) the lockdown announcement where the omitted weekly dummy (w = 0) is the seven days preceding the local lockdown announcement (t = −7 to t = −1) and we use data up to four weeks pre-and post-lockdown (where available). we focus on a short time window given the volatile period we study the control groups are likely to become less suitable comparisons over time. we estimate this using an ols regression weighting observations by their 2019 resident population and cluster standard errors by region. 18 δ k (t reat g * af ter w,g,t ) + γ t + ε g,t we regard our estimation approach as providing informative, descriptive real-time evidence to inform policymakers. for these to be interpreted as causal effects for the effect of local lockdown on local spending would require a 'common trends' assumption that, for example, in the absence the manchester local lockdown, consumption in manchester would have followed the common trend to that in liverpool. 19 such an assumption is unlikely to hold in these data -particularly over longer time horizons there may be spillovers between areas and other government measures being introduced -hence we interpret our short-term results as helpful descriptive evidence which might strongly suggest a causal relationship, due to what is known more generally about virus transmission and the efficacy of social distancing measures. we first examine the effects of local lockdowns on covid-19 cases. the descriptive results are displayed in figure 1 . vertical dotted lines display the timing of the national lockdown in march 2020 (affecting both treatment and control groups) and the local lockdown (day 0) that only affects the (yellow) treatment areas not the (black) control localities. there are clear pre-lockdown differences in case data between lockdown areas and comparison areas, with lockdowns being introduced following a sharp increase in the case rate in each of the city graphs shown. 20 18 resident population estimates are from office for national statistics (ons). 19 see goodman-bacon and marcus (2020) for a fuller discussion of difference-in-differences to estimate causal effects in the higher pre-lockdown case rate in lockdown areas may partially reflect a policy choice by health authorities whereby more tests are purposefully carried out in areas the government is considering introducing local lockdown restrictions. this is consistent with what we observe in covid-19 positive we observe cases continue to rise after the lockdown announcement -as would be expected given the disease's incubation period -but typically find that following the lockdown announcement the rise in cases ceases, case numbers peak and then level-off or decline. this therefore indicates that local lockdowns can be effective at containing covid-19 outbreaks. we estimate the relationship between local lockdowns and the evolution of cases using a difference-in-difference regression model. while our estimates should be interpreted as descriptive, we limit the sample for lockdown cases for which the pre-lockdown data indicates that the comparison geography appears to be a relatively suitable control for the treatment locality. 21 this leads us to drop bolton and leicester as there were far sharper rises in covid-19 incidences before the lockdown than for their control groups. we also drop aberdeen and caerphilly as we have a relatively small number of transactions for these areas. thus we have a remaining sample of six local lockdowns to study: manchester, birmingham, glasgow, greater glasgow, newcastle and preston. the dynamic regression results in table 1 quantifies the relationship between local lockdowns and covid-19 cases. in each case, the coefficient on the local lockdown dummy the first week post-lockdown is positive, and in most cases remains positive after two weeks, consistent with local lockdowns being imposed as cases rise towards a spike. the coefficients at subsequent time horizons fall, in the case of manchester becoming negative in period one month after the imposition of he local lockdown cases as a percent of tests, however, such data are only available weekly and at a higher level of geographic region (upper rather than lower tier local authority) that does not align to areas subject to local lockdowns. 21 roth (2018) highlights the limitations of relying on common pre-trends tests as evidence for common trends assumption for estimates to be interpreted as causal. 13 findings are consistent across these measures. a broader set of treatment and control groups pairs are displayed for the same credit card spending measures in figures a6 (offline) and annex figures a5 (all) , 3 (food and beverage) and a7 (large store chains, 14 day moving average) -due to smaller sample sizes in some of these areas the series are more volatile but show consistent results. we highlight three features from these data. first, the treatment and control regions have similarly-timed and sized declines in spending in march 2020. second, the treatment and control regions typically have similar trends in consumption in the lead up to the local lockdown being announced. third, in contrast to the large spending declines observed for the national lockdowns, we observe comparatively small, if any, spending declines following the local lockdowns. this descriptive analysis therefore indicates that local lockdowns are not having the large negative effect on consumption that the first wave of the virus and national lockdown did. our dynamic regression results (table 2 ) also show this, however, we caveat it as in some weeks, in some lockdowns spending is lower. for example in manchester, it is no more so than the difference observed pre-lockdown and certainly nowhere near the spending declines accompanying the march 2020 national shutdown. we introduce a new real-time source of consumption data that we demonstrate is a highly correlated, leading indicator of official statistics and that also available at transaction level. we use these data to reveal the disparate impacts of the covid-19 pandemic on uk urban and regional geographies. our analysis studies how consumer spending responds to uk local lockdowns using a difference-in-difference approach, comparing nearby cities or small areas. in contrast to the large spending declines observed in march 2020, we do not find large spending declines in response to these local lockdowns. instead we find little (if any) decline in local spending. using the same difference-in-difference methodology we find that these local lockdowns typically appear to turn the tide on rising covid-19 positive cases. we thus conclude that it appears possible for policymakers to use such local lockdowns restricting household mixing to contain covid-19 outbreaks without killing local economies. however, we caveat this by noting that the effectiveness of such measures to mitigate the virus itself are expected to be highly dependent on an effective system of testing to isolate which regions to lockdown and contain infected individuals. it is also expected to depend upon co-ordination across regions by governments to ensure outbreaks in one area are contained and do not spillover into other areas. while our evidence indicates some initial successes from local lockdowns, the uk (along with many european countries) experienced a rapid, nationwide rise in covid-19 cases in late september and october following the end of the summer holidays, reopening of schools and universities (though the cause of the wave's sudden rise is not yet clear) indicating that its system of nationwide containment is not isolating cases early enough to be effective. at the time of writing a rapid, nationwide covid-19 outbreak has resulted in government imposing a new, more restrictive system of local lockdowns that includes forced business closures and there is public debate over imposing a potential circuit-breaker national lockdown. whether this new range of more restrictive national and local measures being imposed or considered result in changes to consumption more like the first national lockdown or the local lockdowns we study here will have profoundly different implications for the uk economy's prospects and other countries facing similar secondary virus waves. note: * p<0.05; * * p<0.01; * * * p<0.001 ols regression as specified in equation 2 with fixed effects for areas and days. areas are weighted by their 2019 resident population. standard errors clustered at area level. each column is for different area subject to local lockdown (with its nearby control area). daily data for 4 weeks pre and post local lockdown (3 weeks post for newcastle). outcome is fable data daily series index for 7 day moving average of offline consumer credit card spending. outcome is de-seasoned by taking ratio of 7 day moving average a year prior and indexed (=1)to its moving average 8-28 january 2020. omitted category is week (days -7 to -1) preceeding lockdown. notes: lower tier local authorities listed. this does not include areas below the local authority level (e.g. blaby, charnwood, carmarthenshire) where parts were locked down. 'control' lists lower tier local authorities chosen as control groups: blank where not used for analysis because region is small and/or no suitable control area exists. bolton announced and immediately introduced requirements on 5/9 but a full local lockdown was subsequently announced on 8/9. consumer responses to the covid-19 crisis: evidence from bank account transaction data covid-19 doesn't need lockdowns to destroy jobs: the effect of local outbreaks in korea debt and the response to household income shocks: validation and application of linked financial account data how does household spending respond to an epidemic? consumption during the 2020 covid-19 pandemic the coronavirus and the great influenza pandemic: lessons from the "spanish flu" for the coronavirus's potential effects on mortality and economic activity the 1918 influenza pandemic and its lessons for covid-19 consumers' mobility, expenditure and online-offline substitution response to covid-19: evidence from french transaction data the effects of coronavirus on household finances and financial distress the measurement of household consumption expenditures consumption and geographic mobility in pandemic times: evidence from mexico tracking the covid-19 crisis with high-resolution transaction data the impact of the covid-19 pandemic on consumption: learning from high frequency transaction data the economic impacts of covid-19: evidence from a new public database built from private sector data real-time economics: a new platform to track the impacts of covid-19 on people, businesses, and communities using private sector data consumer spending responses to the covid-19 pandemic: an assessment of great britain the cost of the covid-19 crisis: lockdowns, macroeconomic expectations, and consumer spending pandemics depress the economy, public health interventions do not: evidence from the 1918 flu initial impacts of the pandemic on consumer behavior: evidence from linked income, spending, and savings data covid-19, shelter-in place strategies and tipping were urban cowboys enough to control covid-19? local shelter-in-place orders and coronavirus case growth consumption effects of unemployment insurance during the covid-19 pandemic did california's shelter-in-place order work? early coronavirus-related public health effects health versus wealth: on the distributional effects of controlling a pandemic using difference-in-differences to identify causal effects of covid-19 policies covid-19 lockdown policies at the state and local level fear, lockdown, and diversion: comparing drivers of pandemic economic decline 2020 macroeconomic implications of covid-19: can negative supply shocks cause demand shortages? the distributional impact of the pandemic trading off consumption and covid-19 deaths the covid-19 shock and consumer credit: evidence from credit card data optimal mitigation policies in a pandemic: social distancing and working from home introduction to the special issue:"new perspectives on consumption measures public health interventions and economic growth: revisiting the spanish flu evidence living with covid-19: balancing costs against benefits in the face of the virus pre-test with caution: event-study estimates after testing for parallel trends consumption in the time of covid-19: evidence from uk transaction data equilibrium social distancing the financial impact of covid-19 in the united states online consumption during the covid-19 crisis: evidence from japan fable & bank of england monthly data fable data 7,14,28 day moving averages are the daily moving average de-seasoned by taking ratio of the moving average a year prior education & heritage sectors c. countryside living d. ethnically diverse metropolitan living e. london cosmopolitan f. services & industrial legacy g. towns & country living h. urban settlements notes: credit card spending is 14 day moving average de-seasoned by taking ratio of the 14 day moving average a year prior. the series is then indexed to its moving average 8 -28 figure a3: credit card spending by geographic regions yorkshire & the humber notes: credit card spending is 14 day moving average de-seasoned by taking ratio of the 14 day moving average a year prior. the series is then indexed to its moving average a uk map of these areas can be found in annex figure 1. figure a4: uk political geography a. urban-rural areas (super groups) b. geographic regions source overall credit card spending in areas subject to local lockdown (yellow) compared to control areas not locked down (black), 7 day moving average a wolverhampton notes: overall credit card spending is a 7 day moving average de-seasoned by taking ratio of the 7 day moving average a year prior. the series is then indexed to its moving average 8 -28 food and beverage credit card spending in areas subject to local lockdown (yellow) compared to control areas not locked down (black) credit card spending is a 7 day moving average de-seasoned by taking ratio of the 7 day moving average a year prior. the series is then indexed to its moving average credit card spending in large store chains areas subject to local lockdown (yellow) compared to control areas not locked down (black) wolverhampton notes: store spending based on transactions tagged to large retail store chain locations. credit card spending is a 14 day moving average de-seasoned by taking ratio of the 14 day moving average a year prior. the series is then indexed to its moving average 8 -28 notes: offline credit card spending is a 7 day moving average de-seasoned by taking ratio of the 7 day moving average a year prior. the series is then indexed to its moving average 8 -28 january 2020. key: cord-283708-k9hquon7 authors: cilloni, l.; fu, h.; vesga, j. f.; dowdy, d.; pretorius, c.; ahmedov, s.; nair, s. a.; mosneaga, a.; masini, e. o.; suvanand, s.; arinaminpathy, n. title: the potential impact of the covid-19 pandemic on tuberculosis: a modelling analysis date: 2020-05-20 journal: nan doi: 10.1101/2020.05.16.20104075 sha: doc_id: 283708 cord_uid: k9hquon7 background routine services for tuberculosis (tb) are being disrupted by stringent lockdowns against the novel sars-cov-2 virus. we sought to estimate the potential long-term epidemiological impact of such disruptions on tb burden in high-burden countries, and how this negative impact could be mitigated. methods we adapted mathematical models of tb transmission in three high-burden countries (india, kenya and ukraine) to incorporate lockdown-associated disruptions in the tb care cascade. the anticipated level of disruption reflected consensus from a rapid expert consultation. we modelled the impact of these disruptions on tb incidence and mortality over the next five years, and also considered potential interventions to curtail this impact. results even temporary disruptions can cause long-term increases in tb incidence and mortality. we estimated that a 3-month lockdown, followed by 10 months to restore normal tb services, would cause, over the next 5 years, an additional 1.92 million tb cases (crl 1.74 2.15) and 488,000 tb deaths (cri 449 541 thousand) in india, 48,000 (33,400 72,320) tb cases and 16,800 deaths (11.9 21.9 thousand) in kenya, and 9,100 (6,980 11,200) cases and 1,960 deaths (1,620 2,350) in ukraine. however, any such negative impacts could be averted through supplementary 'catch-up' tb case detection and treatment, once restrictions are eased. interpretation lockdown-related disruptions can cause long-lasting increases in tb burden, but these negative effects can be mitigated with targeted interventions implemented rapidly once lockdowns are lifted. the emergence of the novel virus sars-cov-2 has caused morbidity, mortality and societal disruption on a global scale. in the absence of pharmaceutical interventions, many countries have resorted to population-wide lockdowns to slow the spread of the virus and to allow their health systems to cope 1 . these lockdowns have had an important effect on sars-cov-2 transmission 2,3 . however, unintended consequences are inevitable with such sweeping measures. in low-and middle-income countries with health systems already under strain, even temporary disruptions in health services can have lasting impact on population health 4, 5 . in the present study we focus on tuberculosis (tb) -globally, the leading cause of death due to an infectious disease 6 . in recent decades tb incidence and mortality have been steadily declining, reflecting ongoing improvements in diagnosis, treatment and prevention 7 . however, in march 2020 a rapid analysis conducted by the stop tb partnership brought attention to severe impacts of covid-related lockdowns on tb care in different countries 8 . for example, in the weeks following the imposition of a nationwide lockdown on march 24, 2020, india reported an 80% drop in daily notifications of tb 8 relative to average pre-lockdown levels. such declines, likely reflecting reductions in access to diagnosis and treatment, could have a lasting impact on tb burden at a country-wide level. missed diagnoses would mean increased opportunities for transmission, while worsened treatment outcomes increase the risk of death from tb. therefore, while lockdowns are an important measure to mitigate the immediate impact of covid-19, it is critical to anticipate (i) the potential long-term impact of these measures on tb and other diseases, and (ii) how this impact might be stemmed, in the short term, by appropriately targeted investment and effort. we therefore aimed to examine these questions using mathematical modelling of tb transmission dynamics. building on earlier modelling conducted for the 2019 lancet commission on tuberculosis 9,10 , we modelled the potential tbrelated impact of covid-related lockdowns -and mitigating effects of potential post-lockdown interventions -in three focal countries: india, the republic of kenya, and ukraine. for each country we drew from previously published models of tb transmission 10 , which were designed to capture essential features of the tb care cascade. for the current analysis, this approach allowed us to model the impact of disruptions acting at multiple points in the care cascade. for india we incorporated the dominant role of the private healthcare sector in providing tb care 11 ; for kenya, the role of hiv in driving tb dynamics 12 ; and for ukraine, the burden of drug resistance 13 . we calibrated each country model to the available data on tb burden, including who estimates of tb incidence and mortality 6 , and on the burden of drug resistance. full details of each model are provided in the supporting information. calibration was performed using markov chain monte carlo (mcmc) simulation [14] [15] [16] , whereby we allowed model parameters to vary over pre-specified prior distributions, using a likelihood function based on the calibration targets listed above to weight simulations according to their fit to the observed data. for each country, we drew 250 samples from the weighted (posterior) density of simulations following burn-in and thinning as described in the supporting information. we then performed model projections on the basis of each of these samples, under the lockdown scenarios described below. for any model projection (for example, incidence over time), we estimated bayesian credible intervals as 2.5 th and 97.5 th percentiles, and central estimates as 50 th percentiles, of the corresponding posterior density. disruptions to tb services can act at all stages of the tb care cascade. during a lockdown, movement restrictions would curtail opportunities for those experiencing tb symptoms to seek care. even once these people are able to visit a provider or health facility, the diagnostic and laboratory capacity needed to support tb diagnosis may be severely reduced -for example, with molecular diagnostic tools for tb being repurposed for covid-19 17 or tb laboratory staff being redirected to covid-19 efforts. national tb programmes are investing significant effort to continue supporting those already on tb treatment, but there are also concerns that lockdown conditions may interfere with the continued supply of drugs 18 . to capture this range of possible disruptions, we performed a rapid consultation amongst experts at the stop tb partnership and the united states agency for international development (usaid). table 1 lists those experts' consensus opinion as to the degree to which tb services could be disrupted by covid-related lockdowns, at each step of the care cascade. there is substantial uncertainty around these possible impacts, and as described below, we performed sensitivity analysis to identify which components of disruptions would have the greatest impact on overall tb burden. depending on its readiness, a country tb programme may take weeks or months to restore tb services to normal after a lockdown. this process may be delayed if, for example, laboratory capacity for diagnosis needs time to be reconstituted for tb, or indeed if there remains a reluctance to seek care amongst those with tb symptoms, as a consequence of fear and stigma caused by the covid 19 pandemic. accordingly, to model the impact of the lockdown and its aftermath, we assumed two phases: a lockdown of given duration, during which all impacts listed in table 1 are in full effect, followed by a 'restoration' period, during which tb services are gradually (for simplicity, linearly) restored to normal. we also assumed that tb transmission would revert to normal at the same time as lifting the lockdown, as a result of contact rates in the community rapidly being restored to normal (although see below for sensitivity analysis). this assumption may be appropriate in high-burden, low-income settings where physical distancing is less feasible than in high-income settings, but also where there are strong economic incentives to restore livelihoods as soon as possible. we present results for two scenarios: a 'moderate' scenario consisting of a 2-month lockdown followed by a 2-month . 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 may 20, 2020. . https://doi.org/10.1101/2020.05.16.20104075 doi: medrxiv preprint restoration period for tb services, and a 'severe' scenario consisting of a 3-month lockdown followed by a 10-month restoration period. in each scenario we simulated the excess tb cases and deaths that would arise, over the period from 2020 -2025, compared against a situation where tb services continue as normal over this period. in doing so, we ignore potential expansions in tb care, for example the scaleup of engagement with the private sector in india that was ongoing prior to the covid-19 pandemic 19 . since our analysis does not include the benefits of continuing these expansions, our model projections should be conservative with respect to the excess tb burden arising from the lockdown. until further data become available (discussed below), we took the assumptions in table 1 as plausible scenarios for disruption. we also analysed how the impact of lockdown may vary, under different conditions for the type and length of disruption. first, we examined model sensitivity to the duration l of the lockdown and r of the restoration period. for a fixed value of r, we simulated excess tb burden (cases or deaths) for a hypothetical range of l between 2 and 6 months. using the gradient of excess tb burden with respect to l, we estimated the additional tb burden that would result, between 2020 and 2025, for every month of lockdown. likewise, we estimated the excess tb burden per month of restoration, by holding l fixed and estimating the gradient of excess burden with respect to r between 1 and 10 months. second, we conducted a 'leave-one-out' analysis, in which we simulated the impact of the lockdown, but in the absence of a single element in table 1 (for example, a scenario where all impacts are in full effect with the exception of diagnosis, which remains at pre-lockdown levels). this analysis allows an assessment of how excess tb burden may vary under more limited disruptions than the full set of scenarios identified in table 1 . in doing so, this analysis also helps to identify which types of disruption have the strongest contribution to excess tb burden. by performing a 'leave-one-out' simulation for each row of table 1 in turn, we aimed to estimate the influence of each type of disruption. additionally, while many of the assumptions in table 1 can be refined as further data become available, the effect of reduced contact rates in particular will be challenging to measure empirically. we therefore conducted additional simulations of excess tb burden with all disruptions in effect, but using an alternative assumption of 25% reduction (rather than 10%) in contact rates during the lockdown period. we additionally simulated a scenario where community contact rates revert to normal over a period of 4 months (rather than immediately), independently of the time taken to restore normal tb services. . 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 may 20, 2020. . https://doi.org/10.1101/2020.05. 16.20104075 doi: medrxiv preprint role of the funding source sa is employed by usaid and san, am, em and ss are employed by the stop tb partnership. the funders otherwise had no role in the study, preparation of the report, or decision to submit the paper for publication. figures s2 -s4 in the supporting information show the model calibrations to each of the targets shown in table 1 . on the basis of these calibrations, following a moderate lockdown we projected that between 2020 and 2025, in india there would be an increase of 473,000 tb cases (95% bayesian credible interval (cri) 429 -529 thousand) and 130,000 tb deaths (95% cri 120 -142 thousand). likewise, in kenya there would be an additional 12,200 cases (95% cri 8,570 -18,200) and 4,550 deaths (95% cri 3,230 -5,960), and in ukraine an additional 2,630 cases (95% cri 2,120 -3,300) and 676 deaths (95% cri 536 -781) (see figures 1 and 2, and table 2 ). overall, this excess burden translates to a 1 -4% increase across countries in tb incidence, and 2 -6% in tb deaths, between 2020 and 2025. both estimates of adverse impact were projected to increase by three-to four-fold in the case of a severe, rather than moderate, lockdown ( table 2 ). in terms of the monthly dynamics, figures 1 and 2 illustrate that increases in mortality would be greater proportionally than increases in incidence, but would also recover more rapidly than incidence upon restoration of normal tb services. increases in incidence lasted far beyond the period of disruption; for example, in india, incidence was projected to remain at least 4% higher than a "business-as-usual" baseline for a period of 32 months, even in the moderate scenario of a two-month lockdown followed by two-month restoration ( figure s5 ). the five-year impact of covid-related lockdowns on tb burden is strongly affected by the durations of the lockdown and restoration periods (table 3) . for example, in india each month of lockdown would give rise to an additional 228,000 tb cases (95% cri 204 -258 thousand) and 64,000 tb deaths (95% cri 58.8 -70.3 thousand) over the next 5 years, while each month to restore normal tb services would give rise to an additional 133,000 tb cases (95% cri 120 -149 thousand) and 34,400 tb deaths (95% cri 31.6 -38.2 thousand). figure s6 in the supplementary information shows the analyses informing these results. in india, the four specific disruptions having the most effect on incidence and mortality are, in order: the probability of diagnosis per visit to a provider; the increase in the initial patient delay before first presenting to a provider; the drop in treatment initiation; and the drop in transmission rate ( figure 3 ). likewise in kenya, the same four factors appear as most influential on the impact of the lockdown, on both tb incidence and mortality. in ukraine, a setting with a high burden of drug resistance, the drop in second-line treatment completion was far more influential on overall impact, though reductions in transmission rate, the drop in drug sensitivity testing, . 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 may 20, 2020. . https://doi.org/10.1101/2020.05. 16.20104075 doi: medrxiv preprint and the drop in the probability of tb diagnosis per visit to a provider were also important considerations. the effect of disruptions in diagnosis, as well as in care-seeking and treatment initiation, is an expansion of the pool of individuals with undetected and untreated tb. figure 4 shows how the size of this pool grows over time; the right-hand panel illustrates the potential impact of a twomonth campaign to reduce the prevalence of untreated tb in india through expanded case finding to reach an augmented monthly notification target, immediately upon easing of lockdown restrictions (i.e., implemented alongside the restoration of tb services). depending on lockdown severity and duration of restoration, such a two-month campaign could, preemptively, bring 5year incidence trends back to pre-lockdown levels. we also conducted analyses to test the sensitivity of model projections, to our assumptions for transmission. figure s7 shows simulations under alternative scenarios, namely transmission that is reduced by 25% (not 10%) during a lockdown, and taking several months (not immediately) to return to normal, once a lockdown is lifted. this additional analysis highlights that short-term increases in tb mortality are likely to occur whatever the effect of the lockdown on transmission, since these increases in mortality are driven by build-ups in undetected tb, rather than by transmission. on the other hand, long-term incidence can be affected by different scenarios for transmission. in particular, when assuming that transmission takes 4 months to return to normal (right-hand panels of figure s7 ), a moderate lockdown scenario represents an example where tb services are restored more rapidly (i.e. within 2 months) than tb transmission, and a severe scenario represents the converse (i.e. service restoration within 10 months). figure s7 illustrates the implications of these scenarios: namely, that the risk of longterm elevations in incidence is greatest when community transmission rates return to normal more rapidly than the restoration of tb services. additional analyses, provided in the supporting information (section 4), illustrate a simple approach for extrapolating from these three focal countries to the global level. this approach suggests, for example, that a severe lockdown scenario could lead to an additional 6,331,000 tb cases, and an additional 1,367,000 tb deaths worldwide between 2020 and 2025. this modeling analysis in three key countries illustrates that even short covid-related lockdowns can generate long-lasting setbacks in tb control. our results suggest that, even in a moderate lockdown scenario, over the next five years tb deaths could see increases of 2-6%, while tb incidence could see increases of 1-4%, in the three countries studied here. this impact would increase roughly threefold under a severe lockdown scenario (figures 1 and 2 , and table 2 ). our results also illustrate how these long-term dynamics depend strongly on the duration of the disruption: in the example of india, each additional month of restoration could cause an additional 133,000 cases and 34,400 deaths over the next five years (table 3) . . 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 may 20, 2020. . https://doi.org/10.1101/2020.05. 16.20104075 doi: medrxiv preprint the reason for these dynamics is illustrated by figure 4 , which shows the accumulation in undetected and untreated tb during a lockdown, as a result of missed opportunities for diagnosis and treatment initiation. this expanded pool of undetected tb continues to seed new infections of latent tb, many of which would take years to manifest as incident tb disease. consequently, service disruptions give rise to a short-term escalation of tb mortality (figure 2 ), followed by a prolonged increase in incidence that could take years to undo (figure 1 ). it follows that this excess burden could be averted through focused efforts to address the problem of undetected tb, immediately upon lifting the lockdown (figure 4 ). in practice, such supplementary measures could involve active case-finding 20,21 , including contact tracing with longitudinal followup 22 . on the patient side, covid-19 and pulmonary tb are both associated with respiratory symptoms. if, during the current pandemic, covid-19 comes to be seen as a "tb-like" disease, public recognition of the importance of recognising tb symptoms may wane once covid-19 is thought to be under control. additional efforts may therefore be needed to address these misperceptions. an additional concern is that covid-19 may carry stigma in many communities, and this stigma may transfer to individuals with tb as well 23 . conversely, there may be opportunities to leverage synergies between the two diseases; for example, integrated tb and covid-19 screening and testing algorithms or combined contact investigation strategies. any such strategies based on respiratory symptoms could use similar infrastructure and staff to mitigate both the direct impacts of sars-cov-2 transmission and the indirect effects of augmented m. tuberculosis transmission. in short, readiness to restore tb services as rapidly as possible, combined with focused efforts to 'catch up' on missed diagnoses, will be critical in limiting any long-term setback to tb care efforts as a result of the covid-19 response. one important uncertainty is the potential impact of the lockdown, on tb transmission. we have assumed that a lockdown would reduce transmission by 10%, and moreover that transmission would revert to normal as soon as a lockdown is lifted. these assumptions reflect expert opinion for the implementation of lockdowns in low-and middle-income settings, but carry substantial uncertainty. as illustrated by section 3 in the supporting information, which tests both of these assumptions, it is likely that short-term increases in tb mortality would be unaffected by alternative scenarios. this is because both factors do little to address the problem of undetected tb, that accumulates during a lockdown. however, our estimates for long-term incidence trends may be affected by alternative scenarios for transmission. in general, the risk of long-term increases in incidence are greatest if community contact rates return to normal at a faster rate than tb services ( figure s7 ). overall, therefore, this sensitivity analysis underlines the practical implications of our analysis: that it is critical to restoring routine tb services as rapidly as possible, alongside 'catch-up' campaigns immediately upon lifting a lockdown. a key implication of our scenario analysis is the centrality of establishing surveillance and other data systems to inform the extent of lockdown-associated disruptions in tb care. for example, . 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 may 20, 2020. . https://doi.org/10.1101/2020.05.16.20104075 doi: medrxiv preprint tb notifications (e.g., ref 24 ) can be monitored in real time at a national and subnational level, to evaluate the depth and duration of any reductions in tb diagnosis at different stages of any lockdown. if these indicators suggest persistent declines in notifications and/or falling levels of treatment success, targeted interventions (e.g., active case finding, treatment support, or expanded access) can be rapidly implemented. as contact investigation for tb is implemented, surveillance of infection and active tb can be established and time trends can be used to inform whether household transmission has increased and/or access to care has declined, again at the local, subnational, and national levels. in the longer term, community-based surveys (e.g., serial surveys of tb infection in young children 25 , can be conducted to explore the impact of lockdowns on tb transmission more broadly. we note that the present analysis focuses only on the potential impact of lockdowns on the tb epidemic, and does not address the potential for direct interactions between tb and covid-19 (for example, increased risk of covid-19 mortality among individuals with tb). for this reason, our estimates for excess mortality in particular are likely to be conservative. for example, early evidence suggests that existing tb infection, whether latent or active, can be a strong risk factor for severe disease resulting from sars-cov-2 infection 26 . moreover, through pre-existing lung damage 27 , past tb infection might also predispose individuals to poorer outcomes from covid-19. further evidence on both potential impacts would be invaluable for future work examining these potential pathogen-pathogen interactions. as with any modelling study, our analysis involves several simplifications. our models do not distinguish age structure, nor pulmonary versus extrapulmonary tb, instead taking an average over these distinctions. for our modelling of kenya, for simplicity we have only captured the transmission dynamics of tb, treating hiv incidence as pre-specified. our model therefore does not capture the potential tb implications of disruptions in hiv care, and for this reason may be conservative. lockdowns are likely to reduce community transmission but at the expense of intensifying and prolonging household and congregate setting exposure. faithfully capturing household contact structure is generally not feasible in compartmental models such as in the current analysis, and instead we have taken a simple approach of an assumed overall net reduction in transmission. as discussed above, practical implications of our analysis remain unchanged by uncertainties relating to transmission. in conclusion, our analysis illustrates how increases in tb burden can take months to manifest, but years to undo. even if a lockdown is a period of curtailed programmatic activity, our results also highlight how this period might be used by country programmes and international agencies to prepare for the timely restoration of tb control activities and even "catch-up" campaigns upon easing of restrictions, to prevent such long-term negative impacts from taking hold. the resilience of systems to end tb worldwide will depend critically on readiness to restore, supplement and monitor tb services as rapidly as possible. . 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 may 20, 2020. . https://doi.org/10.1101/2020.05. 16.20104075 doi: medrxiv preprint author contributions ss, sa and na conceived the study, and na, dd and cp designed the approach. sa, san, am, em, and ss provided expert input in constructing the model assumptions, and validated model findings. lc, hf, jfv and cp performed the analysis, and all authors contributed to the interpretation. lc, hf, na and dd wrote a first draft of the manuscript, and all authors contributed to the final version. we declare no conflict of interest. figure 1. the potential impact of a lockdown on tb incidence in india, kenya and ukraine. shown is monthly tb incidence in each country, in 2020 and 2021, for two lockdown scenarios: (i) a 'moderate' scenario with a 2-month lockdown and a 2-month restoration (orange), and (ii) a 'severe' scenario with a 3-month lockdown and a 10-month restoration (red). bars labeled with "l" and "r" denote, respectively, the lockdown and restoration periods, with numbers giving the number of months in each period. as described in the main text, we assume that the disruptions in table 1 are in full effect during the lockdown period, and that they are reduced to zero in a linear way over the restoration period. shaded intervals show 95% bayesian credible intervals, reflecting uncertainty in pre-lockdown model parameters. cumulative excess tb incidence over the period 2020 -2025 is given in table 2 . . 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 may 20, 2020. . https://doi.org/10.1101/2020.05.16.20104075 doi: medrxiv preprint figure 2 . the potential impact of a lockdown on tb deaths in india, kenya and ukraine. as for figure 1 , but showing monthly tb deaths in each country. as in figure 1 , bars labeled with "l" and "r" denote, respectively, the lockdown and restoration periods, with numbers giving the number of months in each period. shaded intervals show 95% bayesian credible intervals, reflecting uncertainty in pre-lockdown model parameters. excess tb deaths over the period 2020 -2025 are listed in table 2 . . 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 may 20, 2020. table 1 in effect, with the exception of one (given by the label to the left). bars in the figures show the excess tb burden between 2020 and 2025 arising from this scenario, relative to the scenario where all disruptions are in effect. vertical lines mark median excess tb cases and deaths in the 'full-impact' scenario. the largest bars therefore indicate those types of disruption that are most influential, for excess tb burden. left-hand panels show results in terms of excess tb incidence, and right-hand panels show excess tb deaths. error bars show 95% credible intervals, calculated by iterating this process over 250 posterior samples for each country. abbreviations: dst: drug susceptibility test, fl: first-line, hiv: human immunodeficiency virus, ipt: isoniazid preventive therapy, sl: second-line, tx: treatment. . 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 left-hand panel shows, in the example of india, the growth in the prevalence of undetected and untreated tb during the lockdown period, taking the example of a 2-month lockdown followed by a 2-month restoration. as described in the text, this expanded pool of prevalent tb is a source of short-term increase in tb mortality, as well as seeding new infections of latent tb that manifest as incident tb disease over the subsequent months and years. the right-hand panel shows the effect of 'supplementary measures' that are instigated immediately upon lifting the lockdown, and that operate over a two-month period to reach these missed cases and initiate them on treatment as rapidly as possible. in practical terms, such efforts could be guided by notification targets. shown in the figure is the example of a moderate lockdown scenario, followed by supplementary measures that aim to reach a peak target of 14 (95%cri 13-16) monthly notifications per 100,000 population. . 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 may 20, 2020. footnotes: scenarios were constructed through a rapid consultation with experts in the stop tb partnership and usaid, the former using information from a rapid survey of national tb programmes 8 . . 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 may 20, 2020. . https://doi.org/10.1101/2020.05. 16.20104075 doi: medrxiv preprint the scenarios listed here are not predictive, but illustrative on the basis of current information: they offer a basis for examining the potential impact of different types of disruption. 1 for the initial levels of these parameters in each country, see tables s4 -s6 (entries highlighted in yellow) in the supporting information. 2 lockdowns would have the effect of reducing transmission in the community level, but also intensifying and prolonging exposure at the household level. as our models do not incorporate household vs community structure, these scenarios instead aim to capture the net effect of changes in household vs community transmission. in urban slums in particular, where tb transmission is strongest, overcrowding may tend to reduce the effect of any lockdown on community transmission. in section 3 in the supporting information, we present corresponding sensitivity analyses to these assumptions. 3 the initial patient delay is an assumed interval of active, infectious tb, prior to a patient's first presentation for care. it is calibrated to match epidemiological data (see table s1 for data, and tables s4 -s6 for parameter estimates). 4 for simplicity, only the kenya model incorporates the role of hiv/tb coinfection, which is estimated to account for 27% of incident tb. however, we note that ukraine has a high burden of hiv as well; in the present study, our focus in ukraine is on the role of drug-resistant tb. abbreviations: covid-19: coronavirus disease 2019, dr: drug-resistant (i.e. rifampicin-resistant), ds: drug-susceptible, dst: drug susceptibility test, hiv: human immunodeficiency virus, ipt: isoniazid preventive therapy, plhiv: people living with hiv, tb: tuberculosis . 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 may 20, 2020. . https://doi.org/10.1101/2020.05.16.20104075 doi: medrxiv preprint how will country-based mitigation measures influence the course of the covid-19 epidemic? estimating the number of infections and the impact of non-pharmaceutical interventions on covid-19 in 11 european countries the effect of control strategies to reduce social mixing on outcomes of the covid-19 epidemic in wuhan, china: a modelling study malaria morbidity and mortality in ebola-affected countries caused by decreased health-care capacity, and the potential effect of mitigation strategies: a modelling analysis health in financial crises: economic recession and tuberculosis in central and eastern europe world health organization. global tuberculosis report 2019. world health organization tuberculosis 2015: burden, challenges and strategy for control and elimination we did a rapid assessment: the tb response is heavily impacted by the covid-19 pandemic building a tuberculosis-free world: the lancet commission on tuberculosis assessing tuberculosis control priorities in highburden settings: a modelling approach the number of privately treated tuberculosis cases in india: an estimation from drug sales data kenya tuberculosis prevalence survey 2016: challenges and opportunities of ending tb in kenya alarming levels of multidrug-resistant tuberculosis in ukraine: results from the first national survey inference for deterministic simulation models: the bayesian melding approach bayesian melding for estimating uncertainty in national hiv prevalence estimates population health impact and cost-effectiveness of tuberculosis diagnosis with xpert mtb/rif: a dynamic simulation and economic evaluation tuberculosis and hiv responses threatened by covid-19 joint effort for elimination of tuberculosis turning off the tap: stopping tuberculosis transmission through active case-finding and prompt effective treatment how much is tuberculosis screening worth? estimating the value of active case finding for tuberculosis in south africa, china, and indi household-contact investigation for detection of tuberculosis in vietnam world health organization. social stigma associated with covid-19 central tb division india. nikshay dashboard decreasing household contribution to tb transmission with age: a retrospective geographic analysis of young people in a south african township active or latent tuberculosis increases susceptibility to covid-19 and disease severity tuberculosis and lung damage: from epidemiology to pathophysiology 2020 to the beginning of 2025. percentages show increases in cases and deaths relative to a baseline of no lockdown conditions see figure s6, supporting information, for further details. abbreviations: cri-credible interval we gratefully acknowledge support from sara gonzalez andino and shinichi takenaka from stop tb partnership, in the process of development of modelling assumptions. key: cord-268816-nth3o6ot authors: roy, satyaki; ghosh, preetam title: factors affecting covid-19 infected and death rates inform lockdown-related policymaking date: 2020-10-23 journal: plos one doi: 10.1371/journal.pone.0241165 sha: doc_id: 268816 cord_uid: nth3o6ot background: after claiming nearly five hundred thousand lives globally, the covid-19 pandemic is showing no signs of slowing down. while the uk, usa, brazil and parts of asia are bracing themselves for the second wave—or the extension of the first wave—it is imperative to identify the primary social, economic, environmental, demographic, ethnic, cultural and health factors contributing towards covid-19 infection and mortality numbers to facilitate mitigation and control measures. methods: we process several open-access datasets on us states to create an integrated dataset of potential factors leading to the pandemic spread. we then apply several supervised machine learning approaches to reach a consensus as well as rank the key factors. we carry out regression analysis to pinpoint the key pre-lockdown factors that affect post-lockdown infection and mortality, informing future lockdown-related policy making. findings: population density, testing numbers and airport traffic emerge as the most discriminatory factors, followed by higher age groups (above 40 and specifically 60+). post-lockdown infected and death rates are highly influenced by their pre-lockdown counterparts, followed by population density and airport traffic. while healthcare index seems uncorrelated with mortality rate, principal component analysis on the key features show two groups: states (1) forming early epicenters and (2) experiencing strong second wave or peaking late in rate of infection and death. finally, a small case study on new york city shows that days-to-peak for infection of neighboring boroughs correlate better with inter-zone mobility than the inter-zone distance. interpretation: states forming the early hotspots are regions with high airport or road traffic resulting in human interaction. us states with high population density and testing tend to exhibit consistently high infected and death numbers. mortality rate seems to be driven by individual physiology, preexisting condition, age etc., rather than gender, healthcare facility or ethnic predisposition. finally, policymaking on the timing of lockdowns should primarily consider the pre-lockdown infected numbers along with population density and airport traffic. we process several open-access datasets on us states to create an integrated dataset of potential factors leading to the pandemic spread. we then apply several supervised machine learning approaches to reach a consensus as well as rank the key factors. we carry out regression analysis to pinpoint the key pre-lockdown factors that affect post-lockdown infection and mortality, informing future lockdown-related policy making. population density, testing numbers and airport traffic emerge as the most discriminatory factors, followed by higher age groups (above 40 and specifically 60+). post-lockdown infected and death rates are highly influenced by their pre-lockdown counterparts, followed by population density and airport traffic. while healthcare index seems uncorrelated with mortality rate, principal component analysis on the key features show two groups: states (1) forming early epicenters and (2) experiencing strong second wave or peaking late in rate of infection and death. finally, a small case study on new york city shows that days-to-peak for infection of neighboring boroughs correlate better with inter-zone mobility than the interzone distance. states forming the early hotspots are regions with high airport or road traffic resulting in human interaction. us states with high population density and testing tend to exhibit during pre-and post-covid periods to show that the odds of mortality of whites and blacks are statistically equivalent [23] . myers et al. analyzed the covid-19 positive patients in california to investigate its prognosis in the higher age groups and individuals with preexisting conditions [24] . zoabi et al. applied ml on 51,831 covid-19 positive patients to understand the effect of gender, age and contact to show that close social interaction is a strong feature for covid-19 transmissibility [25] . khan et al. applied regression tree, cluster analysis and principal component analysis on worldometer infection count data to study the variability and effect of testing in prediction of confirmed cases [26] . finally, pan et al. studied the effects of the myriad public health interventions (such as lockdown, traffic restriction, social distancing, home quarantine, centralized quarantine, etc.) on 32,583 covid-19 patients, with respect to their age, sex, residential location, occupation, and severity [27] . contributions: while it is evident that factors such as gender, race, age, testing, social contact and distancing have been analyzed in a piecemeal manner, there is no comprehensive study that combines the demographic, economic, and epidemiological, ethnic and health indicators for infection and mortality from covid-19. to address this gap, we carry out a machine learning-based analysis with the following three objectives. 1. we curate a dataset of diverse features (detailed in sec. 2.1) from 50 states of usa. this dataset is somewhat unique, since, in addition to the above features, it includes factors such as airport traffic, homeless and variations in lockdown dates. also, note that the lockdown was enforced on the us states at around the same time, when each state was at a different stage of the covid-19 infection cycle. 2. we analyze the variation of covid-19 infection spread and mortality rates using a set of standard supervised ml methods. we rank the key discriminatory factors based on the importance score calculated from randomized decision trees. we combine the findings to identify the most vulnerable age groups and us states. we also show the effect of testing and lockdowns on the infection spread dynamics. 3. we utilize multiple linear regression to gauge the extent to which the key pre-lockdown factors affect the post-lockdown infected and death numbers. this study assigns weights to features and drive mitigation efforts and large scale policymaking. our data-driven experiments using supervised methods demonstrate that population density, testing [28] and airport traffic [29] are key factors contributing to infection and mortality rates. furthermore, high age group (40 and beyond, and specifically exceeding 60) population are more vulnerable. principal component analysis on the key features show two groups: highly affected us states (1) forming early epicenters and (2) showing consistent or newly peaking rate of infection and death. multiple regression analysis shows that the postlockdown numbers are most influenced by the pre-lockdown infected and death numbers followed by population density and airport activity, while overall healthcare index of a state does not seem to play a part in the overall death count. similarly, the race of individuals did not play any significant role in the infection or mortality numbers. despite increased testing rates, the fraction of individuals tested positive drop approximately three weeks into the lockdown, suggesting that the social distance measures has had an impact on curbing spread. finally, we discuss the role of mobility and distance in infection spread. in the absence of large-scale inter-state mobility data, our case study on the boroughs of new york city show that peaks of infection correlate better with inter-zone mobility than the interzone distance. all the experiments have been performed using scikit-learn, which is a popular machine learning library in python [30] . let us discuss the details of the two datasets used in this work. 2.1.1 data from us states. our dataset has been carefully curated from several open sources to examine the possible factors that may affect the covid-19 related infection and death numbers in the 50 states of usa. the individual open-access data sources as well as the integrated (curated) dataset has been shared on github (https://github.com/satunr/covid-19/tree/master/us-covid-dataset). below, we discuss a summary of the features and output labels of the integrated dataset. • gross domestic product (in terms of million us dollars) for us states [31] (filename: source/ gdp.xlsx, feature name: gdp). • distance from one state to another (is not measured in miles but the euclidean distance between their latitude-longitude coordinates between the pair of states [32] ) (filename: source/data_distance.xlsx, feature name: d(state1, state2)). • gender feature(s) is a fraction of total population representing the male and female individuals [33] (filename: source/data_gender.csv, feature name: male, female). • ethnicity feature(s) are the fraction of total population representing white, black, hispanic and asian individuals (we leave out other smaller ethnic groups) [34] (filename: source/ data_ethnic.csv, feature name: white, black, hispanic and asian). • healthcare index is measured by agency for healthcare research and quality (ahrq) on the basis of (1) type of care (like preventive, chronic), (2) setting of care (like nursing homes, hospitals), and (3) clinical areas (like care for patients with cancer, diabetes) [35] (filename: source/data_health.xlsx, feature name: health). • homeless feature is the number of homeless individuals of a state [36] (filename: source/ data_homeless.xlsx, feature name: homeless). the normalized homeless population of each state is the ratio between its homeless and total population. • total cases (and deaths) of covid-19 is the number of individuals tested positive and dead [37] (filename: source/data_covid_total.xlsx, feature name: total cases and total death). the normalized infected/death is the ratio between the infected/death count to total population of the given state. • infected score and death score is obtained by rounding normalized total cases and deaths to discrete value between 0-6 (feature name: infected score, death score). • death-to-infected is a feature measuring impact of death in terms of the difference between death and infected scores. it is calculated as max(death score -infected score, 0). • lockdown type is a feature capturing the type of lockdown (shelter in place: 1 and stay at home: 2) in a given state [37, 38] (filename: source/data_lockdown.csv, feature name: lockdown). • day of lockdown captures the difference in days between 1st january 2020 to the date of imposition of lockdown in a region [39] (filename: source/data_lockdown.csv, feature name: day lockdown). • population density is the ratio between the population and area of a region [40] (filename: source/data_population.csv, feature name: population, area, population density). • traffic/activity of airport measures the passenger traffic (also normalized by the total traffic across all the states of usa [41] (filename: source/data_airport.xlsx, feature name: busy airport score, normalized busy airport). • age groups (0-80+) in brackets of 4 year (also normalized by total population) [40] (filename: source/data_age.xlsx, feature name: age_to_, norm_to_, e.g. age4to8); we later group them in brackets of 20 for the purposes of analysis. • peak infected (and peak death) measures the duration between first date of infection and date of daily infected (and death) peaks [40] (feature name: peak infected, peak death). • testing measures the number of individuals tested for covid-19 (total number, before and after imposition of lockdown) [38, 42] (filename: source/data_testing.xlsx, feature name: testing, pre-lockdown testing, post-lockdown testing). • pre-and post-infected and death count measures the number of individuals infected and dead before and after lockdown dates (feature name: testing, pre-infected count, pre-death count, post-infected count, post-death count). • days between first infected and lockdown date (feature name: first-inf-lockdown). the above features, their abbreviations and summary statistics (i.e., mean, standard deviation, maximum and minimum) are enlisted in table 1 . note that, for gender and ethnicity we report the fraction of the total state population falling in each category. the new york city (nyc) datasets (https://github.com/ satunr/covid-19/blob/master/us-covid-dataset/nyc_dist_mob.xlsx) show the inter-borough distance and mobility as well as covid-19 infected (https://github.com/satunr/covid-19/blob/master/us-covid-dataset/nyc-inf.xlsx) and death counts (https://github.com/ satunr/covid-19/blob/master/us-covid-dataset/nyc-dth.xlsx) for the 5 boroughs of nyc, namely, manhattan, queens, brooklyn, bronx and staten island. table 1 . summary of features and their statistics (i.e., mean, standard deviation (dev.), maximum (max.) and minimum (min.)). the features in the order shown under "feature name" are: gdp, inter-state distance based on lat-long coordinates, gender, ethnicity, quality of health care facility, number of homeless people, total infected and death, population density, airport passenger traffic, age group, days for infection and death to peak, number of people tested for covid-19, days elapsed between first reported infection and the imposition of lockdown measures at a given state. factors affecting covid-19 infected and death rates inform lockdown-related policymaking • mobility data (based on traffic volume counts collected by dot for new york metropolitan transportation council (nymtc) [43] ) shows the number of trips from one borough to another. • covid-19 data shows the number of covid-19 infected and death counts for each borough [44] . we acquire the daily infected and testing counts across us from january-july, 2020 [45] . this dataset is part of the covid tracking project that collect covid-19 statistics on the numbers on tests, cases, hospitalizations, and patient outcomes from every us state and territory by voluntary public participation. we use the scikit-learn library kbinsdiscretizer to group the continuous feature values into discrete values by creating balanced clusters using the quantile strategy [46] . 2.1.5 supervised learning methods. supervised machine learning algorithms learn a function that maps the input training data (i.e., features) to some output labels [47] . in this work, we consider the following supervised learning techniques. (refer [48] [49] [50] [51] [52] [53] [54] for the details on these ml approaches.) • support vector machine (svm) is used for classification and regression problems that maps the inputs to high-dimensional feature spaces. svm operates on hyperplanes-decision boundaries that help classify the data points. the objective is to maximize the separation between the data points and the hyperplane. svm is memory efficient and effective for datasets with fewer data samples [55] . • stochastic gradient descent (sgd) is an iterative approach that fits the data to an objective function [56] . as the name suggests, it is a stochastic variant of the popular gradient descent (gd) optimization model [57] . in gd, the optimizer starts at a random point in the search space and reaches the lowest point of the function by traversing along the slope. unlike gd that requires calculating the partial derivative for each feature at each data point, sgd achieves computational efficiency by computing derivatives on randomly chosen data points. • nearest centroid (nc) is a simple classification model that represents each class by the centroid of its members. subsequently, it assigns each data point to the cluster whose centroid is the closest to it. nc is particularly effective for non-convex classes and does not suffer from any additional dependencies on model parameters [58] . • decision trees (dts) are a classification and regression technique that assigns target labels based on decision rules inferred from data features [59] . dt maintains the decision rules using a tree. a data point is assigned to a class by repeatedly comparing the tree root with the data point value to branch off to a new root. • gaussian naive bayes (nb) are a class of fast, probabilistic learning techniques that apply the bayes' theorem to assign labels to the data points [60] . while supervised ml approaches generally yield reliable prediction accuracy, they often suffer from overfitting or convergence issues [47, 61] . each of the above approaches has its own advantages and disadvantages. svm works well when the underlying distribution of the data is not known. however, it is prone to overfitting when the number of features is much greater than the number of samples. sgd needs low convergence time for a large dataset, but it may require to fit a number of hyperparameters. conversely, dt involves almost no hyperparameters, but often entails slightly higher training time. unlike dt, nb requires less training time but works on the implicit assumption that all the attributes are mutually independent. finally, nc is a fast method but is not robust to outliers or missing data. in the context of our work, we intuit that the discriminatory feature(s) will yield a high accuracy irrespective of the underlying supervised ml algorithm used. • accuracy function measures the fraction of matches between the predicted and actual labels in a multi-label classification, i.e., the ratio of correctly predicted observations to the total observations. it can be calculated as: in the above equation, tp, tn, fp, fn denote true positive, true negative, false positive and false negative, respectively. • extra trees classifier is an estimator that fits randomized decision trees (called extra-trees) on data samples. the memory and computation overhead of this approach can be controlled by regulating the size of the extra trees. the nodes in the tree are split into sub-trees resulting in high accuracy (i.e., drop in impurity). thus, feature importance is measured as total reduction in impurity affected by that feature [62] . • multiple regression (mr) is a statistical tool to capture the linear relationship between the independent and the dependent variables x and y of a function y = g(x). in our context, mr generates a linear relationshipŷ where b fi is the coefficient that captures the contribution of feature f i towards the dependent variable y, while β 0 and � are the intercept and error terms, respectively. given any pair of vectors v andv (jvj ¼ jvj ¼ n), we apply the following standard statistical operations: • mean centering subtracts the mean μ from each element of a vector v, i.e., v 0 = v − μ(v). this standardization adjusts the scales of magnitude by making the new mean 0 and helps compare data from varied sources or having different datatypes. • mean squared error (mse) is calculated as 1 • pearson correlation coefficient (pcc) between v andv measures the strength of a linear association between two variables, where the value pcc = 1 is a perfect positive correlation and −1 is perfect negative correlation. • positivity rate ρ is the ratio between the number of individuals tested positive to the number of tests performed daily [63] . this section is classified into the following three subsections: (1) and (2) table 2 . unless otherwise stated, the feature set comprises gdp, gender, ethnicity, health care, homeless, lockdown type, population density, airport activity, and age groups, whereas the output labels consist of infected and death scores on a scale of 0-6. we apply supervised machine learning (ml) approaches to identify the key factors affecting covid-19 infected and death counts. for each supervised ml technique, we perform an exhaustive search of all possible combinations of any 5 features and identify the feature subset (s) with the highest accuracy (discussed in sec. 2.2) as the most important features. fig 1 shows the scores for different supervised methods. although proposing a machine learning algorithm that works best on covid-19 data is not the purpose of this study, it is worth reporting that decision tree classifier (dt) slightly outperforms the other algorithms for both cases of infected and death scores. we create a pool of all features participating in at least one combination for output labels of infected and death scores. fig 2 shows a heatmap of the importance i for all such features against each supervised technique. for infected score as output label (top figure), homeless (home), population density (pd), airport activity (air), testing (test), white (wht), etc. have the highest i. for death score as output label, pd, air, test and age groups above 50 years (age50_54 and age80_84) exhibit the highest importance. we apply the extra trees classifier to generate the impurity-based rank for the features (discussed in sec. 2.2). fig 3a shows the top 5 important features corresponding to the infected and death scores, respectively. it is interesting that for both cases, the same set of features, namely, population density, days to peak, airport traffic, testing and high age groups, are identified. also note that the same features exhibit a very high participation in the 5-feature combinations shown in fig 2. next, as a validation exercise, we apply dimension reduction on the factors affecting covid-19 infected and death rates inform lockdown-related policymaking table 1 we discussed in sec. 2.1, that our initial dataset groups ages into brackets of 4 (0-4, 4-8, and so on). our results from supervised learning (sec. 3.1) and extra trees (sec. 3.2) suggest that high age groups are important factors affecting the infected and death scores of covid-19. to understand the effect of covid-19 infected and death scores on low and high age groups, we create two feature sets for population of age �40 and >40. fig 4a shows that for both cases of infected and death, the accuracy (acc) is higher for higher age groups. we explore this by repeating the above experiment, this time, with a feature set of groups 40-60 and >60. fig 4b depicts that acc for age group 60+ is marginally higher, suggesting that the elderly are amongst the most vulnerable, however the difference in mortality rates in this case was not statistically significant. we carry out a study to identify the pre-lockdown factors of any region (us states in our case) that contribute to the overall post-lockdown infection and death numbers. we partition the total infected and death numbers for each state into pre-and post-lockdown infected and death counts. we then create a feature set consisting of population density, airport business, pre-lockdown infected, pre-lockdown death, days between first infected to lockdown and age group above 80. the features represent the set of observable factors for the administrative and health bodies and were already shown to possess high feature significance in the previous factors affecting covid-19 infected and death rates inform lockdown-related policymaking section. the output labels are the post-lockdown infected and post-lockdown death numbers. we perform the following experiments: 3.4.1 identification of discriminating features. we carry out a simple preprocessing step to convert each feature entry to percentile (with respect to the feature vector) and rank the us states in the decreasing order of infected and death scores (fig 5) . we calculate the weighted average percentile of features for the top and bottom k = 10 us states using the formula where p(f i ) and ρ(f i ) are the percentile and rank of the i th feature value, while r is the number of us states (equal to maximum rank). we intuit that the feature exhibiting the maximum difference in weighted average percentile for top and bottom k covid-19 affected us states are the discriminating ones. fig 6a shows the percentile difference suggesting that airport and population density are the most significant, while days between first infected to lockdown and age group of 80+ are the least discriminating. we apply multiple regression (mr) (see sec. 2.2) to measure the weightage of each of the above features in the observed post-lockdown infected (post_inf) and post-death numbers (post_dth). we eliminate the days between first infected to lockdown (fst-lock) and age group 80+, which are the least discriminating features from the percentile analysis (see fig 6a) . as a prerequisite for mr, we need to eliminate features that are mutually correlated. fig 6b shows that pre-inf and pre-dth are highly correlated, and hence we run two separate batches of mr: (1) population density, airport business, pre-lockdown infected and (2) population density, airport business, pre-lockdown death. we explore the effect of testing and lockdown on infection spread. we utilize positivity ratio ρ (defined in sec. 2.3) to gauge how widespread the infection spread is [63] . we acquire the daily infected and testing count in us (see sec. 2.1.3) and plot the mean daily ρ across all states over the period of february-july 2020. fig 7a shows that the testing increased over a period time, while the positivity ratio dropped post lockdown (shown in red dotted line). while, testing (and, by extension, positivity ratio) is an effective epidemiological indicator, it cannot curb infection spread by itself. however, fig 7a shows that the ρ has dropped approximately three weeks into the lockdown, suggesting that the latter had an impact on curbing spread by minimizing social contact. table 3 shows that pre-infected and pre-death with high coefficients contribute highly towards factors affecting covid-19 infected and death rates inform lockdown-related policymaking the post-lockdown infected and death numbers, followed by population density and airport traffic. this finding is further supported by the p values reported for the respective features. note that the r 2 scores for all the four cases are >0.8, suggesting that the output features capture a high proportion of the variance in the input features. overall, pre-infected count has higher coefficient and r 2 score and emerges as a marginally better discriminating feature of post-lockdown effects than the pre-death count. factors affecting covid-19 infected and death rates inform lockdown-related policymaking in sec. 3.2, we perform pca on the feature set of the key factors to show that states with high infection and death numbers stand out of the cluster of other states. these states include some erstwhile hotspots forming group 1 (such as new york city, new jersey, massachusetts, connecticut, rhode island) as well as states experiencing a steady infection and death count and also a strong second wave forming group 2 (such as texas, washington, california, georgia, arkansas, utah and colorado) (fig 3b) . in the pca analysis, pc1 and pc2 account for 41% and 21% variance, respectively. we explore how each feature influences each component to show that pc1 is driven by factors such as airport activity and high age groups (70 and beyond), while pc2 is dominated by population density, airport, age (80+) and testing. notice in fig 3b, though both groups 1 and 2 exhibit high spread across pc1, group 2 forms a slightly denser cluster than group 1, implying that it exhibits an even mix of pc1 and pc2 features. we intuit that the early peaking in infection in group 1 states is due to high road and airport mobility leading to high mixing and infection spread that is manifested in the elderly population. group 2 shows enduring infection spread due to high population density and testing, in addition to airport activity and populations with higher age group. we study how demographics affect covid-19 numbers to show that states with higher age groups (particularly 60 and beyond) numbers are the most vulnerable. finally, we split the infected and death numbers on the pre-and post-lockdown epochs and apply multiple linear regression to show that pre-lockdown infected and death, population density and airport contribute highly to the post-lockdown numbers. this analysis can be particularly effective in pinpointing the most vulnerable states and recommending lockdown policies on starting dates and duration to curb pandemic spread. note that our present study pertains to the identification of the discriminatory features with respect to the date of lockdown. there exists several unanswered questions regarding the impact of length, scheduling strategies, lockdown types and extent of lockdowns on pandemic spread that need to be answered. such an analysis requires a richer feature set as well as a sound understanding of the dynamics of infection spread in terms of healthcare, distance, mobility, etc. as a preliminary study, we first explore whether there is any relationship between the health care index (health) of a us state and the number of transitions from infected to death (dth/inf) in this state. the pearson's correlation coefficient (see sec. 2.3) between the two factors is 0.11, suggesting that the overall mortality numbers is largely unrelated to the healthcare facility and may solely depend on the infected individual's attributes, such as age, comorbidities, infection severity, etc. second, since proximity plays a role in infection spread, neighboring regions should peak at nearly the same time. we posit that mobility may play an even greater role in the spread, than a static measure like distance between a pair of regions. in the absence of a inter-state mobility dataset, we create two feature sets for the nyc boroughs dataset (see sec. 2.1): (1) inter-borough distance and (2) inter-borough mobility. each borough b has a distance and mobility vector d b = {d b1 , d b2 � � �} and m b = {m b1 , m b2 � � �} where d bi and m bi are the probabilistic measure of distance and mobility between a borough b with borough i. we calculate the correlation of the mean squared error (see sec. 2.3) of the distance/mobility vectors of any pair of boroughs b 1 and b 2 against the absolute difference of their peak to infected or peak-to-death features. fig 7b suggests that mobility yields a higher correlation (0.44) than distance (0.22) suggesting that mobility is a slightly more informative feature to analyze infection spread. we are currently working towards broadening the scope of this study in different directions. first, this work attempted to apply ml analysis on a wide range of features, making the the states of united states the ideal choice, specifically from the standpoint of data availability. in future we would like to extend this work by running these experiments on epidemiological, demographic and economic data of different countries. it would be interesting to report the variation in the discriminatory features identified for different countries. second, we identify population density, testing, airport activity and pre-lockdown infected count as key features driving the post-lockdown infection and death numbers. we plan to utilize these findings to design policies on the timing, duration and stringency of lockdown for future pandemics. third, all the input features discussed in this work are static or time invariant. it is imperative to analyze the evolution of dynamic features (such as gdp and unemployment rates) from the pre-covid to the post-covid timelines to uncover the long-term economic effects of covid-19. machine learning is emerging as an important tool to predict the dynamics of spread of covid-19 and identify the key factors driving infection and mortality rates. while existing works study the effects of gender, race, age, testing, social contact and distancing separately, we present an unified analysis of the demographic, economic, and epidemiological, ethnic and health indicators for infection and mortality rates from covid-19. we curate a dataset of us states comprising features (from varying sources discussed in sec. 2.1) that may potentially impact infection and death rates of covid-19. we run several supervised machine learning techniques to identify and rank the key factors correlating with infection and fatality counts. population density, testing rate, airport traffic, high age groups emerge as significant, while ethnicity, gender, healthcare index, homeless and gdp have little or no impact on pandemic spread and mortality. coronavirus: what have been the worst pandemics and epidemics in history coronavirus world map: which countries have the most cases and deaths epidemiology, causes, clinical manifestation and diagnosis, prevention and control of coronavirus disease (covid-19) during the early outbreak period: a scoping review covid-2019 and world economy. covid-2019 and world economy covid-induced economic uncertainty is this the second wave of covid-19 in the u.s.? or are we still in the first? how will country-based mitigation measures influence the course of the covid-19 epidemic? the lancet in beijing it looked like coronavirus was gone. now we're living with a second wave daily covid-19 cases in india continue to soar, japan's tokyo in fears of 2nd wave of infections a fiasco in the making? as the coronavirus pandemic takes hold, we are making decisions without reliable data it's time to get real about the misleading data analysis of factors associated with disease outcomes in hospitalized patients with 2019 novel coronavirus disease factors affecting covid-19 transmission the origin, transmission and clinical therapies on coronavirus disease 2019 (covid-19) outbreak-an update on the status prediction models for diagnosis and prognosis of covid-19 infection: systematic review and critical appraisal artificial intelligence and machine learning to fight covid-19 machine learning using intrinsic genomic signatures for rapid classification of novel pathogens: covid-19 case study coronavirus (covid-19) classification using ct images by machine learning methods wrong but useful-what covid-19 epidemiologic models can and cannot tell us prediction of epidemic trends in covid-19 with logistic model and machine learning technics modified seir and ai prediction of the epidemics trend of covid-19 in china under public health interventions the association of race and covid-19 mortality. eclinicalmedicine, 100455 characteristics of hospitalized adults with covid-19 in an integrated health care system in california covid-19 diagnosis prediction by symptoms of tested individuals: a machine learning approach. medrxiv hossain countries are clustered but number of tests is not vital to predict global covid-19 confirmed cases: a machine learning approach. medrxiv randomized placebo-controlled trials of remdesivir in severe covid-19 patients: a systematic review and meta-analysis. medrxiv center for disease control and prevention. covid-19 testing overview scikit-learn: machine learning in python world population review. gross domestic product list of geographic centers of the united states population distribution by gender population distribution by race agency for healthcare research and quality. health care quality: how does your state compare? 2013 ahar: part 1-pit estimates of homelessness in the u cdc covid data tracker covid-19 cases covid19 us lockdown dates dataset united states census. state population by characteristics list of the busiest airports in the united states center for disease control and prevention. previous u.s. viral testing data nyc-covid19 borough level breakdown scikit-learn-preprocessing -kbinsdiscretizer machine learning: a review of classification and combining techniques support vector machine stochastic gradient descent scikit learn developers (bsd license) scikit learn developers (bsd license) scikit learn developers (bsd license). naive bayes scikit learn developers (bsd license) multiple linear regression support vector machine-a survey stochastic gradient descent an overview of gradient descent optimization algorithms a local mean-based k-nearest centroid neighbor classifier simplifying decision trees. international journal of man-machine studies an empirical study of the naive bayes classifier scikit-learn classifier tuning from complex training sets covid-19 testing: understanding the "percent positive the news tribune. washington state reports 455 new covid-19 cases if trends persist, houston would become the worst affected city in the us, expert peter hotez says dph reports almost 900 new cases of covid-19 in ga hundreds test positive for covid-19 at tyson foods plant in arkansas covid-19 cases rise as hospitalizations remain low in colorado utah confirms 394 new coronavirus cases; 3 more deaths on sunday the authors would like to acknowledge the editor/reviewers for critically assessing the materials and providing suggestions that significantly improved the presentation of the paper. furthermore, they acknowledge the department of computer science, virginia commonwealth university for its computational resources. validation: satyaki roy.visualization: satyaki roy. writing -review & editing: preetam ghosh. key: cord-275827-r86ygqmy authors: lapeyre-mestre, maryse; boucher, alexandra; daveluy, amelie; gibaja, valerie; jouanjus, emilie; mallaret, michel; peyrière, helene; micallef, joëlle title: addictovigilance contribution during covid-19 epidemic and lockdown in france date: 2020-06-23 journal: therapie doi: 10.1016/j.therap.2020.06.006 sha: doc_id: 275827 cord_uid: r86ygqmy abstract addictovigilance is a safety monitoring targeted at substances with potential for abuse and dependence. this vigilance was involved during the period of covid-19 epidemic due to the significant changes in access to drugs and psychological disruption caused by the pandemic and lockdown. this article aims to present the different steps implemented by the french addictovigilance network in collaboration with the french health authorities from march to may 2020, including monitoring of potential harmful events, and scientific communication. the first events were identified through the continuity of the networking between the french addictovigilance centres and their partners: community pharmacies, general practitioners, specialized structures and emergency wards. as soon as the lockdown began, first cases of overdoses (lethal or not) were reported with opioids, mainly with methadone, and other opioids (heroin, oxycodone, tramadol or antitussive codeine). lockdown-related noteworthy events consisted in clinical cases or other relevant information for which lockdown clearly played an important role : among the many substances identified at least once, pregabalin, benzodiazepines, cannabis, cocaine and nitrous oxide were the most significant in terms of prevalence, seriousness or particularly specific to the lockdown context. despite significant decrease in the activity and travel limited to vital needs, community pharmacies continued to identify falsified prescriptions in this period, highlighting an increase in suspicious requests for pregabalin, codeine and tramadol. in parallel, the french addictovigilance network continued its communications efforts in the period, issuing a newsletter on tramadol, a press release on methadone and naloxone, and participating in the covid-19 frequently asked questions (faqs) of the french society of pharmacology and therapeutic website (https://sfpt-fr.org/covid19 ). covid-19 epidemic has been an important challenge for addictovigilance, and has proved that this monitoring is highly essential for alerting health professionals and health authorities to points of vigilance in the field of psychoactive substances. any safety monitoring system is part of a global approach aimed at identifying emergence or spread of a health risk. this health security approach involves the early detection of signals and their most rapid integration into an action system allowing an adapted, effective and early intervention to preserve the health of populations. in the context of pharmacovigilance and drug safety, new or unexpected adverse drug reactions should be detected as early as possible in order to further inform and secure the use of the drug, giving the general population and health professionals the opportunity of evidence-based information about these risks. in the context of covid-19 epidemic, the french regional pharmacovigilance centres network ensured this mission [1] , with an assessment maintained in its continuity, based on a pharmacological and medical characterization of cases, shared with a population-based approach integrating pharmacoepidemiological methods when possible, contributing to optimizing the level of evidence. sharing and collaboration, both within and beyond the french pharmacology and therapeutics scientific community, was integral within these special weeks and beyond (see frequently asked questions [faq] at https://sfpt-fr.org/covid19 ) [2] [3] [4] . in the addictovigilance context, the field is even wider and more heterogeneous [5] [6] [7] . in the first weeks of epidemic spread, most of the interrogations were related to the disease itself and to drugs with supposed antiviral properties or interactions with the immune system. concerns about substances of abuse appeared as soon as lockdown occurred in france on march 17, 2020 . this article the french addictovigilance network was set up in the 1990s, in order to benefit from a proactive vigilance system targeted at substances with potential for abuse and dependence (except tobacco and alcohol), and to participate in a proactive and coordinated manner in the activities of the world health organization (who) expert committee on drug dependence [8, 9] . this vigilance is based on spontaneous notification by healthcare professionals of any serious case of misuse, abuse and drug dependence involving psychoactive substances, regardless of their nature or status [5, 10] . in addition to this passive monitoring subject to under-reporting, other sources of information have been developed to improve vigilance: systematic data collection on falsified prescriptions from pharmacies ("ordonnances suspectes indicateur d'abus possible", osiap survey) [11, 12] and on secure prescription forms for narcotic drug prescriptions ( "antalgiques stupéfiants et ordonnances sécurisées", asos survey) [13] , systematic data collection from patients seen in addiction specialized structures ("observatoire des produits psychotropes illicites ou détournés de leur usage médicamenteux", oppidum survey) [14] , analysis of toxicological data on chemical submission [15] or on deaths in a medico-legal framework ("décès en relation avec l'abus de médicaments et de substances", drames survey) [16] . addictovigilance can broaden the assessment of the potential for abuse and dangerousness of substances by specific analyses on large databases from the national health data warehouse [17, 18] , or on ad hoc field studies [19] [20] [21] [22] [23] [24] . the identification of a potential signal from one or more of the sources described above makes it possible to anticipate an emerging problem and to assess its magnitude using a multi-source approach (fig. 1) [5, 10] . on march 16, in his first address on the extend of the epidemic in france, the french president announced, in a message broadcast to the nation, the implementation of travel restrictions, lockdown, and a state of emergency involving the redeployment of the entire healthcare sector to prioritize covid-19 care from the following day. among the different measures launched by the government, several ones were intended to ensure continuous access for care, while limiting outing to what was strictly necessary (urgent medical care). in these conditions, both public and private medical sectors (general and specialized practitioners, nurses, most of other health professionals) decreased their activities, together with addiction specialized structures, in order to insure social distancing and prophylactic barrier measures to reduce the risks of viral contamination. for example, in many areas in france, several first line harm reduction structures [25] and addiction specialized centres modified their way of functioning, with limited access hours, redeployment of nurses and doctors for covid care, remote consultations, etc., all these changes leading to a degraded operating mode. some other structures may have also closed their doors, in particular those offering conviviality space with coffee and food for homeless and vulnerable isolated people, because of the impossibility to ensure social distancing. the rules for renewing prescriptions have been modified by several decrees (the first being published in the official journal on march 14, 2020 [26] ), in order to prevent the health risks related to the abrupt interruption of chronic exposure to drugs, in a context of a reduced availability of prescribers during covid-19 epidemic. pharmacists were invited to issue even if the period of validity of a renewable prescription has expired, within the framework of the initially planned dosage, a number of boxes per prescription line guaranteeing the continuation of treatment, for a period not exceeding one month. these measures include specific provisions concerning medicinal products liable to be abused or misused, such as anxiolytic or hypnotic drugs, opioid maintenance drugs and other narcotic drugs or drugs falling under the regulations of narcotics. along the successive decisions of the president and government over time, these decrees were intended to be prolonged during the period of the national state of health emergency. table 1 summarizes the different situations concerning psychotropic and narcotic drugs (at the date of may 31, 2020). from march 17, 2020, some important problems rapidly appeared: because of the strict lockdown and repeated controls for any outing or trip, border shutdown for all extra european countries, but also with our immediate neighbours, drug trafficking has been drastically impacted, raising fears of an increase of episodes of withdrawal syndromes in the population of drug users. opioid maintenance treatment (omt) should be considered as an essential treatment during the covid-19 pandemic, as significant risks to the community exist with an interruption of the stable provision of opioid treatment. difficulties for omt drug provision have been expected with permanent changes of the prescribing and dispensing rules for narcotic drugs in the first days of lockdown, leading some patients to stock large amount of methadone at home. there may be an increased risk of opioid j o u r n a l p r e -p r o o f 6 overdose arising from i) erratic access to omt, ii) erratic access to illicit opioid supplies and iii) increased access to takeaway doses of methadone, which would have required the systematic prescription for take-home naloxone supplies [27] . there was also a growing concern about the risk of overdose with methadone (or of accidental exposure because of lockdown and provision of takeaway methadone at home), as methadone was already the first substance involved in drug abuse-related deaths before the disease outbreak, with an increasing trend in the last years [28] . unfortunately, despite drug approval for forms of naloxone directly available without medical prescription in 2017, the level of use of takeaway naloxone from specialized structures or community pharmacies remains very low [27] . psychological disturbances may occur due to the lockdown, with an increasing risk of misuse and abuse of psychoactive drugs in the population of drug users (including patients on omt particularly vulnerable to these disruptions), but also in the general population [29] . distress may result in some people increasing their substance use and subsequently require treatment (for example, alcohol use may increase). changes in illicit drug supply may occur due to a range of complex interacting factors, with an increased demand for services. alternately, some people who use drugs may be less likely to request services during the pandemic, with an escalation of substance use during a time of distress. some not evidence-based and potentially deleterious "guidelines" were launched in order to anticipate withdrawals, with several dangerous recipes for substitution or techniques to make provisions of narcotic drugs. such practices may bring new patterns of problematic use, including access to new psychoactive substances sold on the internet, with free home delivery services for using up stocks of illicit drugs. finally, in relation with the covid-19 itself, concerns arose about risk of drug-drug interactions and qt prolongation with methadone potentially combined with chloroquine and hydroxychloroquine or azithromycin. actually, when infected by sars-cov-2, older people, men and those with medical comorbidities (chronic pulmonary disease, cardiovascular disease, cerebrovascular disease, diabetes and a compromised immune system) present a much higher likelihood of acute respiratory distress, renal failure and death. due to the respiratory and pulmonary tropism of sars-cov-2, people who smoke or vape tobacco or cannabis products were expected to be more at risk of pulmonary complications. immune-suppressed people, for example, due to hiv infection or other chronic medication conditions, are also at increased risk for sars-cov-2 infection. consequently, drug users with these conditions may be a subgroup more at risk. we described the different events and facts collected and observed from mid-march to may 31, 2020. in the first days of lockdown, several concerns emerged in the field. the first events were identified through the continuity of the networking between the french addictovigilance centres and their partners (i.e. community pharmacies, general practitioners, specialized structures and emergency wards). by the second week of lockdown, several cases of methadone overdoses for people at home were reported, and falsified prescription forms to obtain hydroxychloroquine and azithromycin were also identified as osiap by different pharmacies on the french territory. these early signals have been transmitted to national health authorities, leading to the implementation of a weekly specific monitoring of noteworthy cases or events related to the covid-19, related to the lockdown, and of all falsified or abnormal prescription forms reported through the osiap survey during the period. this weekly monitoring was closely done between the french addictovigilance network and the ansm [30] . the lessons of this weekly monitoring by conference calls and shared minutes of the meeting are presented in the following paragraphs. the fig. 2 summarizes the highlights of this monitoring. detecting and identifying signals are a cornerstone for addictovigilance actors: they need to be able to label a piece of information received as a signal [5, 10, 20, 25, [31] [32] [33] . signals suggesting a public health risk are collected and analysed in continuous manner in a surveillance process implemented by watchdog or public health structures, in a perspective of alert, anticipation and early action. in this framework, a signal is defined as a piece of information concerning a health phenomenon or exposure to a risk or hazard, which requires investigation in order to validate it and decide whether or not it should be considered as an alert. the signals observed in addictovigilance may be related to human cases (unusual deaths, symptoms or syndromes grouped in clusters); to psychoactive substances or associations thereof likely to have serious health consequences (presence of adjuvants, degree of purity, novelty of the substance or its usage) and to new ways of administration or new settings of use. monitoring such noteworthy events is an important issue in addictovigilance. simad-covid was the national periodical assessment with the aim to proactively monitor and share occurrence of fatal and non-fatal overdoses due to opioid medications (methadone, opioid analgesics) or opioid substances (heroin) or other illicit drugs (cocaine). as soon as the lockdown began, first cases of overdoses were reported with opioids, mainly with methadone, and to a less extent, with heroine and other opioid analgesics (oxycodone, tramadol) or cough syrups containing codeine. until may 31 methadone was the most reported drug among overdoses. interestingly, several characteristics of methadone overdose have emerged: i) accessibility of methadone by storage from family/friends at home was often reported ii) occurrence of overdose among opioid naïve subjects (never previously exposed to opioids or return to use after cessation) iii) occurrence among vulnerable subjects (homelessness, migrants, patients with psychiatric comorbidities) iv) methadone used outside its labelling in france, for anxiolytic or analgesic purposes iv) take-home naloxone was exceptionally used in the period. it is important to note that during this period the price for street methadone remained relatively low, suggesting continued accessibility during the lockdown period compared to illicit drugs. heroin overdoses were also observed in several areas, often among previous heroin users (around 40-45 years old) leading to severe opioid toxidromes (acute renal failure, rhabdomyolysis, haemodialysis). the same trend was observed with cocaine leading to cardiogenic complications including a patient with covid-19 myocarditis. overdoses were reported among young adults after tramadol use alone or associated with other drugs (cannabis) or after concomitant codeine and promethazine use (purple drank). lockdown related noteworthy events "simad confinement"" consisted in clinical cases or other relevant information for which lockdown clearly played an important role, and concerned all other substances, whatever their nature (medications, illicit drugs, diverted drugs). during the lockdown period and until may 31, 2020, 231 reports were collected by the 13 french addictovigilance centres all over the country, including oversea territories. among the many substances identified at least once in these reports, pregabalin, benzodiazepines (including z drugs), cannabis, cocaine and nitrous oxide (n2o) were the most significant in terms of prevalence, seriousness or particularly specific to the lockdown context. -first signals of abuse of pregabalin (a gabapentinoid close to gabapentin, approved for the treatment of neuropathic pain, epilepsy and generalized anxiety disorder) were reported in france from 2011 with falsified prescriptions, medical nomadism and diverted use for psychoactive effect [34] [35] [36] . the french addictovigilance monitoring of pregabalin has shown, at the end of 2018, a dramatic increase in the number of cases of abuse, with the emergence of a population of young abusers. during the whole lockdown period and then afterwards, reports came from medical doctors who were urgently requested for prescription of pregabalin by young people, often minors, including migrants. this pregabalin addiction was not clearly identified before by these health professionals, since in the recent past reports came only from community pharmacists reporting abnormal prescription of lyrica ® . during the period, several cases of overdose were reported with pregabalin, including one requiring hospitalization with dyspnoea and hallucinations in a 17-year-old male. -benzodiazepines and z-drugs were expected to be highly consumed during the beginning of the lockdown in france, because of social isolation or psychological troubles due to the lockdown with the potential increase of marital conflicts and domestic violence. no withdrawal syndrome was reported (renewal of prescriptions was possible along the period), but abuse or misuse (with alcohol or other psychoactive substances) were reported. clonazepam alprazolam, oxazepam and zolpidem were the most frequently reported. -several reports concerned n2o indicating persistent diverted use during the lockdown due to i) a shortage of other substances in some areas and ii) a need to consume due to inactivity. on the other hand, difficulties to easily obtain large quantities of n2o cartridges led a 24-year-old male to abuse cocaine because of his craving. during the lockdown, it would appear that home deliveries have been made easier with internet orders. neurological complications with sensory-motor axonal polyneuropathies were also observed in the period, highlighting the spread of this new phenomenon of n2o addiction that has appeared in recent months [37] . -unexpectedly, reports concerning cocaine were numerous (more over than with heroine or cannabis), while supply constraints could be considered as the same as for other illicit substances. actually, this accessibility varied according to the regions, with cocaine easily available in some ones and with a wide disparity in cocaine concentration. the above described case of switching n2o to cocaine illustrates this greater availability of cocaine, with modified supply chains (home delivery instead of buying on the street from dealers). -cannabis supply was expected to be more difficult during lockdown. some patients reported withdrawal symptoms due to supply difficulties or an increase in prices, while others abused cannabis in a context of anxiety related to the outbreak. cases of accidental poisoning in children under 2 years of age who have accidentally ingested cannabis have been also reported. in addition to these most frequent substances, other reports confirm that after a short period of waiting, the drug trade has adapted to lockdown, and cases of abuse, misuse or deleterious consequences of use were reported with synthetic cathinone 3-mmc (n = 7), amphetamines (n = 8), lsd, ketamine and ghb (n = 1 each). finally, even if the number of reports seems quite low, it should be borne in mind that there is often a delay in reporting (cases that have occurred since lockdown break have not been reported by may 31, 2020) and that under-reporting in this area is very significant [38] . the two first reports collected through the osiap survey concerned out of date and falsified hydroxychloroquine prescription forms (presented during the first week of lockdown), in the context of media coverage about its hypothetic efficacy on sars-cov-2 [1, 2] . this first signal has been forwarded to the ansm at the end of march. from this date, all suspected falsified prescription forms identified by community pharmacies and reported to the 13 addictovigilance centres were centrally analysed weekly and compared to the information collected at the same period in 2019. as a reminder, osiap is one of the national program implemented by the french addictovigilance network in the 1990s to record all falsified prescriptions presented to a network of community pharmacies located all over the country [11, 12] . this monitoring program has been useful to identify addictovigilance signals or characterize the abuse potential of prescription drugs [32, [39] [40] [41] . usually, osiap are periodically collected each year (in may and november) on a voluntary basis by sentinel pharmacies [12] . outside these proactive collection periods, osiap are continuously reported by community pharmacies, regional health authorities or medical/pharmacy councils. the osiap intensive data collection planned for may 2020 was cancelled due to the lockdown. between march 16 and may 31, 2020, 178 falsified prescription forms were reported by community pharmacies to the french addictovigilance network, in a context of a significant decrease in the activity and travel limited to vital needs. this frequency must be considered with caution, as falsified prescriptions are often reported with a significant delay each year. in comparison, 634 prescription forms were collected in the same period in 2019, including the intensive data collection in may [12, 42] . fig. 3 presents the main frequently reported drugs during the covid-19 monitoring by weeks, compared to the same period in 2019 (estimated through the information available on may 31, 2020). during this period, the most frequently reported drugs were pregabalin, antitussive codeine syrup and analgesic codeine and tramadol. pregabalin and codeine syrups were mainly requested by a population of young males. this profile was similar to that observed in the covid and the lockdown noteworthy events, highlighting the emergence of a little-known population to health professionals [43] [44] [45] . the french addictovigilance network has published a national newsletter on addictovigilance news for several years ("bulletin d'addictovigilance"), which was issued four times in 2019 (january, april, september and october) and once in 2020 (january). table 2 summarizes the different topics discussed in these newsletters, which highlight the emergence or confirmation of addictovigilance signals in the recent months. in retrospect, the majority of bulletins have addressed substances that had been a problem during lockdown. throughout the lockdown and then, communication by the french addictovigilance network remained active with release of new national communications. the last issue of the national addictovigilance bulletin was entitled: "limitation of the prescription period of tramadol: how did we get there". this bulletin presented a summary of the data collected in france on tramadol between 2013 and 2018 and summarized the key elements which have led in particular to limit the duration of prescription of this drug. from april 15, 2020, the maximum prescription period for analgesics containing tramadol has been reduced from 12 to 3 months. continuation of treatment beyond 3 months will require a new prescription. following the results of the national addictovigilance monitoring of methadone, the french addictovigilance network has published a press release on the need to maintain access to methadone during the lockdown period, while ensuring the safety of its use. methadone is a mu opioid receptor agonist indicated for the substitution of opioid dependence. in france, for at least the past ten years, it has been the most frequently retrieved substance during the toxicological analyses of those involved in deaths linked to the excessive use of psychoactive substances (drames survey). the lockdown period may increase the risks linked to exposure to this drug in naïve-opioid subjects including children and those around them not treated with methadone. it should be remembered that the potentially lethal dose of methadone ingestion in a person who has never used opioids is estimated at 1 mg/kg body weight. the press release focused on the risk of overdose, due to the larger dispensed quantities, methadone "storage", consumption of larger quantities of methadone or other respiratory depressants (alcohol, benzodiazepines, other opioids, etc.), resort to illegal obtaining, risk of overdose in the event of resumption of methadone after a few days off, risk of serious poisoning in children or naïve subjects. the press release also highlighted the risk of qt prolongation increased because high doses of methadone itself and because of combination with drugs or substances which also modify qt: domperidone, macrolides (erythromycin, clarithromycin, etc.), antidepressants (citalopram, escitalopram), antihistamines (hydroxyzine), antipsychotics (haloperidol, quetiapine), as well as drugs currently tested against covid-19 in hospitals (hydroxychloroquine, azithromycin, lopinavir/ritonavir) or other psychoactive substances such as cocaine. in order to minimize these risks, the press release insisted on warning about purchase of these drugs outside the pharmaceutical circuit, and on the need to report treatment with methadone in case of hospitalization for sars-cov-2 suspicion. the press release also insisted on the urgent need to increase the distribution of naloxone to methadone consumers (see brochure about where and how find naloxone; fig. 4 ). on march 16, 2020, the french society of pharmacology and therapeutics has launched a national faqs website at https://sfpt-fr.org/covid19, focused on the proper use of drugs during the covid-19 pandemic [4] . the french addictovigilance network has joined the scientific council and has participated to document the responses to each question related to addictovigilance. one topic of the faqs was about opioid maintenance treatment, because drugs approved in this indication (methadone and buprenorphine) should be considered as essential medications during the covid-19 pandemic, and significant risks to the community exist with an interruption of the stable provision of opioid treatment. another topic was related to the accessibility of naloxone take home in france. another topic gave information on the risk to switch to other substances (cannabidiol or gabapentin) to manage cannabis withdrawal or to switch to opioid analgesics outside medical management for non-cancer pain [46] conclusion covid-19 epidemic has been an important challenge for addictovigilance. only part of the events that took place during this period have been reported to the french addictovigilance network, and it is likely that in the coming weeks or months the number of overdoses or deaths related to substance abuse will be higher than described in this article. this is of particular concern for methadone, heroin and pregabalin, but also for cocaine and nitrous oxide which seem to be more accessible than expected in this period. this addictovigilance monitoring has proved to be indispensable for warning health professionals at the local and regional level in order to limit the risk for users, and for alerting health authorities at the national level to points of vigilance in the field of psychoactive substances. adverse drug reactions of hydroxychloroquine: analysis of french prepandemic sars-cov2 pharmacovigilance data off-label" use of hydroxychloroquine, azithromycin, lopinavir-ritonavir and chloroquine in covid-19: a survey of cardiac adverse drug reactions by the french network of pharmacovigilance centers french society of pharmacology t. non-steroidal anti-inflammatory drugs, pharmacology, and covid-19 infection genesis of an emergency public drug information website by the french society of pharmacology and therapeutics during the covid-19 pandemic signal identification in addictovigilance: the functioning of the french system social media mining for toxicovigilance: automatic monitoring of prescription medication abuse from twitter comment on: an insight into z-drug abuse and dependence: an examination of reports to the european medicines agency database of suspected adverse drug reactions from psychoactive medicines to addictovigilance in french public health code the french system of evaluation of dependence: establishment in a legal system safety signal detection by the french addictovigilance network: innovative methods of investigation, examples and usefulness for public health medical prescriptions falsified by the patients: a 12-year national monitoring to assess prescription drug diversion network of centers for e, information p. survey of forged prescriptions to investigate risk of psychoactive medications abuse in france: results of osiap survey tamperresistant prescription forms for narcotics in france: should we generalize them? surveillance system on drug abuse: interest of the french national oppidum program of french addictovigilance network french network of centers for e, information on p. chemical submission: results of 4-year french inquiry décès directement liés aux drogues interest of large electronic health care databases in addictovigilance: lessons from 15 years of pharmacoepidemiological contribution ten-year trend of opioid and non-opioid analgesic use in the french adult population a capture-recapture method for estimating the incidence of off-label prescriptions: the example of baclofen for alcohol use disorder in france identification and tracking of addictovigilance signals in general practice: which interactions between the general practitioners and the french addictovigilance network? parachuting psychoactive substances: pharmacokinetic clues for harm reduction medical complications of psychoactive substances with abuse risks: detection and assessment by the network of french addictovigilance centres use of new psychoactive substances to mimic prescription drugs: the trend in france identifying life-threatening admissions for drug dependence or abuse (iliadda): derivation and validation of a model les caarud, lieux privilégiés d'émergence de signaux pour l'addictovigilance arrêté du 19 mars 2020 complétant l'arrêté du 14 mars 2020 portant diverses mesures relatives à la lutte contre la propagation du virus covid-19 intérêt de la mise à disposition de la naloxone auprès des usagers de drogues pour le traitement d'urgence de surdosage d'opioïdes améliorer la balance bénéfices/risques de la méthadone en respectant ses spécificités pharmacologiques psychopathological consequences of confinement pharmacovigilance et addictovigilance dans le contexte du covid-19 : une surveillance renforcée detection of signals of abuse and dependence applying disproportionality analysis early signal of diverted use of tropicamide eye drops in france pregabalin use disorder and secondary nicotine dependence in a woman with no substance abuse history patterns of gabapentin and pregabalin use and misuse: results of a population-based cohort study in france drug abuse monitoring: which pharmacoepidemiological resources at the european level? warning on increased serious health complications related to non-medical use of nitrous oxide use of multiple sources and capture-recapture method to estimate the frequency of hospitalizations related to drug abuse evidence of clonazepam abuse liability: results of the tools developed by the french centers for evaluation and information on pharmacodependence (ceip) network slow-release oral morphine sulfate abuse: results of the postmarketing surveillance systems for psychoactive prescription drug abuse in france example of an investigation of an "emergent" phenomenon in addiction vigilance: the case of methylphenidate medical prescriptions falsified by the patients: a 12-year national monitoring to assess prescription drug diversion pharmaciens d'officine, étudiants en pharmacie et demandes de médicaments à base de codéine : étude observationnelle disproportionality analysis for the assessment of abuse and dependence potential of pregabalin in the french pharmacovigilance database detecting the diverted use of psychoactive drugs by adolescents and young adults: a pilot study site de l'association française des centres d'addictovigilance the french addictovigilance network would like to acknowledge all persons in the 13 addictovigilance centres who participated in the active monitoring during this period (all health professionals who reported cases during the period, and persons in charge of psychoactive drugs at the ansm (aldine fabreguettes, emilie monzon, charlotte pion, nathalie richard). authors have no competing interest to declare key: cord-262787-3a3c8ee1 authors: ray, debashree; salvatore, maxwell; bhattacharyya, rupam; wang, lili; mohammed, shariq; purkayastha, soumik; halder, aritra; rix, alexander; barker, daniel; kleinsasser, michael; zhou, yiwang; song, peter; bose, debraj; banerjee, mousumi; baladandayuthapani, veerabhadran; ghosh, parikshit; mukherjee, bhramar title: predictions, role of interventions and effects of a historic national lockdown in india's response to the covid-19 pandemic: data science call to arms date: 2020-04-18 journal: nan doi: 10.1101/2020.04.15.20067256 sha: doc_id: 262787 cord_uid: 3a3c8ee1 importance: india has taken strong and early public health measures for arresting the spread of the covid-19 epidemic. with only 536 covid-19 cases and 11 fatalities, india a democracy of 1.34 billion people took the historic decision of a 21-day national lockdown on march 25. the lockdown was further extended to may 3, soon after the analysis of this paper was completed. objective: to study the shortand long-term impact of an initial 21-day lockdown on the total number of covid-19 cases in india compared to other less severe non-pharmaceutical interventions using epidemiological forecasting models and bayesian estimation algorithms; to compare effects of hypothetical durations of lockdown from an epidemiological perspective; to study alternative explanations for slower growth rate of the virus outbreak in india, including exploring the association of the number of cases and average monthly temperature; and finally, to outline the pivotal role of reliable and transparent data, reproducible data science methods, tools and products as we reopen the country and prepare for a post lock-down phase of the pandemic. design, setting, and participants: we use the daily data on the number of covid-19 cases, of recovered and of deaths from march 1 until april 7, 2020 from the 2019 novel coronavirus visual dashboard operated by the johns hopkins university center for systems science and engineering (jhu csse). additionally, we use covid-19 incidence counts data from kaggle and the monthly average temperature of major cities across the world from wikipedia. main outcome and measures: the current time-series data on daily proportions of cases and removed (recovered and death combined) from india are analyzed using an extended version of the standard sir (susceptible, infected, and removed) model. the esir model incorporates time-varying transmission rates that help us predict the effect of lockdown compared to other hypothetical interventions on the number of cases at future time points. a markov chain monte carlo implementation of this model provided predicted proportions of the cases at future time points along with credible intervals (ci). results: our predicted cumulative number of covid-19 cases in india on april 30 assuming a 1-week delay in people's adherence to a 21-day lockdown (march 25 april 14) and a gradual, moderate resumption of daily activities after april 14 is 9,181 with upper 95% ci of 72,245. in comparison, the predicted cumulative number of cases under "no intervention" and "social distancing and travel bans without lockdown" are 358 thousand and 46 thousand (upper 95% ci of nearly 2.3 million and 0.3 million) respectively. an effective lockdown can prevent roughly 343 thousand (upper 95% ci 1.8 million) and 2.4 million (upper 95% ci 38.4 million) covid-19 cases nationwide compared to social distancing alone by may 15 and june 15, respectively. when comparing a 21-day lockdown with a hypothetical lockdown of longer duration, we find that 28-, 42-, and 56-day lockdowns can approximately prevent 238 thousand (upper 95% ci 2.3 million), 622 thousand (upper 95% ci 4.3 million), 781 thousand (upper 95% ci 4.6 million) cases by june 15, respectively. we find some suggestive evidence that the covid-19 incidence rates worldwide are negatively associated with temperature in a crude unadjusted analysis with pearson correlation estimates [95% confidence interval] between average monthly temperature and total monthly incidence around the world being -0.185 [-0.548, 0.236] for january, -0.110 [-0.362, 0.157] for february, and -0.173 [-0.314, -0.026] for march. conclusions and relevance: the lockdown, if implemented correctly in the end, has a high chance of reducing the total number of covid-19 cases in the short term, and buy india invaluable time to prepare its healthcare and disease monitoring system. our analysis shows we need to have some measures of suppression in place after the lockdown for the best outcome. we cannot heavily rely on the hypothetical prevention governed by meteorological factors such as temperature based on current evidence. from an epidemiological perspective, a longer lockdown between 42-56 days is preferable. however, the lockdown comes at a tremendous price to social and economic health through a contagion process not dissimilar to that of the coronavirus itself. data can play a defining role as we design post-lockdown testing, reopening and resource allocation strategies. software: our contribution to data science includes an interactive and dynamic app (covind19.org) with shortand long-term projections updated daily that can help inform policy and practice related to covid-19 in india. anyone can visualize the observed data for india and create predictions under hypothetical scenarios with quantification of uncertainties. we make our prediction codes freely available (https://github.com/umich-cphds/cov-ind-19) for reproducible science and for other covid-19 affected countries to use them for their prediction and data visualization work. 622 thousand (upper 95% ci 4.3 million), 781 thousand (upper 95% ci 4.6 million) cases by june 15, respectively. we find some suggestive evidence that the covid-19 incidence rates worldwide are negatively associated with temperature in a crude unadjusted analysis with pearson correlation estimates [95% confidence interval] between average monthly temperature and total monthly incidence around the world being -0.185 [-0.548 the lockdown, if implemented correctly in the end, has a high chance of reducing the total number of covid-19 cases in the short term, and buy india invaluable time to prepare its healthcare and disease monitoring system. our analysis shows we need to have some measures of suppression in place after the lockdown for the best outcome. we cannot heavily rely on the hypothetical prevention governed by meteorological factors such as temperature based on current evidence. from an epidemiological perspective, a longer lockdown between 42-56 days is preferable. however, the lockdown comes at a tremendous price to social and economic health through a contagion process not dissimilar to that of the coronavirus itself. data can play a defining role as we design post-lockdown testing, reopening and resource allocation strategies. software: our contribution to data science includes an interactive and dynamic app (covind19.org) with short-and long-term projections updated daily that can help inform policy and practice related to covid-19 in india. anyone can visualize the observed data for india and create predictions under hypothetical scenarios with quantification of uncertainties. we make our prediction codes freely available (https://github.com/umich-cphds/cov-ind19) for reproducible science and for other covid-19 affected countries to use them for their prediction and data visualization work. four months since the first case of covid-19 in wuhan, china, the sars-cov-2 virus has engulfed the world and has been declared a global pandemic. 1 the number of confirmed cases worldwide stands at a staggering 1,930,780 (as of 9:20 am est april 14, 2020, microsoft bing coronavirus tracker 2 ). of these, 10,815 confirmed cases are from india (figure 1) , the world's largest democracy with a population of 1.34 billion (compare china at 1.39 billion and usa at 325.7 million). 3 india has been vigilant and wise in instituting the right public health interventions at the right time including sealing the borders with travel ban/canceling almost all visas, closing schools and colleges in certain states and diligently following up with community inspection of suspected/exposed cases with respect to adherence of quarantine recommendations ( table 1) . on march 24, india took the historic decision of a 21-day national lockdown starting march 25, when it had reported only 536 covid-19 cases and 11 fatalities. in the subsequent days we have seen a steady growth in the number of new cases and fatalities, with growth rates slower than other affected countries but in 21 days, the curve has not yet "turned the corner" or showed a steady decline in the number of newly diagnosed cases (figure 2 ). while india seems to have done relatively well in controlling the number of confirmed cases compared to other countries in the early phase of the pandemic (figure 2) , there is a critical missing or unknown component in this assessment: "the number of truly affected cases," which depends on the extent of testing, the accuracy of the test results and, in particular, the frequency and scale of testing of asymptomatic cases who may have been exposed. the frequency of testing has been low in india. according to the indian council of medical research (icmr), only 130,792 subjects have been tested as of april 9. 4 when there is no approved vaccine or drug for treating covid-19, entering phase 2 or phase 3 of escalation will have devastating consequences on both the already overstretched healthcare system of india, and india's large at-risk sub-populations (supplementary table 1 ). as seen for other countries like the us or italy, covid-19 enters gradually and then explodes suddenly. we provide a table listing other highly affected countries along with their first reported case, initial interventions, crude fatality rates, and active case counts in supplementary table 2 for reference. in this article, we take a data-driven approach to explore five extremely time-sensitive and important questions that india faces today in light of the covid-19 outbreak and the national lockdown: (a) how many cases can india expect at the end of the lockdown period? (b) when will the curve in india reach its apex and will the number of cases go back up after lockdown is lifted? (c) can summer temperatures thwart the outbreak in india? (d) how can the government and the people of india prepare for this crisis during and after the lockdown? (e) how critical is it to have reliable data, data science methods and tools as we envision a long-term strategy during and after the lockdown? this work is the result of the collective public health conscience of a group of interdisciplinary researchers in different parts of the us and in india. we convened virtually after being quarantined in our homes with alternating waves of fear and inspiration surrounding us. we decided to channel our collective energy to study the defining public health and economic crisis of our time and use our data science expertise to search for answers and solutions that can help covid-19 related policymaking in india. this is our contribution and public service as data scientists. our data science product includes two articles on medium pre 5 and post 6 lockdown announcement, providing critical information for policymakers (reuters, 7 times of india, 8 the guardian, 9 the economic times 10 ) and an interactive app that daily updates forecasts as new case counts are coming in, and publicly available codes for reproducible research. we used the current daily data on number of covid-19 cases, recoveries and deaths in india to predict the number of cases at any given time. 11 we obtained the data (up to april 7) from the 2019 novel coronavirus visual dashboard operated by the johns hopkins university center for systems science and engineering (jhu csse). 12,13 for our temperature analysis, these counts were aggregated to a month-level for each country, that is, we look at the total number of new cases in the months of january, february and march for each country. we obtained the monthly average temperature for major cities in the countries with covid-19 outbreak from wikipedia. 14 we analyzed the data from india with standard epidemiologic tools of modeling disease transmission and estimating the theoretical number of cases at any time. one such epidemiologic model is the susceptible-infected-removed (sir) model, which is guided by a set of differential equations relating the number of susceptible people, the number of infected people (cases) and the number of people who have been removed (either recovered or dead) at any given time. recently, this standard sir model was extended to incorporate time-varying transmission rates or timevarying quarantine protocols and is known as the esir model. 11 when using the esir model with time-varying disease transmission rate, it can depict a series of time-varying changes caused by either external variation like government-initiated macro isolation measures, community-level protective measures and environment changes, or internal variations like mutations and evolutions of the pathogen. the r package for implementing this general model for understanding disease dynamics is publicly available at https://github.com/lilywang1988/esir. to implement the esir model, a bayesian hierarchical framework is assumed where the proportions of infected and the removed people are modeled using a beta-dirichlet state-space model while a latent dirichlet distribution is assumed for the underlying unknown prevalence of the three states. priors for the basic reproductive number r0, disease removal rate (consequently, the transmission rate) and the underlying unobserved prevalence of the susceptible, infected and removed states at the starting time are considered. using the current time series data on the proportions of infected and the removed people, a markov chain monte carlo implementation of this bayesian model provides not only posterior estimation on parameters and prevalence of all the three compartments in the sir model, but also predicted proportions of the infected and the removed people at future time point. the posterior mean estimates of the unobserved prevalence at both observed as well as future time points come along with 95% credible intervals (ci). to get predicted case-counts from the predicted prevalence, we used 1.34 billion as the population of india, thus treating the country as a homogeneous system for the outbreak. 15 we made projections of the cumulative number of cases over a time horizon to assess the shortterm impact of lockdown as well as the long-term impact of lockdown and post-lockdown activities. for the short-term forecast on april 30, we assumed lockdown is implemented until april 14 with either a 1-or a 2-week delay in people's adherence/compliance to lockdown restrictions. we compared these projections with two hypothetical scenarios: (a) no nonpharmaceutical intervention (i.e., a constant disease transmission rate over time since the first case was reported in india), (b) a moderate intervention with social distancing and travel bans only (i.e., a decreased transmission rate compared to no intervention). for the no intervention and the moderate intervention scenarios, we chose the transmission rate and the removal rate such that the means for the prior distribution of the basic reproductive number r0 (the expected number of cases generated by one infected person assuming that the whole population is susceptible) are 2.0 and 1.5 respectively [the change in r0 was created based on what we saw in wuhan 16 ]. the value of 2.0 was estimated based on the early phase data in india. for the current scenario of lockdown, our chosen mean for r0 prior starts with 2.0 during the period of no intervention, drops to 1.5 during the period of moderate intervention, and further drops to 0.8 during the 21-day lockdown period, and moves back up to 1.5 after the lockdown ends as described in figure 3 (assuming a gradual, moderate resumption of daily activities). for the longer-term forecast until june 15, we considered three hypothetical post-lockdown scenarios: (i) people return to normal activities due to the urgent desire for reconnecting after lockdown; (ii) people return to moderate activities as they did during the period with social distancing and travel ban intervention; and (iii) people make a cautious return out of fear for the coronavirus and partake in subdued activities. for these three scenarios, we assume mean for r0 prior moves back up from 0.8 to 2.0, 1.5 and 1.2 respectively three weeks after lockdown ends on april 14. we compared these post-lockdown scenarios with another hypothetical scenario involving perpetual social distancing and travel ban only without any lockdown (we fixed the mean for r0 prior at 1.5 over the entire intervention interval). the changes to r0 values across our simulation scenarios are depicted in figure 3 . to assess the long-term impact of lockdown duration, we considered four scenarios: 21-, 28-, 42-, and 56-day lockdown periods. in all scenarios, we assume mean for r0 prior remains at 0.8 for the duration of the lockdown and returns to 1.5 three weeks after the lockdown period ends (analogous to the "moderate return" scenario). the changes to r0 values across our simulation scenarios are depicted in supplementary figure 4 . there are many hypotheses regarding the slow growth rate of covid-19 cases in many countries, particularly low-and middle-income countries (lmics). some of these hypotheses include the use of bacille calmette guerin (bcg) vaccine, 17 younger population, 18 high daily temperature, 19 use of anti-malarials 20 and host genetics. 21 here, we only explore the temperature hypothesis related to covid-19 incidence. we assessed any correlation between country-wise average monthly temperature and total incidence of covid-19. the monthly average temperature for major cities across the world was used to compute the monthly average temperature for each country experiencing covid-19 outbreak by averaging across the major cities within a country. missing data for average temperature for certain countries was manually appended from www.weatheratlas.com. 22 we computed the pearson correlation coefficient, , between the average monthly temperature and total monthly incidence during each month of january, february and march. we used the fisher's z-transformation to compute z = 0.5 log + ,-. ,/. 0. the standard deviation of , which is known under certain normality assumptions, is used to construct a 95% confidence interval for . the inverse transform of is then used to obtain the 95% confidence interval for the correlation . all calculations were carried out in the rstudio platform. under national lockdown (march 25 -april 14), our predicted cumulative number of covid-19 cases in india on april 30 are 9,181 and 11,626 (upper 95% ci of 72,245 and 84,245) assuming a 1-or 2-week delay (i.e., either a quick or a slow adherence), respectively, in people's adherence to lockdown restrictions and a gradual, moderate resumption of daily activities post-lockdown (figure 4, supplementary figure 1 ). in comparison, the predicted cumulative number of cases under "no intervention" and the "intervention involving social distancing and travel bans without lockdown" are 358 thousand and 46 thousand (upper 95% ci of nearly 2.3 million and 0.3 million) respectively. we are reporting only the upper credible limit here and elsewhere since the lower credible limits are very close to 0 due to the large uncertainty in our predictions arising from many unknowns. we also believe that our point estimates are at best underestimates due to potential under-reporting of case-counts and our model not taking into account the population density, agesex and contact network structure of the whole nation. increase in testing and community transmission may lead to a spike in a single day and that may shift the projection curve significantly upward. regardless of the exact numbers it is clear that, the 21-day lockdown will likely have a strong effect on reducing the predicted number of cases in the short term. we took a close look at what might be coming in the next few months, based on what we have seen in other countries and an epidemiological model that has been gainfully employed to assess the effect of interventions in hubei province. 11 we estimated that 26 (upper 95% ci 135) and 177 (upper 95% ci 2869) cases per 100,000 are avoided by may 15 and june 15, respectively, by instituting a 21-day lockdown with a 1-week delay in people's adherence and a cautious release compared to perpetual social distancing and travel ban (without lockdown) ( figure 5 ). this boils down to preventing roughly 343 thousand (upper 95% ci 1.8 million) and 2.4 million (upper 95% ci 38.4 million) covid-19 cases nationwide by may 15 and june 15, respectively. without some measures of suppression after lockdown is lifted, the impact of lockdown in bringing down the case-counts (the now ubiquitous term, "flattening the curve") can be negated by as early as the first week of june. in fact, in figure 5a , the pre-intervention ("normal") curve first passes the social distancing and travel ban curve on june 8. in particular, if people immediately go back to pre-intervention ("normal") activities post-lockdown, a surge in the predicted case-counts is expected in the long-term beyond what we would have seen if there were only social distancing and travel ban measures without lockdown (7.1 million when post-lockdown activity returns to pre-intervention levels vs. 6.9 million under social distancing and travel ban without a lockdown period on july 31; figure 5 ). we estimated that 195 (upper 95% ci 3494) and 380 (upper 95% ci 6463) cases per 100,000 are avoided by june 15 and july 15 respectively if people are cautious in their activities post-lockdown compared to the scenario where people return to normal preintervention activities. long-term forecasting under slow adherence (2-week delay) can be seen in supplementary figure 2 . . cc-by-nd 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/10.1101/2020.04.15.20067256 doi: medrxiv preprint we took the quick adherence epidemiological models and compared the 21-day lockdown with hypothetical 28-, 42-, and 56-day lockdown scenarios (figure 6 ). when comparing a 21-day lockdown with a hypothetical lockdown of longer duration, we find that 28-, 42-, and 56-day lockdowns can approximately prevent 238 thousand (upper 95% ci 2.3 million), 622 thousand (upper 95% ci 4.3 million), 781 thousand (upper 95% ci 4.6 million) cases by june 15, respectively. a 28-day lockdown does not appear to have a significant impact on cumulative case counts when compared to a 21-day lockdown. however, purely from an epidemiologic perspective, there appears to be some evidence that suggests a 42-or 56-day lockdown would have a more meaningful impact on reducing cumulative covid-19 case counts in india. we note that longer lockdown periods are accompanied by increasing costs to individuals -notably economicand must be considered. our models suggest that some form of post-lockdown suppression (e.g., extension of social distancing measures, limits of gathering size, etc.) is necessary to observe longterm benefits of the lockdown period. lockdown duration study under the slow adherence (2-week delay) scenario can be found in supplementary figure 3 . we did explore some alternative assumptions and conducted thorough sensitivity analysis before settling on the models presented above. in one example, we assumed that there are actually 10 times the number of reported cases to date to reflect potential underreporting of cases due to lack of testing. in another scenario, we assumed these cases occurred in metropolitan areas to reflect a potential intensification of case clustering. in yet a third scenario, we hypothesized that r0 prior starts with 2.5 instead of 2.0 (i.e., a single infected individual would infect 2.5 susceptible individuals, on average, instead of 2). these scenarios did not appreciably change our conclusions in broad qualitative terms, though the exact quantitative projections are quite sensitive to such choices. across these scenarios, the projected total number of infected cases by the entire first phase of the pandemic varied between 2-15% of the population, again showing the significant variability in these numbers. the estimates we present here may appear conservative and are at best underestimates, and, in all cases, our confidence in these projections decreases markedly the farther into the future we try to forecast. it is extremely important to update these models as new data arise. spatial plots for the average monthly temperatures accompanied by total monthly incidence across all countries from january through march indicate a suggestive pattern of increase in community spread across cities and regions specifically along narrow north east-west directions (figure 6 ). countries in these regions consistently exhibit similar weather patterns. however, in the context of india, a gradual rise in the number of cases is observed starting from january through march. . cc-by-nd 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 . the estimates and the 95% confidence interval for the correlation coefficient for january was -0.185 [-0.548, 0.236] with 24 countries having non-zero incidence, for february was -0.110 [-0.362, 0.157] with 56 countries having non-zero incidence, and for march was -0.173 [-0.314, -0.026] with 175 countries having non-zero incidence. although the estimates were negative, the 95% confidence intervals either include zero or the upper limit is close to 0, indicating weak evidence for any claim of negative association between case counts and daily temperature. any such affirmation will require further data and investigation that accounts for many possible sources of confounding. our projections using current daily data on case counts until april 7 in india show that the lockdown, if implemented correctly in the end, has a high chance of reducing the number of covid-19 cases in the short term and buy india invaluable time to prepare its healthcare and disease monitoring system. in the long-term, we need to have some measures of suppression in place after the lockdown is lifted to prevent a massive surge in the number of cases that can quickly overwhelm an already over-stretched indian healthcare system resulting in increased fatalities. specific vulnerable populations will be at higher risk of severity and fatality from covid-19 infection: older persons and persons with pre-existing medical conditions (e.g., high blood pressure, heart disease, lung disease, cancer, diabetes, immunocompromised persons). 24,25 supplementary table 1 provides a description of the approximate number of individuals in these high-risk categories in india. beyond the fragile population characterized by health and economic indicators, we have to remember that healthcare workers and first responders at the front line of this pandemic are amongst the most vulnerable. 16 it is important to note that a massive surge in the number of cases can quickly overwhelm an already over-stretched indian healthcare system. the estimated capacity of hospital beds in india is 70 per 100,000, 26 which is an upper bound on treatment capacity. given an average occupancy rate of 75%, only a quarter of these are available. 27 moreover, critically ill covid-19 patients (about 5-10% of those infected) will require icu beds and ventilator support. india has only 35-58 thousand icu beds with very high occupancy rates and at most 1 ventilator per 2 icu beds. 28 from a purely public health perspective, this analysis shows the impact and necessity of lockdown and subsequent measures of suppression after lockdown is lifted. all the people in india, regardless of their vulnerability to covid-19, need to adhere to the public health guidelines issued by the ministry of health and family welfare in india, and continue to be cautious in their post-lockdown activities to guarantee a long-term benefit of the national lockdown. currently, there are many hypotheses regarding differential covid-19 infection rates and mortality rates across countries. one such hypothesis is that the bcg vaccine -developed a century back for tuberculosis -has a protective effect on the prevalence of covid-19 and related . cc-by-nd 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 . mortality. 29 a recent pre-print found covid-19 attributable mortality in countries with bcg policy is 6 times lower than those without a bcg policy in an ecological analysis, after accounting for country-specific confounders like economic status, percentage elderly (those aged â�¥65 years) in the population, and relative position of each country along the epidemic trajectory. 17 however, the authors caution against over-interpretation of this negative association between bcg use and covid-19 due to limitations of country-level analysis and many sources of unmeasured confounders. another hypothesis is that much like the flu virus, summer temperatures will thwart the covid-19 outbreak. our analysis, based on current data, suggests we cannot rely on the hypothetical prevention (with inconclusive evidence) governed by meteorological factors and need public health actions, regardless of the seasonal weather. the management of this covid-19 crisis requires strong partnership of the government, the scientific community, the health care providers and all citizens of india (and all global citizens). long term surveillance and management of covid-19 crisis is needed with not just public health in mind but also to take care of the economic, social, and psychological trauma that it will leave on the people. reviving the economy will be critical in the coming months. below we recommend some healthcare, social and economic reforms that can counter the negative impact of severe public health interventions, some of which india has already begun to make progress. (a) aggressively increase the number of tests administered daily as there are often asymptomatic cases who are spreading the infection without knowing. it is of utmost importance that india adopt widespread testing to identify and isolate the infected. rt-pcr diagnostic test can provide reliable and faster diagnosis of the sars-cov-2 virus. 30 large scale antibody testing should be launched to assess the true scale of this pandemic. 31 the instrument of isolating nearly everyone with a near universal lockdown not only leads to livelihood losses for millions of families but also starvation for others. as we reopen the country, testing high contact, high density areas and setting up a clever surveillance system is critical. immediately prepare to protect the health care workers and first responders who are at the front line of this pandemic. this involves ensuring a steady supply chain of medical resources (masks, gloves, gowns, ventilators), and protecting our healthcare workers physically and psychologically. full gears (protective suit, medical goggle, cap, face shield, mask and gloves) are absolutely essential when seeing suspected cases. these protection strategies worked in china. 32 (c) reduce all non-essential medical care and expand number of hospital beds, icu beds and ventilators. (d) continue to set up covid-19 testing mobile labs, hospitals and mobile cabins (e.g. by converting stadiums and trains into quarantine and treatment facilities). 33 ensure the healthcare facilities have adequate supply of medications that are currently . cc-by-nd 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 . being recommended. for instance, antiviral drugs "remdesivir and chloroquine are highly effective in the control of 2019-ncov infection in vitro" indicating promise for treating covid-19 patients. 34 recently though, the study 35 finding hydroxycholoroquine as an effective treatment for covid-19 has been retracted for bad study design and not meeting expected scientific standards. 36 (f) use pragmatic real-time data for optimally deploying surveillance, community inspection and health care resources. this is key with limited resources. economic recommendations: (j) provide livelihood assistance over the quarantine period to those who test positive. this will incentivize people to get tested and comply with social isolation protocols. for many people in india, loss of several weeks of earnings can be economically devastating and since symptoms are mild for most infected people, it is unreasonable to expect that all people will tightly follow restrictions unless economically protected. to get a ballpark idea of the fiscal burden involved, assume 1 million detected cases, quarantine of 6 weeks per patient, and inr 10,000 monthly compensation. this adds up to a bill of inr 15 billion, which is roughly 2.5% of the annual healthcare budget of the central government. (k) during periods of social distancing and lockdowns, there is grave livelihood threat to a lot of poor people even if they are uninfected -street hawkers, auto drivers, barbers and shopkeepers, etc. providing a universal basic income (ubi), or some mildly means tested version of it, over the period of disruption is needed for a successful lockdown. (l) to prevent shutdowns in badly affected sectors, the government may provide goods and services tax (gst) credit to firms based on the difference between past and current sales. once the pandemic is over and normal business resumes, expansionary monetary and fiscal policy will be needed to revive macroeconomic health. there are many epidemiological models to predict the course of an infectious disease and even many that are india-specific. 37-39 some use age-structure, contact patterns, spatial information to finesse their prediction. some consider the possibility of a latent number of true cases, only a fraction of which are ascertained and observed. 16 the model we used here is an extension of a . cc-by-nd 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 . standard sir model, called esir model, 11 where we can create hypothetical intervention scenarios in a time dependent manner. the goal of any intervention is to reduce the chance that an infected person meets a susceptible person. we create models for declines/drops in contact probabilities when an intervention is rolled out. thus, there is some intrinsic ad-hocery to our assumptions. any statistical model is wrinkled with such assumptions. similarly, the predictions themselves have large uncertainty (as reflected by the upper credible limits). as we interpret the numbers from any model, let us use caution in not over-interpreting them. a rigorous quantitative treatment often allows us to analyze a problem with clarity and objectivity, but we recommend focusing more on the qualitative takeaway messages from this exercise rather than concentrating on the exact numerical projections or quoting them with certainty. we see tremendous role of data and data science in governing policy as india reopens post lockdown. the release from lockdown will not be in a binary switch on/switch off process but a modulated slow-varying process. we see the following roles and opportunities for pragmatic use of data science in the post-lockdown phase. (a) flexible, athletic, data driven policymaking will need up-to-date numbers and projections at hand, which require granular data, automation and data transparency (b) understand uncertainty in numbers: all models are wrong, some are useful, but note that takeaway messages for intervention forecasting are often the same 40 (c) using technology to create body temperature/expected health status map (e.g., healthweather.us) 41 (d) assess adherence to social distancing using mobile networks, google (e.g., google mobility reports) 42 (e) use survey to identify potential super-spreaders, manage contact network, oversample high risk areas for testing (f) install syndromic surveillance in hospitals, medical claims systems to set up alert; establish expected number of respiratory and flu-like illnesses so departures can set off an alarm. (g) use community health workers in rural areas, community dwellings to identify and isolate cases and conduct cluster testing (h) targeted communication strategies: regarding tests, treatments, contagion level. misinformation and incorrectly analyzed data lead to panic (i) accurate and consistent reporting of case counts and deaths due to covid-19 are extremely critical our statistical modeling and forecasts are not without limitations. we have very few data points and a long time-window to extrapolate for the long-term forecasts. the uncertainty in our predictions is large due to many unknowns arising from model assumptions, population . cc-by-nd 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 . demographics, the number of covid-19 diagnostic tests administered per day, testing criteria, accuracy of the test results, and heterogeneity in implementation of different government-initiated interventions and community-level protective measures across the country. we have neither accounted for age-structure, contact patterns or spatial information to finesse our predictions [37] [38] [39] nor considered the possibility of a latent number of true cases, only a fraction of which are ascertained and observed. 16 increase in frequency and scale of testing, and community transmission of the sars-cov-2 virus may lead to a spike in a single day and that can shift the projection curve significantly upwards. covid-19 hotspots in india are not uniformly spread across the country, and state-level forecasts 43 may be more meaningful for state-level policymaking. we are assuming that the implementation and effects of public health interventions and policies are the same everywhere in india by treating india as a homogeneous unit. future opportunities for improving our model include incorporating contagion network, age-structure, estimating seir model, incorporating test imperfection, and estimating true fatality/death rates. regardless of the caveats in our study, our analyses show the impact and necessity of lockdown and of suppressed activity post-lockdown in india. rather than over-interpreting exact numerical projections, we recommend focusing more on the qualitative takeaway messages. one ideological limitation of considering only the epidemiological perspective of controlling covid-19 transmission in our model is the inability to count excess deaths due to other causes during this period, or the flexibility to factor in reduction in mortality/morbidity due to some other infectious or flu-like illnesses, traffic accidents or health benefits of reduced air pollution levels. a more expansive framework of a cost-benefit analysis is needed as we gather more data and build an integrated landscape of population attributable risks. finally, in our strong commitment to reproducibility and dissemination of our research, we have made the code for our predictions available at github (https://github.com/umich-cphds/cov-ind-19) and created an interactive and dynamic r shiny app (covind19.org) to visualize the observed data and create predictions under hypothetical scenarios with quantification of uncertainties. these forecasts will get updated daily as new data come in. we hope these products will remain our contribution and service as data scientists during this tragic global catastrophe, and the model and methods will be used to analyze data from other countries. our epidemiologic and mathematical calculations make a convincing case for enforcing the 21day national lockdown of the largest democracy in the world, acting early, before the growth of covid-19 infections in india starts to accelerate. we also notice the public health benefit of extending the lockdown by 3-5 weeks in our projections. measures of suppression are needed postlockdown to get long-term benefits from the lockdown. however, these draconian public health . cc-by-nd 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/10.1101/2020.04.15.20067256 doi: medrxiv preprint measures come at a tremendous price to social and economic health that can last months or even years after the restrictions on social mobility are lifted. thus, there is an urgent need for social and economic immunity: not just coverage for testing and treatment for covid-19 for everyone in india, but subsidies and incentives for the common man to survive the consequences of the severe interventions that are needed to stop the coronavirus from creating a massive catastrophe in india. we also illustrate the critical role of data in aiding policy decisions. finally, our message to the public is to proceed with prudence and caution, and not panic or drown in despair. we should draw hope from the success of south korea and china and the initial promising containment in india. we need to support the community around us and help the government of india to manage the crisis with the best strategies, resources and science. the lockdown has given us time to prepare and act, let us make the best use of it. we are still in a state of national and global emergency and it will take a considerable time for humanity to recover from this global pandemic and return to normalcy. in the meantime, we root for public health, for innovation and science, for home testing kits [there is none yet], 44 for fda approved drugs [solidarity trial], 45 and for a vaccine [5 clinical trials ongoing]. 46 in these frightening times, we find inspiration in the power of the common people and the magic of human kindness. 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 . 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 . . cc-by-nd 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 . . cc-by-nd 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 . figure 7 . left: country-wise total monthly incidence of covid-19 in the months of january, february and march. the horizontal lines approximately indicate the equator, the tropic of cancer and the 60n latitude. right: average monthly temperature (in c) during the months of january, february and march. these maps were created by smoothing the counts as well as the average monthly temperature across geographical locations by spatial interpolation. . cc-by-nd 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 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 . . cc-by-nd 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 . cc-by-nd 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/10.1101/2020.04.15.20067256 doi: medrxiv preprint supplementary figure 1 . short-term daily growth in cumulative case counts in india assuming a 2-week delay in people's adherence to restrictions. observed data are shown for days up to april 7. predicted future case counts for april 8 until april 30 are based on observed data until april 7 using the esir model. . cc-by-nd 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 . figure 3 . cumulative (a) and incidence (b) graphs for forecasting models assuming a 2-week delay under 21-, 28-, 42-, and 56-day lockdown scenarios using observed data through april 7. . cc-by-nd 4.0 international license it is made available under a author/funder, who has granted medrxiv a license to display the preprint in perpetuity. 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/10.1101/2020.04.15.20067256 doi: medrxiv preprint who director-general's opening remarks at the media briefing on covid-19 -11 microsoft bing covid-19 tracker. microsoft corporation data for india, united states sars-cov-2 (covid-19) testing: status update 09 aprilr predictions and role of interventions for covid-19 outbreak in india historic 21-day lockdown, predictions for lockdown effects and the role of data in this crisis of virus in india india faces spike in coronavirus cases, says study, in test for health system epidemiologic models show we need aggressive measures in the early phase...lockdown buys us time overcome by anxiety: indians in lockdown many can ill afford. the guardian india could see a reduction in the number of coronavirus cases by next week: study. the economic times an epidemiological forecast model and software assessing interventions on covid-19 epidemic in china an interactive web-based dashboard to track covid-19 in real time list of cities by average temperatures evolving epidemiology and impact of non-pharmaceutical interventions on the outbreak of coronavirus disease differentiall covid-19-attributable mortality and bcg vaccine use in countries case-fatality rate and characteristics of patients dying in relation to covid-19 in italy high temperature and high humidity reduce the transmission of covid-19 regulators split on antimalarials for covid-19 covid-19): are you at higher risk for severe illness? the world bank hospital utilizationn statistics: thirty-five year trend analysis, a measure of operational efficiency of a tertiary care teaching institute in south india coronavirus: does india have enough ventilators, hospital beds? the times of india can an old vaccine stop the new coronavirus? the new york times centers for disease control and prevention. cdc 2019-novel coronavirus (2019-ncov) real-time rt-pcr diagnostic panel new blood tests for antibodies could show true scale of coronavirus pandemic analysis of 25,000 lab-confirmed covid-19 cases in wuhan: epidemiological characteristics and non-pharmaceutical intervention effects india turns trains into isolation wards as covid-19 cases rise remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-ncov) in vitro hydroxychloroquine and azithromycin as a treatment of covid-19: results of an open-label non-randomized clinical trial hydroxychloroquine-covid-19 study did not meet publishing society's "expected standard prudent public health intervention strategies to control the coronavirus disease 2019 transmission in india: a mathematical model-based approach age-structured impact of social distanncing on the covid-19 epidemic in india covid-19 for india updates coronavirus statistics: what can we trust and what should we ignore? the guardian covid-19 community mobility reports food and drug administration. coronavirus (covid-19) update: fda alerts customers about unauthorized fradulent covid-19 tests world health organization. who director-general's opening remarks at the media briefing on covid-19 -27 covid-19 vaccine tracker. regulatory affairs professionnals society hypertension (women)* 132 â�  based on 2020 est. of 1.38 billion from un department of economic and social affairs * age-standardized defined as within 5-kilometer distance of home or work abbrev.: copd, chronic obstructive pulmonary disease international diabetes federation; nicpr, national institute of cancer prevention and research ⶠdate of 1 st case data obtained from jhu csse time series data on covid-19 â�  microsoft bing covid-19 tracker the authors will like to thank the university of michigan advanced research computing services for enabling daily updates to our models and allocating us abundant computational resources. we will also like to thank professor matthew fox from the boston university school of public health for his valuable comments on our rshiny app, key: cord-308493-3fsn7awq authors: günther‐bel, cristina; vilaregut, anna; carratala, eduard; torras‐garat, sonia; pérez‐testor, carles title: a mixed‐method study of individual, couple and parental functioning during the state‐regulated covid‐19 lockdown in spain date: 2020-07-17 journal: fam process doi: 10.1111/famp.12585 sha: doc_id: 308493 cord_uid: 3fsn7awq during the recent covid‐19 outbreak in spain we explored the individual and relational wellbeing of people confined together with their partners and/or children during the first three weeks of state‐regulated lockdown. adults 18 years or older (n=407) completed an online survey that included demographic, household, and employment information along with standardized measures of psychological distress (state‐trait anxiety, beck depression) and relationship functioning – either the dyadic adjustment scale if there were no children in the household or a basic family relations evaluation questionnaire (cerfb) measuring conjugal, parental, and co‐parental functions. qualitative analyses of responses to an open‐ended question about perceived changes in couple or family dynamics during lockdown revealed 9 specific themes comprising two overarching categories: relational improvement and deterioration. the overall prevalence of improvement themes (61.7%) exceeded deterioration themes (41.0%), with increased (re)connection and conflict atmosphere cited most often. quantitative analyses found elevated levels of state anxiety but not trait anxiety or depression during lockdown. consistent with the qualitative results, couples having no children at home reported high levels of dyadic adjustment, but with children present cerfb parental functioning exceeded conjugal functioning, a pattern sometimes associated with child triangulation into adult conflicts. although correlates of psychological distress (e.g., unemployment, perceived economic risk) were relatively stable across sub‐groups, predictors of relationship functioning varied substantially with household/parental status (e.g., telecommuting and employment facilitated conjugal functioning only for couples with children). according to the world health organization (who, 2020a), a pneumonia of unknown cause was first detected in wuhan, china in december 2019. after rapid escalation, the who (2020b) declared the novel coronavirus disease (covid-19) a global pandemic. in the context of this outbreak, governments around the world have taken measures to prevent and control the covid-19 infection. starting in china, one such measure has been to place entire cities under mass quarantine. spain has been one of the worst affected european countries. the virus spread to all spanish regions, with the communities of madrid and catalonia suffering the highest number of cases. on march 14th 2020, the spanish government formally declared a state of alarm over covid-19 and ordered a state-regulated lockdown defined as home confinement, in which everyone, including those who were asymptomatic, had to stay confined at home for the next two weeks. the only exceptions were for basic activities like buying food or medicine, attending medical centers, or commuting to work (boe, 2020) . schools, universities and other education institutions were also on lockdown, going online in most cases. one week into the state of alarm, as covid-19 taxed the country's oversaturated health care system, spain's government tightened quarantine even further, ordering all non-essential workers to stay home for two additional weeks with the possibility of extending the emergency measures until outbreak remission. in addition to the biomedical and epidemiological benefits of mandatory mass quarantine, it is prudent to consider possible psychological and behavioral impacts (rubin & wessely, 2020) . in a prompt and clarifying article, brooks, webster, smith, woodland, et al. (2020) have reviewed scientific literature relevant to effects of quarantine on individual mental health, aiming to facilitate decision making in the current global crisis. based on 24 quantitative and qualitative studies across 10 countries where quarantines involved direct or potential exposure to sars, ebola, h1n1 influenza pandemic, middle east respiratory syndrome, or equine influenza, brooks et al. (2020) conclude that quarantine often has negative psychological effects, including anxiety, depressed mood, irritability, insomnia, symptoms of post-traumatic stress, and emotional exhaustion. indeed, a recent large-scale survey during the covid-19 pandemic in china found that well over a third of the general population experienced significant psychological distress, though it is unclear how much of this was related to lockdown (qiu, this article is protected by copyright. all rights reserved shen, zhao, wang, et al., 2020) . in may of this year a similar proportion of american adults reported clinical anxiety or depression according to the u.s. census bureau (fowers & wan, 2020) . the literature cites many factors that could have negative mental health consequences during or after quarantine including fear of infection (bai, lin, lin, chen, et al., 2004; desclaux, badji, ndione, & sow, 2017) , boredom and isolation (cava, fay, beanlands, mccay, & wignall, 2005; digiovanni, conley, chiu, & zaborski, 2004) , financial insecurity (jeong, yim, song, min, et al., 2016; mihashi, ostubo, yinjuan, nagatomi, et al., 2009) , limited access to basic supplies (jeong et al., 2016) , and confusing public information (blendon, benson, desroches, raleigh & taylor-clark, 2004; jeong et al., 2016) . another factor particularly relevant to the present study is quarantine duration (hawryluck et al., 2004; marjanovic, greenglass, & coffey, 2007; reynolds, garay, damond, moran, et al., 2008) . for example, hawryluck et al. (2004) found significantly more symptoms of posttraumatic stress among people locked down for more than 10 days compared with those in quarantine less than 10 days. in spain we collected data through the first three weeks of intensive covid-19 lockdown. although the literature emphasizes psychological effects of quarantine on individuals, there are good reasons to consider implications for couple and family relationships as well (sprang & silman, 2013) . for example, a global times (2020) newspaper article reported unprecedented divorce rates in some districts of xi'an, the capital of northwest china's shaanxi province, as a direct repercussion of covid-19. similarly, in qualitative studies, participants described ebola containment measures in liberia as creating mutual distrust, even between family members (pellecchia, crestani, decroo, van den bergh, & al-kourdi, 2015) ; and toronto health care workers quarantined following exposure to sars reported disruptions in parental roles and routines, "creating stress for the entire family" (robertson, hershenfield, grace, & steward, 2004, p.404) . from our interpersonal-systems perspective, a more general reason to widen the lockdown lens is that individual and family functioning are inextricably interwoven, especially for children and adolescents but also for adults. in fact, an enormous body of research links family conflict and dysfunction to psychological distress, physical health symptoms, and a wide variety of behavior problems (e.g., cummings, koss & davies, 2015; repetti, taylor & seeman 2002) . similarly, cohesive and supportive family processes not only protect individuals from accepted article negative effects of life stress (hobfoll & spielberger, 1992) but also generate a variety of positive outcomes (e.g., conger & conger, 2002; joel wong, uhm & li, 2012) . relevant to the covid-19 pandemic, disruptions of family functioning in the wake of widespread socioeconomic stress such as the great recession of 2007-09 (forbes & krueger, 2019; margerison-zilko et al., 2016) , as well as natural disasters such as floods and earthquakes (cao, jiang, li, lo & li, 2013; mcdermott & cobham, 2012) , have had multiple negative impacts on survivors' behavioral health just as stable and cohesive family relations protect against these. while the quarantine literature emphasizes mainly deleterious effects, it is possible that positive as well as negative repercussions of lockdown could occur at the level of intimate relationships. on the one hand, home confinement can easily create conditions for conflict or estrangement as household members readjust work, school, and recreational activities; face possible contagion and financial strain; and spend virtually all of their time together in limited physical space. on the other hand, such proximity might also create opportunities for increased closeness, communal problem solving, and deeper personal relationships. the title of a may 24th new york times article -"the virus has wrecked some families. it has brought others closer" (wilson, 2020)essentially captures this mixed picture. as couples and families face the demands of a new (crisis) situation, interactional discontinuities may sometimes lead to more resilient as well as deteriorated functioning (patterson, 2002; walsh, 2007) . soon after covid-19 disrupted spain we were able to organize a sizable on-line study of repercussions for individuals, couples, and families. although the sample of convenience did not rigorously represent the spanish population, we hoped to gain preliminary information about the individual and relational wellbeing of people confined together with their partners and/or children during the first three weeks of state-regulated lockdown. in addition to demographic, household, and covid-related employment information, the survey included standardized spanish-language measures of psychological distress and relationship functioning: the former were the state-trait anxiety inventory (stai; spielberger, gorsuch & lushene, 2008; buela-casal, guillen-riquelme & seisdedos-cubero, 2015) and the beck depression inventory (bdi; beck, steer & brown, 1996; sanz & vazquez, 2011 this article is protected by copyright. all rights reserved important qualitative (mixed-method) component of the study is that participants also responded to an open-ended question about perceived changes in couple or family dynamics since the beginning of home confinement. the relationship aspects of the study were of special interest to us as couple and family therapists, and including the cerfb followed naturally from our involvement in a research and development project ("family relational diagnosis in mental health") funded by the spanish government (ministerio de economía, industria y competitividad, 2017). the cerfb attempts to operationalize central constructs in linares' (1996 linares' ( , 2002 linares' ( , 2012 basic family relations theory, where partially orthogonal conjugal functions and parenting functions converge to create optimal (or sub-optimal) conditions for relational nurturing, a crucial determinant of child mental health. according to the theory, combining the bi-polar conjugal (harmony-disharmony) and parenting (preservation-deterioration) dimensions yields four prognostically significant quadrants: functionality (both dimensions high) allows for mature and balanced child development; triangulation (parenting high, conjugal low) facilitates child involvement in couple conflicts; deprivation (conjugal high, parenting low) sustains satisfactory couple relations at the expense of child nurturing; while chaotization (both dimensions low) maximizes conditions for child psychopathology (linares, 2002) . because the cerfb scales apply only with children present, we used the das to assess conjugal relationship quality when parents had no children (couple only) or when children were no longer at home (empty nest). more specific aims of the study were to (a) compare lockdown responses from the pandemic convenience sample to benchmarks for established measures of individual, couple and parental functioning; (b) describe via qualitative analysis the ways in which participants felt their couple and family relationships had improved and/or deteriorated during the first few weeks of lockdown; (c) identify demographic, household, and employment-related correlates of pandemic relationship functioning and psychological distress with special attention to variations across couples with children at home, couples with no children, and couples with empty nests; and (d) explore possible changes in relationship functioning over time during the first three weeks of lockdown. this article is protected by copyright. all rights reserved a total of 407 participants recruited through facebook and other social media platforms completed our online survey between 24 march and 7 april 2020, weeks 2 and 3 of the stateregulated home confinement. inclusion criteria were: (a) aged 18 or higher, (b) currently living in spain, (c) living with one's romantic partner and/or one's children (including divorced parents currently living with children in a shared custody arrangement). the participants were predominantly female (77.0%) and ranged in age from 22 to 77 years (m = 42.7; sd = 12.7). most were also well-educated (76.7% had university degrees) and resided in cities (67.2%) rather than smaller urbanized towns (27.5%) or rural areas (5.1%). the crucial variable of household-parental status governing which relationship measure(s) they would complete distributed as follows: partnered parents living with children (47.4%), partners in couples without children (37.3%), partners in couples whose children were not at home (9.6%), and divorced parents (5.7%). children's ages ranged from 5 months to 51 years, and 49.7% of parents had more than one child at home. a majority of participants (69.9%) were at least partly employed at the time of the survey, with 54% telecommuting, 7.3% working entirely on site, and 8.8% doing both. while only 7.1% had experienced covid-related job loss, 17.0% of the sample was currently unemployed and 5.9% had retired. occupations varied widely, with 33.4% of the sample in some way affiliated with the health professions, 11.3% working in an educational capacity, and 47.2% in general commerce or self-employed. some participants (14.0%) reported that a member of their household was experiencing a health problem at the time of the survey, and 11.8% indicated they were currently receiving psychological or psychiatric treatment for problems such as anxiety, depression, substance abuse, adhd, relationship issues, or wanting "personal growth." the project received ethics approval from the ethics research committee of the school of psychology, education, and sports sciences, blanquerna, ramon llull university (certificate # 1920005p). before beginning the online survey, each participant reviewed information about the study's purpose and procedures, including assurance of confidentiality, and provided her or his informed consent. responses to an initial question about household and parental status distributed participants according to which standardized relationship measures they would complete later in this article is protected by copyright. all rights reserved the survey. while all 407 participants completed the stai and bdi, those in couples with no children or all children away from home completed the das (n=191); partnered parents with a child in the household did the full cerfb (n=193); and divorced parents completed the cerfb parenting and co-parenting scales but not the conjugal function scale (n=23). although a common measure of couple functioning for participants with and without children at home would have been ideal, allocating scales as we did made the online survey more time efficient. fortunately, previous validation research has found high correlations between cerfb conjugal functioning and the das (e.g., r = .74 in ibañez, 2016) . the sequence of survey questions proceeded from demographic, household, and current employment information (including perceived economic risk) to the standardized measures of psychological distress and relationship functioning. instructions throughout the survey reminded participants to focus on the lockdown period in considering their responses. a final, open-ended question eliciting data for qualitative analysis asked, "what changes have you perceived in your couple or family dynamics since the beginning of home confinement (march 14th 2020)?" accompanying this was an apology for not conducting a face-to-face interview and a request to answer in as much written detail as possible: "the more information you provide, the better." we discontinued the survey on 8 april 2020, when state-regulated restrictions first began to ease. the state-trait anxiety inventory (stai; spielberger et al., 2008) , validated for use in spain by buela-casal, et al. (2015) , is a 40-item self-report instrument that assesses anxiety as both a state (20 items) and a trait (20 items). with items in a 0-3 response format, state and trait anxiety scores range from 0-60, and benchmark cut points for adult spanish men and women provide a basis for classifying subscale scores from "very high" to "very low" (buela-casal et al., 2015) . internal consistency coefficients for the current lockdown sample were α = .93 and .84 for state and trait anxiety, respectively. the beck depression inventory (bdi; beck et al., 1996) , validated in spanish by sanz & vázquez (2011), consists of 21 self-report items measuring the presence and severity of this article is protected by copyright. all rights reserved depression. as with the stai, benchmark cut points for the spanish population permit classifying bdi scores as reflecting minimal, mild, moderate or severe depression (sanz & vázquez, 2011) . reliability for the lockdown sample was α = .85. participants with no children at home completed the spanish version of the dyadic adjustment scale (das; spanier, 1976, 2017), a 32-item questionnaire measuring general couple relationship quality. the das also has consensus, satisfaction, affectional expression, and cohesion subscales, but because these were highly intercorrelated we used only the das total score (α = .94) in the main analyses. although clinical cut points for the das are not participants with children at home completed the basic family relations evaluation questionnaire (cerfb; ibáñez et al., 2012; vilaregut et al., 2019) , a 25-item parent-report instrument inspired by linares' (2002 linares' ( , 2012 theoretical ideas about relational nurturing. the original cerfb includes a 14-item parenting function scale (α = .92), measuring the quality of parent-child relations (e.g., "i feel that my children return my affection"), and an 11-item conjugal function scale (α = .91) reflecting the quality of how parents relate to each other as a couple (e.g., "my partner knows how to treat me"). responses are on 5-point likert scales ranging from 1 (never) to 5 (always). to further assess collaboration between the two parents with regard to childrearing, we included 16 additional items from a preliminary co-parenting scale (α = .85) currently undergoing validation (e.g., "we make a good team as parents"). clinical norms for cerfb scales are not yet available but, as with the spanish das, descriptive statistics from validation studies provide tentative benchmarks for evaluating levels of parental and conjugal functioning in the lockdown sample (campreciós, 2015; ibáñez et al., 2012; roca et al., 2020; vilaregut et al., 2019) . interpretation may be complicated, however, because identifying linares' patterns of functional and dysfunctional parenting requires taking both dimensions (and ideally the views of both parents) into account. this article is protected by copyright. all rights reserved given the exploratory nature of the study, we approached the research aims with the basic premise that combining qualitative and quantitative forms of evidence provides a better understanding than either method does by itself (creswell & plano clark, 2011) . indeed, both qualitative and quantitative data figure prominently in the resultsand despite homage to validated quantitative measurement methods (above), our most direct evidence of actual lockdown "effects" on couple and family functioning came from participants' qualitative (written) descriptions of what had changed. in the spirit of mixed-method research, we then used dichotomous variables representing the presence or absence of specific qualitative themes in participants' responses to explore quantitative associations with other study variables. after dropping 23 written responses reporting no couple/family change (e.g., "everything continues as usual") and 53 blank responses, we used braun & clarke's (2006) method of thematic analysis, assisted by atlas.ti software for mac (v. 8), to code 329 descriptions of change. the total qualitative data set consisted of 13,226 words, with individual responses ranging from 2 words (e.g., "closer now") to 353 words. the thematic analysis involved identifying interesting data features, or codes; clustering codes and searching for potential themes; and finally, naming and defining the themes. to facilitate accuracy and trustworthiness, two authors served as co-coders in an ongoing consensual review process, and the full team reviewed emerging results to reach on the final thematic configuration. on the quantitative side (using spss statistics, v. 20), preliminary analyses justified creating two composite variables that would simplify later examination of psychological distress and couple relationship functioning: composite distress was a z-score combination of stai state, stai trait, and bdi scores, which intercorrelated highly with all r s > .66. similarly, the composite measure of couple functioning combined z-transformations of das total scores and cerfb conjugal function scores, which were not available for the same participants here but had correlated highly in previous research. as dependent variables, the two composites helped clarify multivariate and moderated influences on central study constructs. we approached the main study aims by first examining descriptive statistics for psychological distress (stai, bdi) and relationship functioning (das, cerfb) in the lockdown sample with an eye toward areas of possible discrepancy and/or alignment with benchmarks for the broader population. in light of sampling limitations both here and in the standardization studies, however, such comparisons can only be approximate, with conclusions about lockdown this article is protected by copyright. all rights reserved effects necessarily tentative. next, having defined qualitative change themes (as described above) and coded their presence/absence across participant responses, simple tabulations and cross-tabulations illuminated the prevalence of various improvement and deterioration themes in the full sample and across the four main participant groupings: partnered parents with children at home, partners in couples with no children, parents with all children away from home, and divorced parents. finally, we employed a variety of univariate, multivariate, and moderation analyses to identify demographic, household, and employment-related correlates of pandemic relationship functioning and psychological distress, again with attention to variations across household/parental subgroups. in addition to cerfb scores, das scores, and the two composites, these correlational explorations included the qualitative change themes and a rough approximation of the linares cerfb parenting styles (functionality, triangulation, etc.). also of interest were possible changes in relationship functioning over time as the lockdown progressed, including system-symptom links between relationship quality and individual distress. 13.4% of lockdown participants but only 4.6% were in the moderate or severe range. when asked about their perception of economic risk, just over half of the respondents were at least moderately concerned with economic risk during the pandemic (53.5%) but less than a quarter were very concerned (15.7%) or extremely concerned (6.6%). in general, the lockdown experience appeared to generate moderate to high levels of situational anxiety and uncertainty but not much chronic distress among the adults who completed the on-line survey. this article is protected by copyright. all rights reserved cerfb norms provide no firm basis for locating parenting styles in the linares' quadrants, a rough comparison of lockdown means with standardization data suggests that at least half of the reports in our sample would be most consistent with the triangulation style, where parents tend to maintain harmony by involving children in their conflicts. such classification is approximate at best, however, especially without reports from both parents or confirmation from a child. to summarize, with no children in the household, the quality of couple relationships during the covid-19 lockdown appeared no worse and possibly better than would have been the case without lockdown. with children present, however, our data raise the possibility that preservation of family harmony may have sometimes occurred at the expense of relational nurturing. two overarching thematic categoriesperceived improvement and perceived deteriorationemerged from our qualitative analysis of participants' free-form descriptions of how family and couple dynamics had changed during the lockdown. table 2 shows the prevalence of these themes in the subsample of 329 respondents who indicated that some degree of lockdown-related family change had occurred. (note that a given response could include more than one theme; in fact, 24 cited both improvement and deterioration.) interestingly, the overall prevalence of improvement themes (61.7%) exceeded the prevalence of deterioration themes (41.0%), with increased connection/cohesion (44.7%) and conflict (21.9%) cited most often. relative to the das and cerfb data above, this appears more consistent with the dyadic adjustment of couples having no children at home than with the cerfb parenting picture. table 2 also indicates (via chi-square comparisons) that some themes distributed less evenly than others across subgroups defined by household composition and parental status. for example, the theme of family (re)connection was more prevalent for parents with no children and those with children at home than for parents who were divorced or had children not at home. the most striking differences, however, involved deterioration themes such as couple/family distance (most common with children away and negligible with children at home), conflict atmosphere (less frequent with emancipated children) and unbalanced needs (most likely with children with children at home). overall, participants tended to use more words when describing deterioration themes than improvement themes, with unbalanced needs (r = .24, p < .01), negative expectations (r = .21, p < .001), and conflict atmosphere (r =.14, p = .01) entailing the thickest descriptions. on the improvement side, longer responses were associated with balanced needs (r = .26, p < .001) and teamwork spirit (r = .16, p = .003). as noted above, we created two composite dependent variables representing psychological distress and couple functioning to facilitate identifying correlates of individual and relational wellbeing during lockdown. the distress composite is available for the full sample while the latter includes all participants except 19 who had divorced. the couple composite is useful because the survey entailed different (though highly correlated) measures of couple functioning depending on whether or not the respondent had a child at home. thus, although it was not possible to compare these subgroups directly, the z-score composite permits examining differential (moderated) prediction of conjugal functioning. this article is protected by copyright. all rights reserved table 3 provides an overview of predictor variables associated with various measures of individual, couple, and parental functioning. point bi-serial correlations in the first few rows, where dichotomous dummy variables represent household/parental subgroups, reveal few group differences in composite individual or couple functioning apart from the small group of divorced parents and parents with a preschool-age child reporting more psychological distress. a strikingly different picture of couple functioning emerges in relation to pandemicrelated employment, where stronger associations appear when respondents have children at home (cerfb conjugal function) compared to when they do not (das total score). of particular interest is how couple adjustment relates to telecommuting and (un)employment. general linear model (glm) moderation analyses examining these association across three groups of couples (no children, child at home, empty nest) found significant group x telecommute (f = 5.23, p = .006) and group x employed (f = 4.44, p = .012) interaction terms, with group means suggesting that telecommuting and employment were positively related to couple functioning when respondents were parents (regardless of whether children were at home or emancipated) but in the opposite direction when they were not. table 3 also highlights correlates of individual distress and relationship functioning that did not vary appreciably (interact) with household/parental status: psychological distress was generally higher among women, recipients of psychological treatment, unemployed respondents, those in business professions or perceiving economic risk, and those with a health problem at homebut lower among health professionals. among the cerfb family relationship measures, parental functioning showed significant correlations with some of the same predictors and was positively related to education; and if anything, marital and co-parental functioning was better among younger couples. the last two rows in the top panel of table 3 show potentially important associations involving lockdown duration, couple functioning, and psychological distress. first, marital functioning for couples with children at home systematically improved with days in lockdown, which was not the case for parenting functions, psychological distress, or for couples without children at home. this cross-sectional temporal trend remained after statistically controlling for parallel changes over days in other demographics such as urban location, employment status, or having a preschool child at home. note also that effective marital functioning with a child at home correlated positively with telecommuting and having a job but not with psychological this article is protected by copyright. all rights reserved distress. in this way, couple functioning in the context of active parenting during lockdown was unique among the relationship measures. the strong associations between relationship quality and individual distress in table 3 are consistent with a large couple and family relations literature, but we do not know if they are in any way unique to quarantine. an indirect way to approach this is to ask whether the linkage between relationship quality and individual functioning systematically changed over the course of quarantine, as would be indicated by statistical interactions involving lockdown days. glm analyses along these lines, with psychological distress as the dependent variable, did in fact show suggestive relationship-quality x lockdown-duration interaction terms for das total dyadic adjustment (f = 5.37, p = .022) and cerfb co-parenting (f = 4.50, p = .035) in the direction of system-symptom linkages strengthening as the lockdown progressed. the lockdown study period was relatively short, however, so this intriguing evidence is tentative at best. finally, because the cerfb results in table 3 do not speak directly to the linares model of relational nurturing that guided development of the instrument, we were interested in how parenting styles representing functionality, triangulation, deprivation and chaotization might have operated during the lockdown. this seemed important because the apparent parentalconjugal discrepancy in our sample suggests that some degree of triangulation may have been in play. although the cerfb standardization data offers no cut points for defining linares quadrants with any precision, we attempted to approximate the quadrants in a relative manner by splitting the conjugal and parental distributions at their medians. interestingly, participants in the high-parenting/low-conjugal (triangulation) quadrant reported less distress and fewer deterioration themes than other participants (supplementary table s1 , online). triangulation represented in this manner was also more common among university graduates and health professionals. because the most direct evidence of lockdown effects on family relationships comes from participants' free-form reports of improvement and deterioration, we were interested in which qualitative themes were most and least likely to come from which participants. the bottom panel of table 3 shows strong associations between perceptions of relational improvement/deterioration and some but not all of the standardized individual, couple, and parental functioning measures. in particular, the qualitative reports aligned most solidly with this article is protected by copyright. all rights reserved psychological distress and with the dyadic adjustment of partners with no children in the household. with children at home there were no significant correlations for cerfb conjugal functioning and only marginal ones for parenting and co-parenting. of the specific qualitative themes, family (re)connection dominated criterion connections on the improvement side while conflict and distance did so on the deterioration side. table s2 , online) indicate that respondents reporting improvement were more often employed and had no emancipated children or preschoolers at home. deterioration themes, on the other hand, were associated with receiving psychological treatment, the presence of preschoolers, and coping with health problems in the household. as for markers of specific qualitative theme categories, improvements in (re)connection were more prevalent among health professionals and respondents with no preschoolers; education and employment predicted improved communication; expressiveness themes correlated with telecommuting and low perceived economic risk; city dwellers were high on teamwork; younger participants described more conflict and fewer distance themes; females reported more conflict; health professionals described fewer negative expectations; an empty nest predicted more distance and less conflict; health problems in the household portended distance themes; and having children at home meant less distance and more unbalanced family member needs as noted above. this mixed-method exploration of the recent covid-19 lockdown in spain adds depth and complexity to the quarantine literature by highlighting repercussions for relationships as well as individuals in the direction of improved as well as compromised functioning. although survey respondents experienced moderate to high levels of situational anxiety during the first three weeks of state-regulated lockdown, they also reported high dyadic (couple) adjustment relative to a norm group and cited more instances in which their couple and family dynamics had improved rather than deteriorated. this contrasts with previous studies that have emphasized predominantly negative psychological effects of mass quarantine on individuals (brooks et al., 2020) , and occasionally also on couple and family relations. the results also suggest that quarantine ramifications for couples and families are more complex than those for individual children and adults. one aspect of this complexity is that positive as well as negative forms of change appear more likely at the level of close relationships this article is protected by copyright. all rights reserved (with opportunities for collective coping, interpersonal need satisfaction, etc.) than at the level of individual adjustment. positive individual change (e.g., post-traumatic growth) is certainly possible, of course, but the quarantine literature has not emphasized this. another indication of greater complexity is that measures of couple and parental functioning varied with household composition and parental status in ways that measures of individual wellbeing (psychological distress) did not. for example, conjugal relations during lockdown appeared more harmonious when there were no children in the household, and moderation analyses indicated that covidrelated employment variables (e.g., telecommuting) predicted successful couple functioning in different ways depending on parental status. in general, the relationship implications of lockdown were clearer for participants with no children at home than for those actively engaged in parenting. for the former, dyadic adjustment scores were higher than those from a benchmark comparison group and correlated highly with both the absence of psychological distress and the presence of improvement themes in participants' descriptions of how relationship dynamics had changed. for active parents, on the other hand, benchmark cerfb comparisons were more ambiguous, with parenting quality tending to exceed conjugal quality, and qualitative improvement/deterioration themes did not map so clearly onto relationship functioning. although tentative, the apparent prevalence in our sample of a parenting style linares (2002) associates with triangulation raises the possibility of a subtle downside to family relations during lockdown. if greater appreciation of parent-child than parent-parent (conjugal) bonds does in fact invite triangulation of children into adult conflicts, this could undermine child development in ways that would not be apparent without direct assessment of child functioning. the fact that provisional triangulation correlated with less psychological distress and fewer family conflict themes seems also to underscore the subtlety of this ultimately dysfunctional pattern. as noted earlier, however, the cerfb norms as yet provide no firm basis for identifying linares parenting styles, so without converging reliability and validity evidence we must offer this interpretation cautiously. if nothing else, the possibility of increased risk for triangulation during lockdown provides a hypothesis for future research. what might it mean that participants without children at home more often reported increased couple cohesion during lockdown? from an interpersonal-systems perspective, the relative absence of third-party involvements with children, relatives, friends, or colleagues, this article is protected by copyright. all rights reserved coupled with sustained proximity and more time for shared conjugal activities, provides a plausible explanation. a more psychological explanation, from the author of a recent survey of u.s. couples conducted around the same time as ours, is that people simply want more closeness in their important relationships and turn to partners for support under covid-19 stressso in effect they are getting what they want (g. lewandowski, as cited in bonos, 2020) . the current data are cross-sectional and can only indirectly address any changes that may have occurred over time during the relatively brief (two-week) study period. such change was of interest because at least one prior study found increased symptoms of psychological distress among individuals quarantined at least 10 days (hawryluck et al., 2004) . this was not the case in the present study, where stai, bdi and composite measures of individual distress did not changeyet the cerfb measure of conjugal functioning actually showed an opposite trend of improvement over lockdown days. in other words, participants challenged by managing children at home, whose conjugal relations were probably not as good on average as those of other participants, nonetheless tended to report more couple-level resilience in the third week of lockdown than in the second. our rough indicator of triangulation decreased over days as well, implying the possibility of positive spillover for children. intriguingly, moderation results also suggest that the correlation or linkage between individual distress and couple relationship functioning may have strengthened as the lockdown progressed. the direction of influence in this would be ambiguous, however. in addition to supplementing quantitative analyses, the qualitative themes add texture and detail to how participants experienced relational improvement and deterioration during lockdown. on the deterioration side, some of the themes echo aspects of previously identified stressors linked to quarantine at the individual level. for example, experiencing loneliness and couple/family distance connects to boredom and isolation (cava et al., 2005; digiovanni et al., 2004) ; and negative expectations incorporates fear of the disease (bai et al., 2004; desclaux et al., 2017) , economic uncertainty (jeong et al., 2016; mihashi et al., 2009) , and information overload (blendon et al., 2004) . although a few participants cited instances of both improvement and deterioration, most followed one of the two paths in ways that sometimes resonated with our clinical experiences during the first few months of the pandemic. as if to illustrate the improvement pathway, a stayat-home mother who had come to therapy distressed by a disobedient, disrespectful and this article is protected by copyright. all rights reserved occasionally violent child reported that she, her husband, and the difficult son had been having "so much fun together" (for a change) since the lockdown began, adding that "while we were playing, several picture frames fell of the wall and broke without even bothering me!" more often, however, we encountered deterioration themes in the form of a lockdown family crisis. here a useful strategy has been to contextualize the current problem situation by exploring how family members had (successfully) handled similar situations or had more satisfying relations before the lockdown began. the qualitative improvement themes suggest possible areas to explore. our study has many limitations, with sample representativeness foremost among them. the self-selected sample of convenience was predominantly female, well educated, employed, interested in the topic at hand (family relationships), and probably more functional and economically advantaged than the adult spanish population at large. it is entirely possible that different results (e.g., more evidence of individual distress or relational deterioration) would obtain in other sectors of the population or in other countries or cultures. another limitation is that time-limited, cross-sectional survey data shed little light on enduring effects of quarantine, on how adaptations to lockdown changed or evolved over time, or on what happened during reopening, when home-confinement restrictions began to ease. we plan to address these and other questions through follow-up interviews with study participants. finally, it is difficult to obtain reliable and valid representations of family functioning from the self-reports of only one family member, and shared method variance (e.g., positive or negative response set) could account for some of the quantitative association between individual and couple/family variables. despite these limitations, our exploratory study of the recent state-regulated covid-19 lockdown in spain helps to break new ground by looking beyond the individual to understand psychosocial impacts of quarantine experiences. the ramifications of these experiences for couples and families are clearly important and probably more complex than for individuals. notes. table entries are pearson or point-biserial correlation coefficients reflecting associations between predictor variables and measures of individual, couple and parental functioning. left-most column includes prevalence proportions for dichotomous predictors and qualitative themes. a z-score composite of stai state and trait anxiety measures and bdi depression available for all participants. b z-score composite of total dyadic adjustment (das) and cerfb marital functioning scores, excluding divorced parents. c dyadic adjustment scale (das) total score. d family relations (cerfb) scores for parents with at least on child at home. two-tailed significance levels: *** p < .001, ** p < .01, * p < .05, † p < .10. survey of stress reactions among health care workers involved with the sars outbreak bdi-ii. beck depression inventory-second edition. manual the public's response to severe acute respiratory syndrome in toronto and the united states real decreto 463/2020, de 14 de marzo, por el que se declara el estado de alarma para la gestión de la situación de crisis sanitaria ocasionada por el covid-19 our relationships are relationships are actually doing well during the pandemic, study finds using thematic analysis in psychology the psychological impact of quarantine and how to reduce it: rapid review of the evidence cuestionario de ansiedad estado-rasgo validación y aplicabilidad clínica del cuestionario de evaluación de las relaciones familiares básicas (cerfb) en los trastornos de la conducta alimentaria (doctoral dissertation) family functioning and its predictors among disasterbereaved individuals in china: eighteen months after thewenchuan earthquake. plos one, e60738 the experience of quarantine for individuals affected by sars in toronto resilience in midwestern families: selected findings from the first decade of a prospective, longitudinal study designing and conducting mixed methods research prospective relations between family conflict and adolescent maladjustment: security in the family system as a mediating process accepted monitoring or endured quarantine? ebola contacts' perceptions in senegal factors influencing compliance with quarantine in toronto during the 2003 sars outbreak the great recession and mental health in the united states a third of americans now show signs of clinical anxiety or depression, census bureau finds amid coronavirus pandemic. the washington post chinese city experiencing a divorce peak as a repercussion of covid-19. global times sars control and psychological effects of quarantine accepted article this article is protected by copyright. all rights reserved family stress: integrating theory and measurement validación del cuestionario de evaluación de las relaciones familiares básicas (doctoral dissertation) propiedades psicométricas del cuestionario de evaluación de las relaciones familiares básicas (cerfb) mental health status of people isolated due to middle east respiratory syndrome asian americans' family cohesion and suicide ideation: moderating and mediating effects identidad y narrativa. la terapia familiar en la práctica clínica del abuso y otros desmanes. el maltrato familiar, entre la terapia y el control terapia familiar ultramoderna. la inteligencia terapéutica health impacts of the great recession: a critical review the relevance of psychosocial variables and working conditions in predicting nurses' coping strategies during the sars crisis: an online questionnaire survey family functioning in the aftermath of a natural disaster accepted article this article is protected by copyright. all rights reserved predictive factors of psychological disorder development during recovery following sars outbreak propuesta de resolución provisional del procedimiento de concesión de ayudas a proyectos de i+d+i construcción del cuestionario de evaluación de relaciones familiares básicas y coparentalidad (cerfb-cop). poster session presented at xix jornades anpir understanding family resilience social consequences of ebola containment measures in liberia a nationwide survey of psychological distress among chinese people in the covid-19 epidemic: implications and policy recommendations risky families: family social environments and the mental and physical health of offspring the authors appreciate the assistance of dr. michael rohrbaugh in preparing the manuscript for publication. key: cord-306227-63qvvkvk authors: shammi, mashura; bodrud-doza, md.; islam, abu reza md. towfiqul; rahman, md. mostafizur title: strategic assessment of covid-19 pandemic in bangladesh: comparative lockdown scenario analysis, public perception, and management for sustainability date: 2020-07-18 journal: environ dev sustain doi: 10.1007/s10668-020-00867-y sha: doc_id: 306227 cord_uid: 63qvvkvk abstract: community transmission of covid-19 is happening in bangladesh—the country which did not have a noteworthy health policy and legislative structures to combat a pandemic like covid-19. early strategic planning and groundwork for evolving and established challenges are crucial to assemble resources and react in an appropriate timely manner. this article, therefore, focuses on the public perception of comparative lockdown scenario analysis and how they may affect the sustainable development goals (sdgs) and the strategic management regime of covid-19 pandemic in bangladesh socio-economically as well as the implications of the withdrawal of partial lockdown plan. scenario-based public perceptions were collected via a purposive sampling survey method through a questionnaire. datasets were analysed through a set of statistical techniques including classical test theory, principal component analysis, hierarchical cluster analysis, pearson’s correlation matrix and linear regression analysis. there were good associations among the lockdown scenarios and response strategies to be formulated. scenario 1 describes how the death and infection rate will increase if the bangladesh government withdraws the existing partial lockdown. scenario 2 outlines that limited people’s movement will enable low-level community transmission of covid-19 with the infection and death rate will increase slowly (r = 0.540, p < 0.01). moreover, there will be less supply of necessities of daily use with a price hike (r = 0.680, p < 0.01). in scenario 3, full lockdown will reduce community transmission and death from covid-19 (r = 0.545, p < 0.01). however, along with the other problems gender discrimination and gender-based violence will increase rapidly (r = 0.661, p < 0.01). due to full lockdown, the formal and informal business, economy, and education sector will be hampered severely (r = 0.695). subsequently, there was a strong association between the loss of livelihood and the unemployment rate which will increase due to business shutdown (p < 0.01). this will lead to the severe sufferings of poor and vulnerable communities in both urban and rural areas (p < 0.01). all these will further aggravate the humanitarian needs of the most vulnerable groups in the country in the coming months to be followed which will undoubtedly affect the bangladesh targets to achieve the sdgs of 2030 and other development plans that need to be adjusted. from our analysis, it was apparent that maintaining partial lockdown with business and economic activities with social distancing and public health guidelines is the best strategy to maintain. however, as the government withdrew the partial lockdown, inclusive and transparent risk communication towards the public should be followed. recovery and strengthening of the health sector, economy, industry, agriculture, and food security should be focused on under the “new normal standard of life” following health guidelines and social distancing. proper response plans and strategic management are necessary for the sustainability of the nation. graphic abstract: [image: see text] electronic supplementary material: the online version of this article (10.1007/s10668-020-00867-y) contains supplementary material, which is available to authorized users. performed is 308,940 which is 1877 tests/1 million (iecdr 2020; worldometer 2020) . the number is one of the lowest in the world (table 1) . it is predictable that during a pandemic, a humanitarian crisis may arise in a developing country like bangladesh. in most incidents, it will be the combined effects of a variety of shortages that will likely culminate in the worst outcomes (truog et al. ). this can lead to a shortage of basic needs including foods, goods, and services such as job loss, economic and financial loss, food insecurity, famine, social conflicts, and deaths. besides, an impact on the psychosocial and socio-economic and health and well-being of the citizens may be affected which we had shown in our previous study . while predicting all the subsequent impact of the covid-19 pandemic is challenging, early strategic planning and groundwork for the evolving and established challenges will be crucial to assemble resources and react in an appropriate timely manner. moreover, as the gob has lifted the lockdown without flattening the curve of the pandemic what should be the socio-economic management strategy for the government at this stage. this article, therefore, focuses on the public perception of comparative lockdown scenario analysis and the strategic management regime of covid-19 pandemic in bangladesh. as there is no such prediction on how long the situation prevails, the absence/lack of management strategy for an epidemiological and socio-economic emergency response might be a tool to assess the forthcoming situation under a set of specific scenarios. therefore, the objective of this study is to analyse long-term strategic management of the pandemic in three different lengths of scenarios in a resource-limited setting of the so-called lockdown of the country. the outcome can play a crucial role to formulate emergency response strategy to tackle the covid-19 considering the impact of current socio-economic lockdown for flattening the curve of covid-19 infection in bangladesh, this study identifies three scenarios of lockdown based on literature review, the lockdown situation in bangladesh, and the global lockdown practices in different countries. a total of 54 statements were used to develop the questionnaire to understand the scenario-based impact assessment and management of covid-19 outbreak in bangladesh. we had taken expert suggestions, consultations, and cross-validation of 54 statements for getting perceptions from a different group of people. we have used google form to prepare the questionnaire and conduct the survey online. an online database of target bangladeshi participants was prepared by reviewing different online social platforms of different expert groups in bangladesh, considering their current activities, responsibilities, and engagement related to covid-19 response in socio-economic sectors, planning, and policymaking. the prepared questionnaire with an introductory paragraph outlining the objective of the study was then shared through email, facebook, messenger, linkedin, and whatsapp with selective and relevant people considering the purposive sampling method. the questionnaire survey was conducted from 09 april to 11 april 2020. a five-point (1-5) scenario-based likert scale was employed to test whether each respondent understands the statements described ranging from strongly disagree to strongly agree. the target population was general bangladeshi citizens age 18 years and older. the inclusion of the respondents was different social groups like university faculty and scholars, government officials, non-government officials, development workers or practitioners, doctors, engineers and technologists, youth leaders and students, businessmen and industry officials, banking and finance corporates, researchers, and others. the answers to the survey questionnaires were the voluntary basis. a total of 159 responses were recorded during the survey. according to the survey findings, the ratio of male to female participants was 3 (n = 113):1 (n = 46). the composition of age groups was 44% (18-25 years old), 41% (26-35 years old), and 15% (36-55 years old), respectively. however, the average age of the respondents (n = 159) was 28.44 years (sd ± 6.51). 40% of the youth leader was mostly students as they are the dynamic group in the society, also involved in covid-19 response, volunteer social works, job seeking, research and reporting activities. rest of the 60% participants belonged to various professions of doctors, civil service officials, development practitioners, non-government officials, expert panels, and university scholars. the descriptive statistics (e.g. frequencies, percentages, and student t test) were used to understand the participant's characteristics. applying the statistical package for the social science (spss) v. 25.0, datasets were analysed via a set of statistical tools such as principal component analysis (pca), hierarchical cluster analysis (hca), pearson's correlation coefficient (pcc), multiple regression analysis (mra), and classical test theory (ctt) analysis. pca is one of the population data reduction techniques that indicate each potentiality of variables and their significance level in a huge sample size. before conducting the pca, kaiser-maier-olkin (kmo) and bartlett's sphericity tests were applied to confirm the necessity of this analysis. the results of the kmo > 0.5 (the kmo value was 0.8 in this research) and the significance of bartlett's sphericity test at p < 0.01 verified our datasets to be fitted for the pca ). the number of factors chosen was based on the kaiser's normalization principle, where the only factors with eigenvalues > 1.0 were regarded. from ctt analysis, cronbach's alpha was employed to test the consistency and reliability of the factor loadings in this study. cronbach's alpha validation values ranged from 0.925 to 0.934 (the cronbach's alpha reliability value was 0.925 indicating that these statements are appropriate in social science study (table 2 ) (devellis 1991). the hierarchical cluster analysis (hca) is a crucial means of identifying relationships among all socioenvironmental variables. the hca assists to classify a population into different groups based on the similar characteristics of a set of the dataset that may show causes, effects, and or the source of any undetected socio-environmental crisis. furthermore, hierarchical clustering was adopted to determine the possible number of clusters. the one-way anova test was conducted to confirm the significant differences in the variance at p < 0.05. participants were informed of the specific purpose of the study. participants' consent was taken before the survey and they remained anonymous. the survey was completed only once, and the survey could be completed/terminated whenever they wished. the survey content and procedure were reviewed and approved by the department of public health and informatics, jahangirnagar university. reductions of covid-19 pandemic have been linked with the cessation of public transport, educational institutions, the closing of entertainment and business centres, and the prohibition of public meetings. averagely, cities that carried out control measures of lockdown within the first week of the outbreak reported fewer cases compared to the cities which started control later (tian et al. 2020) . vietnam, the communist country bordering china with a population of 95 million, has been an under-reported low-cost success story of the pandemic, which has had just 268 coronavirus cases and no deaths due to strong public health systems, good governance and transparent communication, strict quarantine approach, and contact tracing. from the first-known international cases on 23 january 2020, vietnam quarantined anyone who arrived from the high-risk area and closed its porous 1400-km border with china to all but essential trade and travel. in february, it quarantined more than 10,000 people in the son loi village due to infections. it also closed all schools, colleges, universities, and all other educational institutions beginning in february. on march 22, the vietnamese government suspended all foreign entries (wef 2020a). it also decided early on to impose a 14-day quarantine on anyone arriving in vietnam from a high-risk area. on april 22, vietnam eased its lockdown effort. unfortunately, in the european region such as worst-affected italy, france, and uk, the region in the americas such as the usa and now brazil experienced most deaths due to the failure to understand the disease and keep continuing their activities such as travelling and tourism which spiked the death rates, or delayed lockdown efforts (table 3 ). the case of new zealand is interesting in the sense that it pursued an eradication tactic historically based on a mitigation model and focused on slower arrival of the virus, followed by a range of measures to flatten the curve of cases and deaths (cousins 2020) . the measures include increased testing, screening, strict quarantine of the infected person, contact tracing, and surveillance. the same measures have been adopted by south korea, taiwan, and many successful countries. luo (2020) , of singapore university of technology and design (http://www.sutd.edu. sg) forecasted using a data-driven model that by 30 may 2020, 99% of the infection case will end, while 100% will end by 15 july 2020 for bangladesh. however, the forecast does not seem to be valid considering the present context in bangladesh. at this background, a scenario-based assessment under different assumptive situation considering the socio-economic and cultural attitude of the country could better identify the impacts. the later section of the article mainly focused on the different scenario and their possible management and their way forwards. in our study, we have considered global lockdown practice, country situation analysis, and expert suggestions to set three scenarios for impact analysis and possible management strategies which are: lockdown scenario 1: withdraw the existing partial lockdown (ls1) lockdown scenario 2: continue the existing partial lockdown (ls2) lockdown scenario 3: full lockdown/shutdown with an exit plan (ls3) a total of 54 scenario-based statements were set, and perception-based statistical analysis was performed. the descriptive statistics are presented in table 4 , which exhibits the percentage, mean, and standard deviation (sd), describing the responses of participants to the related scenarios for all 54 statements from each of the 159 respondents and the direction of each statement in the studied survey datasets. sect. 3.2-3.4 discusses the impacts of different lockdown scenarios. considering the withdrawal of existing partial lockdown, the results of ctt analysis and the corrected inter-item correlation analysis reveal that eight statements have low corrected item-total correlation values (< 0.30). this includes, people will start moving towards table 4 (continued) table 4 (continued) regular life (ls1s1: 0.257); massive movement and a mass gathering of people will be started again (ls1s3: 0.29); community transmission of covid-19 will increase due to people's movement and mass gathering (ls1s4: 0.298); and the number of infected populations will increase (ls1s5: 0.271). bangladesh is entering into this scenario without having any signs of flattening the infection curve. from pearson correlation analysis, a strong significant positive relationship was observed between people will start moving towards regular life (ls1s1) and the formal and informal economic activities will be started (ls1s2) (r = 0.671, p < 0.01) ( table 5) . linear regression analysis reveals that the community transmission of covid-19 will increase due to people's movement and mass gathering (ls1s4) as people will start moving towards regular life (ls1s1, p < 0.01), and crime will rise and more people will die (ls1s15, p < 0.01) exhibited a statistically significant high correlation with the number of infected populations will increase (ls1s5) ( table 6 ). in addition, more unemployment and loss of livelihood (ls1s13) and more people will die (ls1s15) statistically pose a significant positive impact on an irreversible loss to the economy (p < 0.01) of bangladesh. from ctt analysis, continue the existing partial lockdown (ls1m1: 0.276); limited people's movement will enable low-level community transmission of covid-19 (ls2s1: 0.281); infection and death rate will increase slowly (ls2s3: 0.214); increased facilities to the healthcare system for covid-19 treatment will be able to provide health services to the infected peoples (ls3s2: 0.202). from pearson correlation (table 5) , a significant positive relationship was found between the limited people's movement will enable low-level community transmission of covid-19 (ls2s1) with the infection, and the death rate will increase slowly (ls2s3) (r = 0.540, p < 0.01). also, there will be less supply of basic products for daily use (ls2s7) which posed a significant relationship with the price of most of the basic products will be higher than usual (ls2s8) (r = 0.680, p < 0.01). from the linear regression model (table 6) , the association between dependent statements limited people's movement will enable low-level community transmission of covid-19 (ls2s1, r = 0.599, p < 0.01) with poor people will suffer from food and the nutritional deficiency (ls2s9), and gender-based violence will increase (ls2s10). based on management scenario 2, massive awareness and enforcement of proper lockdown and quarantine initiatives were strongly associated with limited people movement will enable low-level community transmission of covid-19 (ls2s1, p < 0.05). from the ctt analysis, among 54 statements, the corrected inter-item correlation analysis showed that only one statement has low corrected item-total correlation values (< 0.30). this adds existing with increased facilities for covid-19 in the health system will be able to provide health services to the infected people (ls3s2: 0.255). the highest interitem correlated value is the loss of livelihood and the unemployment rate will increase due to business shutdown (ls3s5: 0.581), while the lowest value is the number of infections and death will be limited (ls3s3: 0.302). in the case of management of scenario 3, inter-item correlated values are more than 0.478. the high inter-item correlation was observed in the synergy with government, law enforcement agencies, and private sector initiatives (ls3m3: 0.555) and long-term planning and implementation of policies regarding covid-19, psychosocial, and socio-economic loss (ls3m6: 0.48). according to the results of the pearson correlation, there was a statistically significant correlation among scenario 3 where gender discrimination will increase due to covid-19 outbreak with gender-based violence will increase rapidly (r = 0.661, p < 0.01). besides, extremely limited people's movement will reduce the risk of community transmission of covid-19 with the number of infection and death will be limited (r = 0.545, p < 0.01). for management purposes, synergy with government, law enforcement agencies, and private sector initiatives with coordinated emergency relief support (r = 0.632, p < 0.01). also, microfinance support to small and medium enterprises is required for recovery (ls3m4, p < 0.05). for management strategies of scenario 1, deep analysis of the situation should be carried out and go for full lockdown with relief support to the poor and most vulnerable are urgently needed for decision-making in the county due to the rapid community transmission of covid-19 (p < 0.01). first of all, the government should come up with a comprehensive strategic plan accompanied by non-governmental and social organizations and law enforcement to analyse the spread of the virus, identifying the most vulnerable hosts, properly tracked the movement of general people, precise estimation of economic losses from different financial and industrial sectors, educational diminutions and professional and informal employment disruption to picture an integrated scenario of the current situation and future predictions by which the revival of the negative aspects of the country could be managed. there must be two types of the strategic plan on under the category of the emergency response plan (short-term) by ensuring basic supplies to all citizens who are in real needs, motivate and/or force the people to abide by the covid-19 guidelines by the gob and who, prepare a complete but robust list of vulnerable population in terms of covid-19 spreading, co-morbidities, and economic stress, activate all the local wings of the gob such as local government representatives at the village level, and construct a covid-19 response task force to monitor and handle the country situation through application of information and communication technologies (ict). the government should implement those plans with proper timing, transparency, and resources. the gob has already been taking a lot of initiatives to tackle covid-19 pandemic, but there seems lacking proper risk assessment and weak coordination among stakeholders from medical to social welfare. another plan must be focused on the reconstruction or rebuild (long-term) and must follow the guidelines of the sendai framework. the sendai framework for disaster risk reduction 2015-2030 recognizes health at the heart of disaster risk management (drm) at the global policy level (wright et al. 2020 ). this sendai framework has given the rise of the health-emergency disaster risk management (health-edrm) framework an umbrella term used by who (2019). health-edrm thus refers to the "systematic analysis and management of health risks, posed by emergencies and disasters, through a combination of (1) hazard and vulnerability reduction to prevent and mitigate risks, (2) preparedness, (3) response and (4) recovery measures" (djalante et al. 2020 ). this also includes build back the healthcare sector, industrial sector, education, agriculture, research, environment, and finance. however, deep research complied with massive surveillance could help in making decisions whether the lockdown must be further carried on or not and this must have to be based on evidence. miscommunication and miscalculation of the strategy may trigger worsen the situation. communicating the disease risk in the local language is also necessary to increase awareness about the diseases. moreover, in sects. (3.5.1-3.5.3) we have analysed emergency management issues including short to medium-term measures as well as long-term management strategies of covid-19 pandemic lockdown scenarios in bangladesh based on our research outcomes. "lockdown" is an unfamiliar word or term to the people of bangladesh. according to scenario 1, a partial lockdown is a hoax. people recommended to use a more familiar term "curfew" (legal section 144) to maintain strict and there is no alternative to reduce covid-19 transmission. in bangladesh, section 144 of the penal code 1860 prohibits assembly of five or more people, holding of public meetings, and carrying of firearms and this law can be invoked for up to two months (minlaw/gob 2019). this law could have been a much more effective strategy to contain the infection. in total, 78.6% of the participants agreed that community transmission of covid-19 will increase due to the people's movement and mass gathering, 57.9% agreed to continue the existing partial lockdown, whereas approximately 73% of respondents agreed that deep analysis of the situation is required and go for full lockdown with the relief support to the poor and the most vulnerable. overall, the participants had a positive view about lockdown scenario 1 to possibly spread out of covid-19 at the community level. many people expressed their disappointment towards the extreme corruption of the healthcare sector and that it has collapsed before the covid-19 pandemic. respondents advocated the government to consider biomedical waste management for reducing further environmental transmission and that efficient incinerator to be built for hospital waste management. however, the responders also suggested the government to sustain the retail and wholesale kitchen market/bazaar of any area maintaining the health guideline and social distancing. this approach could have positive feedback as already experience in different upazilas in bangladesh with the help of local administrative authorities, magistrates, and police forces. after the 30 days of the partial lockdown, the federation of bangladesh chambers of commerce and industries recommended the opening of the industrial sectors with some guidelines (fbcci 2020). moreover, the fbcci taskforce demanded the covid-19 incentive financial package in a more gettable way from the gob. it could be a very crucial decision to be taken considering the covid-19 contagions and the business development to protect the exports. to maintain livelihood, industrial workers resumed their work from 25 april 2020. however, the gob weakened the lockdown and resumed the industrial activities without proper guidelines or the scientific basis for such a risky decision. the question is why the gob was in hurry to weakening the lockdown and withdraw it without eradicating the disease? predictably, there might be a strong business/financial association to withdraw the lockdown when life and livelihood matters for the poor and middle-class people and to run the country's economy. although gob provided healthcare guidelines and social distancing during work, the infection rates surged significantly among the workers in the industrial zones. most of the covid-19 clusters are majorly distributed in dhaka city, chittagong city, narayanganj, cumilla, gazipur, and the peripheral cities (iedcr 2020; tbs news 2020b). finally, this study confirms that the withdrawal of the partial lockdown will not become positive in terms of covid-19 management in bangladesh, because still, we do not have enough evidence even after the 65 days of lockdown that the transmission is reducing from the peak. overall, the participants had a positive view about lockdown scenario 2 to stop/slow down the spreading out of covid-19 pandemic in bangladesh. in total, 52.2% of respondents in this study agreed that existing health facilities will not be able to provide adequate services to the number of covid-19 patients due to limited community transmission, while 67.9% strongly agreed that there will be a need for emergency food and financial support to the poor communities. about 72.5% strongly agreed that emergency relief to the poor communities in both urban and rural areas should be provided ensuring transparency. around 34 million people, or 20.5% of the population, live below the poverty line and based on the current rate of poverty reduction, bangladesh is projected to eliminate extreme poverty by 2021 (chaudhury 2018 ). yet, as covid-19 pandemic hit the country within 2 weeks poverty rate in bangladesh rose to 40.9% as 25% of family incomes fell (the financial express 2020a). so, it was the choice between life versus livelihood (hussain 2020) . the poor community always lacks food and nutrition due to the injustice and corruption by the local or regional level of political stakeholders in bangladesh. by nature, people of bangladesh are quite unaware and kind of ignorant or does not like to abide by rules. moreover, the public is not confident somehow with the administrative decisions, policies, and their implementation of covid-19 emergency response such as lockdown on their livelihoods. there was also a lack of coordination among the different government stakeholders to tackle emergency healthcare and crisis management in the field. for instance, people usually made different excuses to go outside and a regular crowd was common in the kitchen market, streets, and small bazaars. only the government, semi-government, autonomous institutes/organizations, and educational institutions were maintaining the rules/guidelines. this situation is well visualized in different mass media that people are in movement for relief, road blockage, corruption by the government representatives, mismanagement in relief distribution, biases to party supporters, bureaucratic administrators to look after the response activities, and so on. likewise, the potential danger of covid-19 pandemic from the very beginning has been overlooked by the people due to the presence of misinformation in the social and mass media that it was general flue, and that the virus cannot infect in a humid country like bangladesh. so, the government should try to implement a stringent policy of risk communication and media communication during this emergency to the most vulnerable communities. the vulnerable groups such as disable and disadvantaged persons, young children and orphans, and aged citizens should be taken under protection for their well-being (undp 2020a). right now, doctors, bankers, grocers, police, and armed forces are the most vulnerable profession to the covid-19 infection. until 25 may 2020, 24% doctors, 16% nurse and 6% frontline healthcare workers were covid-19-infected. of the infected, 15 police personnel had so far died, while more than 5000 others are in either isolation or in quarantine (the daily star 2020). although the extension of partial lockdown was not a solution in bangladesh, it could have been an effective option continued to slower the infection rate. the lockdown should have been partially continued with necessary financial support for the vulnerable. it would have been a crisis for a short time, but it would be a saviour for the future (shammi and bodrud-doza 2020). however, to run the economy, the hotspots of the infection and the cluster areas could remain lockdown, while economic activities could have maintained by strongly abiding public health guidelines and social distancing. moreover, for the next couple of years, it will be extremely hard for the country especially as far as the financial issues are concerned to achieve the current development as well as sdg targets and reaching to middle-income countries (undp 2020a). gob should declare the delayed beginning of its 8th five-year national plan due to the covid-19 pandemic as a large part of it seems to be irrelevant at this stage, according to his proposals (the financial express 2020b). increasing surveillance as well as the reallocation of the budget, the distribution of direct cash, and private sector engagement could be some of the options to alleviate the crisis. in total, 54.7% of the respondents in this study agreed that due to full lockdown, the formal and informal business, economic and education sector will be hampered severely. 64.8% agreed that the poor and vulnerable communities both in urban and rural areas will be affected severely. for management purposes, 66% of the respondents thought that coordinated emergency relief support is required. overall, the respondents had a positive viewpoint about lockdown scenario 3 due to the covid-19 outbreak in bangladesh. if we have no other options, a strategic plan and policy should be taken for the revival of the health sector, economy, and education. it is speculative that a full lockdown might end up with famine and starvation. according to the world bank report (2020) prolonged and broad national lockdowns will bring a negative growth rate of the economy in bangladesh and other south asian countries in 2020 due to the covid-19 pandemic. this negative growth rate will continue in 2021 with growth projected to hover between 3.1 and 4.0%, down from the previous 6.7% estimate. a more serious issue that will arise due to the progress of the pandemic is the rate of suicide as a long-term effect on the vulnerable population due to fear and economic hardships (mamun and griffiths 2020) . preventing suicide and counselling mental health issues are therefore be considered by the authority (gunnell et al. 2020) . moreover, the authority should take proper steps to meet the basic emergency services and maintain the basic supply-demand of the daily needs of urban and rural people by transporting the crops and vegetable production from the farms. due to the lockdown, the farmers should not face any crop loss and they should be also brought under the financial and other stimulus plans so they can continue their productions for the future. if the needed government should give them free seeds, fertilizers, electricity for irrigations, and water and other incentives such as no-interest agricultural loans for future food security. the government already declared a financial recovery package with a clear disparity towards the agricultural sector. the financial stimulus package mainly focused on large and export-oriented businesses such as the readymade garment sector (rmg). it seems that this package has arrived a little earlier without any participatory strategic assessment. a strong collaborative need-based assessment is required to tackle the short-term and long-term needs to properly distribute the stimulus package. in this emergency response, the local government must have to come forward with full strength and capacities to implement the work plan for the gob. for overall relationship assessment for effective management of policy implications, governance, and developmental effects, pca (fig. 4) , cluster analysis (fig. 5) , and pearson correlation (table 5 and supplementary table 1 ) significantly show the relationships. pca showed a significant level of controlling factors in bangladesh covid-19 pandemic and how these statements are associated with the various scenarios (table 7) . nine principal components (pcs) were originated based on standard eigenvalues (surpassed 1) that extracted 52.195% of the total variance as outlined in table 7 . however, before pca applying in the tested data, the kaiser-meyer-olkin (kmo) and bartlett's tests of sphericity were conducted to appropriateness for this study. the findings of the kmo value in this research were 0.8 (> 0.50), the confidence level of bartlett's sphericity (bs) test was zero at p < 0.01, suggesting the tested data were fit for pca analysis. the scree plot was used to identify the number of pcs to be retained to the understanding of the inherent variable structure (fig. 4) . the loading scores were classified into three groups of weak (0.50-0.30), moderate (0.75-0.50), and strong (> 0.75), respectively (liu et al. 2003; islam et al. 2017) . the pc1 (first) explained 11.074% of the variance as it covered a significance level of strong positive loading of the lockdown scenarios and management 2 in bangladesh (ls1s5: 0.81 and ls1s6: 0.82). similarly, moderate positively loaded of the lockdown scenarios 1 in bangladesh (ls1s4: 0.608 ls1s7-s9: 0.741-0.795). the pc2 (second) explained 8.305% of the total variance and was loaded with moderate positive loading of lock drown scenarios 3 (ls3s4-9: 636-0.706 and ls3s12: 0.615). the pc3 (third) elucidated 8.305% of the variance and was strong positively loaded of massive awareness and enforcement of proper lockdown and quarantine initiatives (ls2m3: 0.836) and provide emergency relief to the poor communities both in urban and rural areas ensuring transparency (ls2m4: 0.789). furthermore, management scenario 2 and scenario 3 were observed moderate positive loading of pc3 (ls2m2: 0.637; ls2m5: 0.597, ls2m6: 0.642 and ls3m3: 0.547). the pc4 (four) accounted for 6.073% of the total variance and was strong positively loaded of poor people who will suffer food and the nutritional deficiency (ls2s9: 0.855) and moderately loaded in scenario 2 (ls2s5-s6: 0.652-0.657 and ls2s11: 0.637). the pc5 (five) explained 5.072% of the variance and was strong positively loaded of deep analysis of the situation and continue this existing partial lockdown (ls1m2: 0.779) and with moderately loaded in the management scenario 1 (ls1m1: 0.747 and ls1m3: 0.686). pc6 (six) accounted for 4.646 of the total variances and were strong positive loading of existing with increased facilities for covid-19 in the health system will be able to provide health services to the infected peoples and number of infection and death will be limited (ls3s2: 0.812 and ls3s3: 0.863) and with moderately loaded of very limited peoples movement will reduce the risk of community transmission of covid-19 (ls3s1: 0.689). pc7 (seven) explained for 4.419% of the variance and was strong positively loaded with gender-based violence will increase (ls2s10: 0.796) and gender discrimination will increase (ls3s10: 0.863). pc8 (eight) was responsible for 4.301% of the variance and was strong positively loaded with people will start moving towards regular life and formal (ls1s1: 0.866) and informal economical activities will be started (ls1s2: 0.836) and moderate positively loaded of massive movement and a mass gathering of people will be started again (ls1s3: 0.652). cluster analysis (ca) further recognized the total status of scenario variations and how these scenarios influence the socio-economic and development impacts (fig. 5) . all the statements were categorized into five major classes: cluster 1 (c1), cluster 2 (c2), cluster 3 (c3), cluster 4 (c4), and cluster 5 (c5). c1 consisted of five sub-clusters of c1-a, b, and c; c1-a composed of an irreversible loss to the economy and more people will die (ls1s12-ls1s15) c1-b comprised of community transmission of covid-19 will increase due to people's movement and mass gathering and panic will rise in the mass communities (ls1s4-s9). c1-c is comprised of the possibility of the full lockdown of the whole system again and no basic services will be available (ls1s10 and ls1s11). c2 consisted of three sub-clusters of c2-a, and b. c2-a consists of continue the existing partial lockdown and deep analysis of the situation and go for full lockdown with relief support to the poor and most vulnerable (ls1m1-m3) c2-b consists of people will start moving towards regular life and massive movement and a mass gathering of people will be started again (ls1s1-ls1s3). c3 consisted of three sub-clusters of c3-a, b, and c. c3-a contained an existing increase in the health facilities involving private sectors and implement inclusive sustainable quick plan and policies to revive the economy and employment (ls2m2-m6). c3-b consisted of lack of support and improper management will lead to the psychosocial and socio-economic crisis and long-term planning and implementation of policies regarding covid-19, psychosocial, and socio-economic loss (ls2s12 and ls3m6), while c3-c composed of continuous situation analysis of disease outbreak and implement the full lockdown with relief and basic support for human survival and loan support for business and economic recovery (ls3m1-m5). cluster 4 consisted of three sub-clusters of c4-a due to full lockdown, the formal and informal business, economic, and education sector will be hampered severely, loss of livelihood and unemployment rate will increase due to business shutdown, and poor communities in both urban and rural areas will be affected severely (ls3s4-6); c4-b supply and access to basic daily products in urban areas will be reduced drastically, the extreme need for relief and financial support in the urban and rural communities will increase, and people will be involved with conflict and crime to access the basic needs (ls3s7-12); and c4-c there will be less supply of basic products for daily use and price of most of the basic products will be higher than usual (ls2s7-s8). c4-d indicates poor people living from hand to mouth will be severely affected and the formal education system will be hampered. c5 consisted of two sub-clusters of c5-a, b, and c. c5-a contained gender-based violence will increase and gender discrimination and violence will increase ls2s10 and ls3s10. c5-b comprised of limited people's movement will enable low-level community transmission of covid-19 and infection and death rate will increase slowly (ls2s1-s3). c5-c contained limited people movement will reduce the risk of community transmission of covid-19 and the number of infections will be limited ls3s1-s3. the covid-19 pandemic has the most effects on vulnerable populations, ranging from good health and well-being (sdg 3) to quality education (sdg 4) worldwide. disruptions in the routine health care, poverty, and access to food and nutrition will culminate into unavoidable shocks and health system collapse which will increase child mortality and maternal deaths as well as many unwanted deaths (roberton et al. 2020) . the crises in achieving clean water and sanitation targets (sdg 6), weak economic development and the absence of decent jobs (sdg 8), overall inequality (sdg 10), and above all, no poverty (sdg 1), and food insecurity (sdg 2) will be aggravated in many developing countries. the world bank reports that about 11 million people will be forced into poverty by the crisis (wef 2020b). according to undp (2020b), revenue losses in developing countries are estimated to reach $220 billion. the losses would be consequences of the education, human rights, and, in the most extreme cases, fundamental food security and nutrition, with an estimated 55% of the global population not having access to social protection. wider socio-economic effects will likely continue for several months to years across the world which will also significantly impact the economy of bangladesh. global food security will be hampered as one-third of the world's population is in lockdown (galanakis 2020) . both the import of important goods and exports related to the readymade garment sector and others likely will be affected for income and employment. financial protection during outbreak matters. at the initial stage of the covid-19 epidemic, out-of-pocket expenditure posed a substantial financial burden for the poor populations with severe symptoms, even for those under coverage by the social health insurance scheme (wang and tang 2020) . people marginally above the poverty level particularly low-income families, daily and informal low wedge earners, ethnic community groups, people with disabilities, and returnee migrant workers are already started falling below the poverty line due to loss of income and employment. brac an international bangladeshi ngo survey report confirmed to increase a 60% rise in poverty amidst the covid-19 pandemic (brac 2020). the intake of foods, vegetables, and herbs can boost the immune system against the infection disease, while it can stimulate the transmission through the food chain (galanakis 2020) . again, the lack of food will rise to malnutrition, hunger, and famine. approximately 265, million people worldwide will be suffering from acute hunger projected by the un world food programme (wef 2020c). ready-made garment (rmg) sector is going to suffer a serious shortfall as until 24 march 2020, orders of rmg products from 738 garment factories worth us$ 2.4 billion was cancelled. this is the sector where almost 4 million low-income people-of whom over 85% are women-work and another similar number of people indirectly depend on the downstream and upstream services required by the rmg value chain (dhaka tribune 2020a). as the lockdown continues to ensure public health and safety, many rmg workers already lost their jobs and did not receive their salary of the previous months, some of them have been sacked already. food security and social and economic recovery package of the government should focus on immediate response during the lockdown period and outbreak and post-lockdown support mechanisms. in this condition, middle-income families are relying on their savings available. the negative coping mechanism includes skipping meals and nutrition and distressing the whole family. in the prolonged lockdown scenario, they need government and other support measures to continue their lives under lockdown. due to lockdown, the agricultural products in the urban areas are selling at a high price, while the farmers are not getting the fair price of the product in agricultural districts. it was due to the proper decisions and policy of the gob that aman paddy was timely harvested ensuring the safety of migrant workers. otherwise, it would have likely imposed a bigger social and economic implications such as heavy rainfall triggering natural flash flooding. moreover, due to the lockdown transport of animal, poultry and fish feed are hampered. likewise, due to the closure of local restaurants and hotels, the market demand for eggs and chicken had lowered. all this will likely impose further impacts on food production and crop supply chains. to protect the country from famine, the bangladesh government should consider the stimulus package for the farmers with 2% agricultural loan to continue cropping and agricultural production. receiving education has stopped for most of the students in bangladesh. the government of bangladesh postponed all academic and public exams until the indefinite period, considering the growing public concern. distance learning education of the national curriculum through air transmission in the national tv had started though. while urban children can attend virtual classes through the internet, rural and marginalized children are deprived due to limited resources. students from marginalized backgrounds particularly with disabilities will lose out more on their education. considering this, gob should prepare special educational package including counselling for marginalized and disadvantaged students. the severe infection of covid-19 pandemic has devastated the healthcare systems across the globe from a shortage of n95 masks, and personal protective equipment (ppes) for the healthcare workers and putting occupational health risk, allocations of ventilators, icus, and hospital beds to a patient who can benefit most from treatment while letting the older persons to death. the peaked disparity between supply and demand for healthcare properties raised a normative query of equitable resource allocation during the covid-19 pandemic (emanuel et al. 2020 ). thousands of healthcare workers have already been infected worldwide (gan et al. 2020) , and the administrative and managerial departments are likely to place increased burdens and stresses on the frontline healthcare workers (willan et al. 2020 ). bangladesh has no exceptional scenario. on 31 may 2020, gob lifted its partial lockdown after 65 days of general holidays. the gob claimed the withdrawal of lockdown as a test for next 15 days from 31 may to 15 june 2020, but it was decided without having the designated committees' opinion rather only considering the economic considerations. the gob is planning to divide areas around the country that are affected by the covid-19 into three zones based on colour as red, yellow, and green indicating the severity of cluster infections and to prevent the disease spread (the daily star 2020). at present bangladesh is at number 21 considering the infections and mortality from covid-19 (worldometer 2020) . the overall attack rate among the bangladeshi population is 208.9/1 million and more than 20% positive cases have been identified in the recent days reported in the who situation report on 25th may 2020 (who 2020c). among the countries of india, pakistan, nepal, bhutan, sri lanka, thailand and vietnam, bangladesh is at the bottom in terms of the number of covid-19 tests done per million population (newage 2020) . the maldives and bhutan are on the top of the list with each conducting 21,822 tests per millions of people (tbs news 2020c). the testing laboratories are situated in the urban metropolitan areas and often due to fear and social stigma the patients do not want to test. moreover, the incidences of a false negative in one laboratory while positive in another laboratory had been reported in mass media. in addition, the mortality rate from covid-19 infection remains a puzzle which just cannot be explained by the gdp of the country, strength of healthcare governance and availability of equipment like icu or ventilators. the trend of screening and testing (1877/1 million population) and contact tracing the covid-19 patients in bangladesh is not quite enough to conclude that the curve is flattening, or the peak of the curve has reached. thus, at this point, the database does not seems to be robust and it could be chaotic from the epidemiological point of view. after the lockdown is withdrawn, it was speculated that the number of infections will increase as the life and livelihood needed to sustain. on 31st may 2020, bangladesh recorded 40 deaths from covid-19 and 2545 new infections (iedcr 2020). at this stage, gob should increase the icu numbers and strengthen the healthcare departments by recruiting more doctors, nurses, and technicians. rapid testing, screening and diagnosis should be increased which was the advice of who from the beginning. along with isolation, clinical management, and infection prevention and risk communication should be continued to the public. the gob should engage public and private hospital authorities for the treatment of covid-19 infected patients and resume treatment of other critical-care patients who are being deprived of any treatment at present. moreover, as the infection from dengue is also rising government should take special emphasis for dengue treatment and management also. in fig. 6 we have outlined the overall impact and management analysis of the three scenarios: scenario 1, scenario 2, and the scenario after the withdrawn of partial lockdown. community health workers can support pandemic preparation earlier to the epidemics by increasing access to the healthcare services and the healthcare products within the communities. they can communicate disease risks and increase awareness in the comparative lockdown scenarios with impact and management analysis for bangladesh due to covid-19 pandemic respected area in cultural language whereas reducing the weights of the formal healthcare systems. community healthcare workers can also contribute to pandemic preparedness by acting as community-level educators and mobilizers, contributing to surveillance systems, and filling health service gaps (boyce and katz 2019) . it is critical to detect cluster surveillance of covid-19 to better allocate resources and improve decision-making as the outbreaks continue to grow in different districts of bangladesh to improve resource allocation, faster testing stations, stricter quarantines and city/block lockdowns as well as travel bans (desjardins et al. 2020) . it is predictable that environmentally the decrease in air pollution reduces preventable communicable and non-communicable diseases such as covid-19 (dutheil et al. 2020) . likewise, ma et al. (2020) mentioned that the warmer season and lockdown activities were the keys to reduce exposure to novel coronavirus on humans in china. although the relationship between the infection rate and climatic variables is not confirmed in bangladesh, as the partial lockdown failed and continued, the number of infections over the past days indicates that gob should have ensured proper implementation of the lockdown scenario 2 with limited public movement in the hotspots, resulting in lower community transmission of the virus and a slower death rate, while continuing economic activity with strict guidelines. gob was looking forward to exiting from partial lockdown beginning of may, yet no specific exit plans were executed by the government which should be scientifically rational and practically achieved. the exit plans from the lockdown should have been well communicated to the public ensuring transparency. without ensuring safety and security the partial lockdown was withdrawn. public transportation started on 31 may without maintaining any health guideline (tbs news 2020d). coordination among the different stakeholders of the government is necessary, along with increased surveillance and resource allocation to the needy ones, to ensure supply of daily necessities, control price hikes, and reduce the loss of livelihood and unemployment. moreover, very recently cyclone amphan hit bangladesh on 20 may 2020, living the coastal districts flooded and in the mayhem. preliminary losses were estimated to be worth bdt12,744 crores (dhaka tribune 2020b). at this stage detection of covid-19 hotspots by increased testing facilities all over the country must be ensured. the poor and vulnerable communities always lack food and nutrition due to injustice and corruption by local political stakeholders. the vulnerable groups, such as disabled and disadvantaged persons, young children and orphans, and elderly citizens, should be taken under protection for their well-being. they should be provided with food and nutrition for the time being. covid-19 pandemics cause environmental, economic, and social attributes which have only partially been described in bangladesh. to fight this pandemic, it requires remarkable tasks and partnership development in the local and global level. the world must prepare for the likelihood that mitigation measures might fail because lockdown periods in different countries took different times to prevent or suspend the spread of covid-19 (gautam and hens 2020) . collective responsibility is required from the public as well to protect themselves by abiding general health guideline, maintaining hygiene and social distancing, and avoiding going to crowded places and meetings. extremely coordinated and effective planning and strategies for both the ongoing and afterwards response are required from the gob to manage this pandemic and take it as a new "standard of normal". considering the global hard-hit economy, depression, unemployment, job loss, shortfall of rmg export and incoming remittances, the socio-economic and development impacts along with the food insecurity as well as rising poverty due to covid-19 at the community level need to be coordinated in bangladesh. at present, as the lockdown is withdrawn, both lives and livelihoods are in danger which is a long-debate that is going on. along with the pandemic disease, the upcoming seasons of natural disasters from cyclones, tidal floods, flash floods, and landslides of monsoon seasons should be considered to prepare for the emergencies. all these will further aggravate the humanitarian needs of the most vulnerable groups in the country in the coming months to be followed. as the health sector is the most strained at present, it will affect the targets of sustainable development goals of 2030. in addition, quality education will be hampered in the country. the government of bangladesh has already mobilized a noteworthy stimulus package to support the affected industries and community which needs to be coordinated over a longer period of 12-18 months and may be incorporated in the upcoming 8th 5-year plans with substantial revising. however, this package should also include research and innovation, recovery of education. there is no alternative to strengthen the health care facilities and preparedness for the potential humanitarian crisis. moreover, humanitarian support should reach the most vulnerable communities which need to be targeted, outlined, and delivered. finally, economic implications should be subjected to the spatial and geographical locations based on the vulnerabilities. hotspots identified in the delta plan can be considered here. the long-term strategic plan can be integrated into perspective plan 2041 and bangladesh delta plans 2100, for better strategic management. whatever will be the lockdown scenario, the basic supports to the mass people must be ensured and that is not so easy without strong strategic planning and multisectoral collaboration for sustainability including supports from the private sectors and international bodies. community health workers and pandemic preparedness: current and prospective roles covid-19 situation report 9 new zealand has no new coronavirus cases and just discharged its last hospital patient. here are the secrets to the country's success covid-19: preparedness, decentralisation, and the hunt for patient zero at current pace, bangladesh to end extreme poverty by 2021. the economic times new zealand eliminates covid-19 rapid surveillance of covid-19 in the united states using a prospective space-time scan statistic: detecting and evaluating emerging clusters scale development: theory and applications coronavirus: bgmea says orders worth $3.15 billion cancelled so far cyclone amphan: bangladesh may face losses worth tk12,744cr building resilience against biological hazards and pandemics: covid-19 and its implications for the sendai framework covid-19 as a factor influencing air pollution? environmental pollution fair allocation of scarce medical resources in the time of covid-19 federation of bangladesh chambers of commerce & industries the food systems in the era of the coronavirus (covid-19) pandemic crisis preventing intra-hospital infection and transmission of coronavirus disease 2019 in health-care workers. safety and health at work sars-cov-2 pandemic in india: what might we expect? environment global implications of bio-aerosol in pandemic suicide risk and prevention during the covid-19 pandemic. the lancet psychiattry what are the real choices? tbs news bangladesh covid-19 update national preparedness and response plan for covid-19 characterizing groundwater quality ranks for drinking purposes in sylhet district, bangladesh, using entropy method, spatial autocorrelation index, and geostatistics simultaneous comparison of modified-integrated water quality and entropy weighted indices: implication for safe drinking water in the coastal region of bangladesh potential association between covid-19 mortality and health-care resource availability. the lancet global health application of factor analysis in the assessment of groundwater quality in a blackfoot disease area in taiwan response to the covid-19 epidemic: the chinese experience and implications for other countries when will covid-19 end? data-driven prediction. singapore university of technology and design effects of temperature variation and humidity on the death of covid-19 in wuhan, china. science of the total environment first covid-19 suicide case in bangladesh due to fear of covid-19 and xenophobia: possible suicide prevention strategies penal code 1860: section 144 bangladesh worst coronavirus fighter in the region assessing global preparedness for the next pandemic: development and application of an epidemic preparedness index early estimates of the indirect effects of the covid-19 pandemic on maternal and child mortality in lowincome and middle-income countries: a modelling study. the lancet global health retrieved from https ://tbsne ws.net/thoug hts/lockd ownwithd raw-or-not-withd raw-77752 ?fbcli d=iwar0 oanmr goirz omt2a fhrle xytyh 49utc 0i8j9 yocfj v0dba -wapgs 3mfjm covid-19 pandemic, socioeconomic crisis and human stress in resource-limited settings: a case from bangladesh. heliyon icu space limited, critical covid-19 patients at risk half the country's covid-19 infection in dhaka city alone bangladesh lowest in covid-19 testing in south asia social distancing in public transport yet another cruel joke covid-19 pandemic: govt plans to divide country into red, yellow, green zones economic ramifications of covid-19 in bangladesh covid 19: sanem proposes 2-year recovery plan for bangladesh an investigation of transmission control measures during the first 50 days of the covid-19 epidemic in china the toughest triage-allocating ventilators in a pandemic undp bangladesh situation analysis: support to the national response to contain the impact of covid-19 covid-19: looming crisis in developing countries threatens to devastate economies and ramp up inequality combating covid-19: health equity matters strengths, weaknesses, opportunities and threats (swot) analysis of china's prevention and control strategy for the covid-19 epidemic viet nam shows how you can contain covid-19 with limited resources why we cannot lose sight of the sustainable development goals during coronavirus doubl e-acute -hunge r-un-warns ?fbcli d=iwar1 6jlhk fiove fpyf0 jwcdv qqovv j3qsa xwx2t gqv-iwnmy cri9d ydidr 4o health emergency and disaster risk management framework who coronavirus disease (covid-19) dashboard who bangladesh covid-19 situation report no who bangladesh covid-19 situation report no 13 challenges for nhs hospitals during covid-19 epidemic bangladesh must ramp up covid-19 action to protect its people, revive economy covid-19 coronavirus pandemic health emergency and disaster risk management: five years into implementation of the sendai framework covid-19 containment: china provides important lessons for global response publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations the authors would like to acknowledge all the frontline doctors, healthcare workers, emergency responders, security, and armed forces fighting this pandemic.funding this research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. key: cord-220723-yl2tg6q4 authors: lv, zhaofeng; wang, xiaotong; deng, fanyuan; ying, qi; archibald, alexander t.; jones, roderic l.; ding, yan; cheng, ying; fu, mingliang; liu, ying; man, hanyang; xue, zhigang; he, kebin; hao, jiming; espc, huan liu state key joint laboratory of; sources, state environmental protection key laboratory of; complex, control of air pollution; innovation, international joint laboratory on low carbon clean energy; environment, school of the; university, tsinghua; china,; civil, zachry department of; engineering, environmental; university, texas am; usa,; science, centre for atmospheric; chemistry, department of; cambridge, university of; uk,; sciences, chinese research academy of environmental; institute, beijing transport title: significant reduced traffic in beijing failed to relieve haze pollution during the covid-19 lockdown: implications for haze mitigation date: 2020-06-12 journal: nan doi: nan sha: doc_id: 220723 cord_uid: yl2tg6q4 the covid-19 outbreak greatly limited human activities and reduced primary emissions particularly from urban on-road vehicles, but coincided with beijing experiencing pandemic haze, raising the public concerns of the validity and effectiveness of the imposed traffic policies to improve the air pollution. here, we explored the relationship between local vehicle emissions and the winter haze in beijing before and during the covid-19 lockdown period based on an integrated analysis framework, which combines a real-time on-road emission inventory, in-situ air quality observations and a localized chemical transport modeling system. we found that traffic emissions decreased substantially affected by the pandemic, with a higher reduction for nox (75.9%, 125.3 mg/day) compared to vocs (53.1%, 52.9 mg/day). unexpectedly, our results show that the imbalanced emission abatement of nox and vocs from vehicles led to a significant rise of the atmospheric oxidizing capacity in urban areas, but only resulting in modest increases in secondary aerosols due to the inadequate precursors. however, the enhanced oxidizing capacity in the surrounding regions greatly increased the secondary particles with relatively abundant precursors, which is mainly responsible for beijing haze during the lockdown period. our results indicate that the winter haze in beijing was insensitive to the local vehicular emissions reduction due to the complicated nonlinear response of the fine particle and air pollutant emissions. we suggest mitigation policies should focus on accelerating voc and nh3 emissions reduction and synchronously controlling regional sources to release the benefits on local traffic emission control. email: liu_env@tsinghua.edu.cn. the covid-19 outbreak greatly limited human activities and reduced primary emissions particularly from urban on-road vehicles, but coincided with beijing experiencing "pandemic haze", raising the public concerns of the validity and effectiveness of the imposed traffic policies to improve the air pollution. here, we explored the relationship between local vehicle emissions and the winter haze in beijing before and during the covid-19 lockdown period based on an integrated analysis framework, which combines a real-time on-road emission inventory, in-situ air quality observations and a localized chemical transport modeling system. we found that traffic emissions decreased substantially affected by the pandemic, with a higher reduction for nox (75.9%, 125.3 mg/day) compared to vocs (53.1%, 52.9 mg/day). unexpectedly, our results show that the imbalanced emission abatement of nox and vocs from vehicles led to a significant rise of the atmospheric oxidizing capacity in urban areas, but only resulting in modest increases in secondary aerosols due to the inadequate precursors. however, the enhanced oxidizing capacity in the surrounding regions greatly increased the secondary particles with relatively abundant precursors, which is mainly responsible for beijing haze during the lockdown period. our results indicate that the winter haze in beijing was insensitive to the local vehicular emissions reduction due to the complicated nonlinear response of the fine particle and air pollutant emissions. we suggest mitigation policies should focus on accelerating voc and nh3 emissions reduction and synchronously controlling regional sources to release the benefits on local traffic emission control. the unexpected covid-19 epidemic in 2020, which coincided with the spring festival, the most important holiday in china, put the chinese economy into a rapid stall. the spring festival migration reduced the population in beijing to a low level, with an estimated 39% reduction from the 22 million residents in normal times (1) , while the coronavirus pandemic lockdown further reduced activities. the stay-at-home orders were initially started in wuhan from january 23rd, 2020, one day before the eve of the spring festival, and soon applied to the whole country (2) . in addition, for those who enter beijing, a two-week compulsory quarantine was implemented. the spring festival holiday and the coronavirus restrictions led to widespread shutdowns and a near-halt in normal life and economic activity in beijing and its surrounding cities. generally, pandemic lockdowns led to clearer skies in china and other places in the world (3) (4) (5) . however, severe air pollution episodes occurred in beijing during "the most silent spring", leading to query the response of air pollution to anthropogenic activities. the unexpected heavy pollution put doubt on current understanding of the source-receptor relationship in beijing. previous work demonstrated that vehicles contributed 45% to ambient pm2.5 caused by all local sources (6) . following scientific instructions, control measurements were undertaken over the past 5-years to reduce the sources of aerosol pollution (7) . as a result, the pm2.5 annual concentration in beijing decreased from 89.5 μg/m3 in 2013 to 42 .0 μg/m3 in 2019, and heavy pollution days were also reduced from 58 days in 2013 to 4 days in 2019 (8) , providing confidence in source-receptor mechanisms supporting pollution control strategies. however, during the covid-19 lockdown the air quality index (aqi) frequently hit extremely unhealthy levels in beijing, including january 25th-29th and february 9th-13th, with a peak daily pm2.5 concentration reaching 218.3 μg/m3 on february 12th, more than 8 times higher than the world health organization (who)'s recommended level of 25 μg/m3 for 24-hr average concentrations. here we show that the severe haze pollution, which occurred in spite of the significantly reduced human activities, highlights weaknesses in our current source-receptor understanding. the "pandemic haze" in beijing raised great attention from the public and government on the role of vehicle emissions in the burden of air pollution. historically, vehicle emission controls have been used as an effective way to relieve air pollution in megacities (9, 10) . beijing undertook a lot of effort to reduce its traffic emissions through strict controls on new vehicle registration, limited usage based on plate number, upgraded vehicle emission standards and shifting to greener transportation (11, 12) . in beijing, car license plate is regarded as a limited public resource. to get a conventional gasoline car, residents need to be in a bimonthly lottery pool, competing with more than 3 million fellow residents with an odds of around 1/2900. beijing residents believed these restrictions could help improve the air quality, as official reports showed that the vehicles contributed 45% to ambient pm2.5 from local sources in beijing (6) . however, recent air pollution episodes in beijing have made the relationship tangled. what is the role of vehicular emission reduction in pm2.5 pollution during the covid-19 outbreak? is traffic emission control still a necessary and effective way to relieve the winter haze in a megacity like beijing? here, we presented a source-receptor analysis on the covid-19 pandemic haze events in beijing based on emission inventory, air quality observations and numerical models. our study integrated multiple real-time traffic data around the covid-19 outbreak and developed a novel realistic traffic emission inventory for beijing. it was applied in a series of counterfactual modeling experiments by a localized chemical transport modeling system and a tracing-based source apportionment to understand the mechanisms and the role of local vehicle emission reductions for the pandemic haze, and to propose the future development of vehicle emission control strategy. the details of our analytic approach are documented in materials and method and the si appendix. roads within the 5th ring road at morning rush hours. average daily vehicle kilometers of travel (vtk) of light duty vehicles (ldv), heavy duty vehicles (hdv), light duty trucks (ldt) and heavy duty trucks (hdt) decreased by 28.3%, 60.9%, 36.7% and 55.4%, respectively, during the lockdown period. after the 7-day spring festival holiday, the activity of ldts gradually increased in order to meet the urban demand, but was still 29.4% lower than the level during the pre period. compared with the data of the same period around the spring festival in 2019, the traffic speed affected by the pandemic still remained at a high level after the 7-day holiday during the lockdown period, with 6.3%-13.8% higher than that in 2019 if snowy days were excluded (si appendix, fig. s2 ). these increases in traffic speed and decreases in traffic flow in beijing, for such a long time, were significantly more marked than any previous holidays in the past few years (13) . based on the slove model and trackatruck model, the real-time on-road emissions of beijing were calculated for the pre period, the transition period, and the lockdown period. as a consequence of the covid-19 pandemic, vehicle emissions decreased by 50.9%-75.9%, with particularly high reduction (75.9%) for nox emissions ( fig. 1 -c, other pollutants were shown in si appendix, fig. s3 ). the relative reduction of emissions from transportation sector was larger than the averaged decrease of all sectors, which were decreased by 20.6%-69.3% during the lockdown period. the spatial distribution showed that in the lockdown period, vehicle emissions decreased substantially on almost all roads, especially for ring roads during the traffic rush hours and the main freight channels at night (si appendix. fig. s4 ). the diurnal variations of vehicle emissions also showed a significant change during the lockdown period (si appendix. fig. s5 ). two emission peaks were observed on both weekdays and weekends during the pre period, with the highest hourly nox emissions reaching up to 9.2 mg/hour at 17:00 on weekdays. however, during the lockdown period, the hourly on-road emissions showed much smaller variations and the difference between weekdays and weekends became smaller. meanwhile, nox emissions at the evening traffic peak declined to 1.8 mg/hour. in sum, our results indicate that the covid-19 outbreak led to a significant reduction in traffic activities and emissions compared to those in the pre period, and also changed the spatial distribution and diurnal variations of vehicle emissions. the pre period vehicle emission estimates in this study were at the same magnitude with recent chinese government led research (si appendix, fig. s6 ). before covid-19 lockdown, on-road emissions were estimated to be 496.8 mg/day (co), 99.6 mg/day (hc), 165.1 mg/day (nox), and 5.1 mg/day (pm2.5), accounting for 45.8%, 29.3%, 65.6%, and 21.3% of the total anthropogenic emissions, respectively. hdts were responsible for only 12.9% and 9.9% of nox and pm2.5 emissions from all vehicles, much lower than previous estimations (14, 15) , because of the implementation of a hdt low emission zone in beijing. compared to estimations for 2013 based on a similar bottom-up method (14, 15) , nox and pm2.5 emissions from on-road traffic reduced by 43.5%-49.4% and 52.3%-55.3%, indicating the effectiveness of the continuous vehicular emission control measures. all these features influenced the role of local traffic emissions in air pollution. fig. 2 -a shows the observed temporal variations of daily aqi and pm2.5 concentrations before and during the covid-19 outbreak, respectively. the haze pollution became more severe during the lockdown period compared to that in either the pre period or the transition period, with the mean daily pm2.5 level unexpectedly increasing from 48.0 μg/m3 to 99.0 μg/m3. moreover, in the lockdown period, half of the days were polluted with daily pm2.5 concentrations exceeding 75 μg/m3, the level ii standard of the chinese national ambient air quality standards (naaqs). the pm2.5 level remained at more than 150 μg/m3 for two episodes from january 25th to january 28th and february 11th to 13rd (the first of these was excluded from this analysis since it was probably caused by fireworks (16) ). the variation of pm2.5 concentrations on polluted days was seen as an asymmetric "saw tooth" pattern, rising slowly before two days and then falling abruptly (17) . both in the pre period and lockdown period, these "saw tooth" periods were selected as the heavy pollution periods (hpps), and other consecutive clean days (daily pm2.5 level less than 75 μg/m3) were defined as non-heavy pollution periods (nhpps). in this study, we separate episodes of hpps and nhpps to make comparisons between the pre period and the lockdown period. fig. 2 -a also shows the time series of the secondary aerosol enhancement, using the ratio of pm2.5 major secondary components (including sulfate, nitrate, ammonium and organic matter hereafter snao) to the elemental carbon (ec), to eliminate the impacts of the mixing-layer height on pollutant concentrations (18) . the ratios of snao to ec were stable between the pre period and transition period, while a significant rise was found in the lockdown period under either the hpp or nhpp, with an average increase of 51.8% compared with that in the pre period. we further investigated the changes in the diurnal variations of snao/ec between the pre period and lockdown period (fig. 2-b) . the enhancement of secondary aerosols during the covid-19 outbreak was evident during the entire day with a peak level in the early morning (9:00-10:00 a.m.). this was especially true for nitrate aerosols, which were presented at a peak level more than twice that observed in the pre period. in spite of the increases in snao during the lockdown period, the concentrations of no2 and so2, regarded as two major gas-phase precursors of secondary pm2.5 (nitrate and sulfate), declined by 29.7% and 34.6% on average ( fig. 2-c) . focusing on the differences between nhpps, the relative reduction of no2 concentrations reached 57.9% during the lockdown period compared to the pre period. this was consistent with the relative reduction of estimated primary emissions (69.3%), indicating that the large emission reduction of local anthropogenic sources actually decreased the pm2.5 precursor concentrations during the covid-19 outbreak. ozone is one of the most important oxidants in tropospheric chemistry. as shown in fig. 2 -c, the observed surface ozone increased up to 263.3% between hpps during the lockdown compared to the pre period, with a period-averaged enhancement of 62.0%. in addition, we investigated changes in the nitrate radical (no3), the primary oxidant for nighttime secondary aerosol formation (19) (20) (21) . the change in diurnal variations of o3, no3 radical and no2 concentrations were provided in si appendix, fig. s7 . compared to the pre period, no3 radical concentrations also increased especially at night during the lockdown period. these changes during the covid-19 outbreak and the subsequent lockdown indicate that the increased concentrations of oxidants facilitated the chemical formation of secondary fine particles in spite of the significantly reduced gaseous precursors, particularly resulting in a fast nitrate growth during the nighttime. the significant enhancement of the oxidizing capacity was responsible for parts of the rapid growth of secondary aerosols during the lockdown period. our results using the wrf-cmaq modeling system (base scenario for the real-time simulation, see methods for more details) also revealed a significant increase of oxidant concentrations in most areas of beijing during the covid-19 outbreak (si appendix, fig. s8 ). the o3 variation from the pre to lockdown period was investigated using data in the nhpps in order to reduce the disturbance of regional transport on air quality ( fig. 2-d) . the o3 concentration increased by up to 16 parts per billion by volume (ppbv or nmol/mol) covering the northwest to the southeast of beijing. the estimated vehicular nox emission also showed an obvious decrease in the same areas, including urban areas within the 6th ring road and major freight corridors with massive hdts (fig. 2-e ). in addition, the surface ozone in southern downwind areas also increased most likely due to the prevailing north wind in nhpps carrying high o3 and precursor gases concentrations from urban areas. as the o3 formation was voc-limited in urban areas, the relatively larger emission reduction of nox compared to vocs raised the voc/nox ratio ( fig. 1-c) , which consequently led to an increase in o3 concentrations (22) (23) (24) . by contrast, a relatively small emission reduction led to little changes or declines in o3 in western and northern rural areas, where o3 formation is in the transition or nox-limited regimes (22) (23) (24) . when focused on emissions within urban areas from different sources, we found that on-road nox emission reductions due to the lockdown were responsible for over half (53.4%) of total nox emission decrease, while the ratio for vocs was only 17.8%. our results indicate that such an imbalance in emission reduction of nox and voc from vehicles was probably responsible for the enhancement of local atmospheric oxidizing capacity, further facilitating the chemical formation of secondary aerosols during the covid-19 lockdown. a hypothetical scenario (s1) was set up in which the on-road emissions during the lockdown period were assumed to be as usual in the pre period, and other emissions and meteorological conditions were the same as base scenario. the differences between base and s1 just reflected the impacts from vehicular variations. fig. 3 shows the changes in spatial distribution of the oxidants and pm2.5 concentrations between base and s1 from wrf-cmaq modeling results. a significant enhancement in o3 concentration of up to 11 ppbv was seen in urban areas within the 6th ring road and southern areas, induced by the larger vehicular nox emission reduction compared to vocs. compared to fig. 2 -d, which shows the o3 variation from the pre to the lockdown period, the o3 enhancements only caused by vehicle emissions are similar (fig. 3-a) . it indicates that local traffic emission reduction is the main driving force for the enhanced oxidization capacity in urban areas of beijing. compared to changes in o3 concentrations, the increase of no3 radical was relatively small at the surface level because of the sharp decline of ambient no2 concentrations ( fig. 3-b) , while a more obvious enhancement was found in the upper air within urban areas (approximately 46m above the ground) due to relatively weak no-titration effects (si appendix, fig. s9 ). the enhanced oxidants facilitated the formation of secondary organic matter (som) during the day and nitrate aerosols in nighttime (si appendix, fig. s10 ). however, the increased secondary aerosol formation was small, only with a rise by up to 1.6 μg/m3 ( fig. 3 -c), probably because of the inadequate precursors particularly during the lockdown period. while the increased secondary aerosols were still enough to offset the benefit of vehicular reductions in primary emissions, leading to a modest increase of total pm2.5 concentrations by up to 1.4 μg/m3 ( fig. 3 our simulations show that the spatial variations of atmospheric oxidation induced the opposite changes in pm2.5 formation in rural (outside the 6th ring road, except for the southeast) compared to urban areas. the enhancement of o3 in rural areas was relatively small, because 80.7%-82.7% of the vehicle emission reductions in beijing were concentrated in urban areas and the ozone formation regime changed from voc-limited to nox-limited going from urban to rural areas. in addition, the reduction of gas-phase precursors restricted the production of secondary particles in rural, resulting in the decrease of snao and pm2.5 concentrations. such contrasting impacts between urban and rural areas on both snao and pm2.5 were more distinct during the hpp compared to those in nhpp, mainly due to unfavorable meteorological conditions for air pollutant dispersion during the hpp (25, 26) . as a conclusion here, the imbalance in emission abatement of nox and voc from vehicles was an important cause for the rise of local atmospheric oxidizing capacity, resulting in a modest enhancement of secondary aerosols and pm2.5 concentration. however, the slightly increased pm2.5 induced by the giant vehicular emission reduction could not explain the significant growth of secondary aerosols in beijing during the covid-19 lockdown period. the wrf-cmaq modeling system combined with the integrated source apportionment model (isam) were further applied to trace the contributions of emissions from local sources (onroad vehicles, industry, domestic and others) and regional sources outside beijing to pm2.5 and sna (the sum of sulfate, nitrate, ammonium) concentrations individually (fig. 4) . our results show that during the pre period, the local sources only contributed 19.1% and 30.1% of pm2.5 concentrations in winter hpp and nhpp, respectively, which was much less than the results in previous researches with target years of 2012-2014 (44.0%-69.5%) (27) (28) (29) (30) . this was mainly owning to a more significant reduction of local emissions in beijing compared to those of surrounding areas in recent years, particularly for the notable decline in local power and heating, industry, and residential sectors from 2013 to 2017 (7) . the primary pm2.5 from residential sources including both cooking and heating was still the largest contributor among the local sources (hourly averaged 12.0%), followed by "other" local sources (7.5%), in which nh3 emissions forming particulate ammonium were the most important. however, little influence of local vehicles and industry was identified with contributions both less than 3%. as sulfate and nitrate aerosols were more easily transported over long-distance compared with ec, the contribution of regional sources to sna concentration reached 89.1% and 84.9% during the hpp and nhpp during the pre period, respectively. during the lockdown period, the contribution of regional transport to ambient pm2.5 increased to 86.8% and 88.5% during the hpp and nhpp, respectively, and it was responsible for more than 90% of sna concentrations, since more emissions from industry and residential sectors were reduced in beijing compared to its neighboring regions during the covid-19 lockdown (si appendix, table s1 ). our modeling results showed that the enhancement of atmospheric oxidizing capacity during the lockdown period was recognized in not only beijing, but also most areas of the beijing-tianjin-hebei (bth) and its surrounding regions (si appendix, fig. s11 ), consistent with recent research (31) . differing from "significant increased o3 but modest enhanced secondary aerosols" in beijing, the increased oxidants in the case of relatively abundant gaseous precursors outside beijing facilitated more chemical formation of secondary organic and inorganic aerosols, which were transferred into beijing and significantly increased the local secondary pm2.5 concentrations. the contributions of almost all local sources to pm2.5 and sna concentrations decreased sharply during the lockdown period, while the contribution of "other" local sources still remained at a normal level due to the important role of its nh3 emissions in the chemical growth of secondary inorganic aerosols. as a result, although traffic activities and emissions in beijing were significantly reduced during the pandemic, it still could not turn over the aggravated haze pollution, due to the complicated nonlinear response of the fine particle and air pollutant emissions as well as the dominant impacts of regional sources. beijing "pandemic haze" is a challenging case for source-receptor relationships (fig. 5 ). large emissions from industry in the bth region (including boilers, cement, steel production and other industrial processes), as well as the further increased emissions in winter for heating (32, 33) , provide sufficient precursors to form the secondary aerosol in the case of enhanced oxidants, which transported into beijing, resulting in an aggregated haze pollution. even during the lockdown period, most of the heavy industries in hebei province were still in continuous operation with little emission reductions (34) . on the other hand, since beijing is already in the "low concentration pool" of bth region, a significant enhanced local atmospheric oxidizing capacity caused by the imbalanced vehicular emission reduction of nox and voc, only leads to a modest increase of secondary aerosols and pm2.5 concentration due to the inadequate precursors. all these make the relationship of vehicle emissions and air quality in beijing different from that in other megacities (35) . the public were disappointed by the "pandemic haze" because of an expectation that previous efforts on controlling traffic to avoid the pm2.5 pollution would mean that during the lockdown there would be no pollution problems. our modelling results indicate that the local traffic activity had little impacts on the winter haze events in beijing. in addition, based on a series of sensitivity runs modifying the local on-road emissions in winter normal days (pre period), even without vehicle emissions, the ambient pm2.5 concentration could only be reduced by 1.2 μg/m3 on average in urban and southern rural areas of beijing (si appendix, fig. s12 ). although vehicles accounted for 65.6% of local nox emissions, the concentrations of oxidants and fine particles would be consistently enhanced, with the reduction ratio of the vehicular nox emissions rising from 0% to 100%, even when voc emissions from vehicles were reduced by 100%. on the one hand, the abundant nitrate precursors in neighboring regions suppresses the effectiveness of local nox emission control. on the other hand, reducing nox emission favors the enhanced atmospheric oxidation ability to form more secondary particles, since many urban areas in china are prevailing under the voc-limited condition (36) . although reducing vehicular voc and primary pm emissions were both positive in decreasing the pm2.5 levels, unfortunately, traffic control usually leads to greater nox reduction than vocs, which goes to explain why the annual reduction of vehicular emissions has resulted in reductions in vocs that are only half (in percentage terms) the nox reductions over the past years (37) . all these points above explain why traffic control cannot mitigate the winter haze pollution in beijing currently, a point which also needs to be better explained to the public. in past years, the gradually strengthened vehicle emission controls have successfully contributed to the pm2.5 decrease in china (38) . in addition, past experiences from developed countries indicates that emission control on the continuously growing motor vehicle fleet is efficient and ultimately cost-effective to relieve the air pollution in a megacity (9, 10) . the problem for now is the imbalanced control among different source regions and air pollutants. compared to its surrounding regions, local sources in beijing reduced faster in the past years, which lead to the share of regional transport for the air pollution in beijing increasing to 76.1% on average in the winter. as for the differences in local species controls in beijing, the reduction rate of nox and vocs emissions used to be approximately 1:1, while the ratio is still large enough to increase the atmospheric oxidizing capacity under the strong voc-limited condition in winter, which is proved by the enhanced oxidants during the covid-19 lockdown (7, 39) . in addition, nh3 emissions, which are crucial for the formation of secondary inorganic aerosols, were still not effectively controlled, and the increased nh3 concentrations over past years weakened the benefits of the reduction in nitrate from nox emission control in east china (40) . therefore, the key is not judging whether traffic emission control is necessary, but accelerating voc and nh3 emissions reductions and synchronously controlling regional sources to release the benefits on local traffic emission control. targeting any of long-term air quality, climate change or street-level personal exposure, means any measures reducing vehicle emissions are going to be beneficial. to achieve pm2.5 concentration reductions in the short term, vocs and primary pm2.5 should be jointly treated as the priority pollutants to control, or nox emissions should be substantially reduced by the combination with other sources beyond vehicles so as to reach the non-linear tipping-point between changes of nox reduction and oxidants concentration. for the first option, the new emission standard for ldv (china 6) , which will be implemented in july 2020, is expected to dramatically reduce the voc emission from evaporations and will be effective to improve the air quality (41) . the challenge is to maintain popular support for mitigation policies such as reductions in traffic flow or restrictions in vehicle type which themselves lead to significant air quality improvements (pm2.5, no2) but which are not directly visible to the general population in the face of the highly visible haze events in this work, we built an integrated analysis framework to seek for the role of vehicle emissions in winter haze pollution in beijing around the covid-19 outbreak and spring festival (si appendix, fig. s13 ). first, we developed a slove model to estimate the hourly traffic emissions in urban areas of beijing. this model consists of two dynamic databases, including a) hourly road speed and b) the observed meteorological condition, and three static local traffic information databases, including a) fleet composition, b) road basic information and c) vehicle emission factors. the real-time traffic condition data through the application programming interface (api) to amap was obtained to calculate the traffic flow and emissions. detailed descriptions about this model were discussed in the si appendix, section s1. in addition, the emissions from hdts were evaluated based on a more accurate trackatruck model driven by big data (trajectory signals of each hdt from the beidou navigation satellite system), with advantages of considering individual truck differences (42) . besides the on-road emissions calculated in this study, emissions from other sectors were assembled from several recent studies to improve the precision of emission inventory, including urban anthropogenic emissions for beijing, tianjin and its surrounding 26 major cities in northern china from an air pollution prevention plan proposed by the government (referred to "2+26" plan), shipping emissions from our previous research (43) , other anthropogenic emissions in china from multi-resolution emission inventory for china (meic) model (38) and others listed in si appendix, table s2 . the emission changes of other anthropogenic sources during the transition and lockdown period were calculated respectively, based on the changes of related industrial and residential activities (si appendix, sections s2). the weather research and forecasting model-community multiscale air quality model (wrf-cmaq) was applied to simulate the air quality in beijing from january 10th to february 15th in 2020 (44, 45) . the modeling system drew on the 4-nested run with a grid resolution at 1.33 km of the innermost domain, where urban canyon model (ucm) with updated land use data and urban canyon parameters (ucps) was applied to improve the prediction of meteorological field (46) . to reproduce the polluted days, the heterogeneous reactions of so2 were incorporate into the cmaq model to enhance the sulfate formation at a high relative humidity (47, 48) . in addition, the cmaq model with the isam was used to determine the source contribution to the pm2.5 and its species concentrations before and during the covid-19 outbreak. detailed model configurations were described in si appendix, section s3. the ground meteorological observations were obtained from the national climate data center (ncdc, ftp://ftp.ncdc.noaa.gov/pub/data/noaa/) integrated surface database. the hourly air quality data in beijing, including pm2.5 and gaseous air pollutants were from the beijing municipal environmental monitoring center (bjmemc). the pm2.5 component observations were collected from the national research program for key issues in air pollution control. we evaluated the simulated no2, o3, and pm2.5 concentrations against ground-based observations. the model well captured the variations of the air quality with correlation coefficients higher than 0.5 for all species, which was in line with other recent modeling studies in beijing (49, 50) . the model performances in predicting major pm2.5 chemical components were also acceptable with the mean biases (mb) ranging from 0.3 μg/m3 to 4.3 μg/m3. to quantify the influence of vehicle emission reduction on air quality before and during the covid-19 outbreak, we set up a series of scenarios in the wrf-cmaq modeling system with different turbulences in on-road emissions while other configurations remained the same as the base scenario. the base scenario simulated the air quality with real emissions, while in s1 scenario vehicle emissions during the lockdown period were assumed to be as usual in the pre period. in s2-s6 scenarios, the assumed relative reduction of vehicle emissions changed from 0% to 100% in winter normal days (pre period). our study was subject to a few uncertainties and limitations. besides uncertainties of emission inventories and wrf-cmaq models (discussed in si appendix, section s4 and s5), the som were not considered in the source apportionment due to the limitation of the existing isam model. we probably underestimated the contribution of local vehicle emissions to pm2.5 concentrations since (a) aromatics from gasoline vehicle exhausts is a critical determinant of urban secondary organic aerosol formation and (b) synergetic oxidation of vehicular exhaust leads to efficient formation of ultrafine particles (ufps) under urban conditions (51, 52) . moreover, the on-road emissions are instantly diluted throughout the coarse grid cells of the chemical transport model in which the emissions occur. in a finer spatial scale, however, the vehicle emission would significantly affect the human exposure to air pollution due to its close proximity to human activities at a low emission height, and thus result in a serious health burden (53) . quantification of vehicular contribution to human health risk in beijing at a neighbourhood-scale based on a source-receptor model with a higher spatial resolution is necessary and suggested for future investigation. beijing 39% 39% 39% 39% 39% 39% 71% 71% 71% 71% 71% 71% beijing 9% 14% 14% 13% 14% 14% 50% 53% 74% 70% 53% 53% with the rapid development of intelligent transportation system (its) technologies, the floating car data is collected by the digital map providers (such as amap, baidu, and didi in china) from the gps of commercial taxies and the signals received from the map user's mobile devices. therefore, this kind of dynamic traffic big data can represent the real-time on-road conditions, which has been successfully used to estimate vehicle emissions in megacities with high data coverage due to a large number of users (1, 2) . based on the local traffic characteristics of beijing, we developed a street-level on-road vehicle emission (slove) model to estimate the hourly traffic emissions. first, to obtain the hourly spatial mean speed of urban roads, we accessed the real-time traffic data through the application programming interface (api) to the amap (www.amap.com) 12 times per hour to improve the temporal coverage of the data. based on the traffic function and capacity from high to low, the roads were classified into four types, including freeway, urban freeway, artery road, and local road. the data collected on monday morning rush hours in pre period and the covid-19 lockdown period were taken as an example to show the spatial distribution of traffic speeds (fig. s1 ). we found that the improvement of traffic conditions after the covid-19 outbreak was obvious due to the epidemic prevention, indicating that our data was reliable to reflect the real-world road information. the road traffic flow was then estimated based on local speed-volume models. previous studies indicated that the single-regime models from traffic engineering could be applied to describe the relationship between traffic speed and flow in urban roads of chinese cities (1) (2) (3) (4) . although there are many other advanced models, most of them are too complicated to use in a megacity like beijing. in this study, we assumed that the same speed-volume model could be applied to roads with the same road type because of similar traffic capacity. the speed-volume model in the urban freeway was established based on the observations from 45 sites with more than 200,000 samples collected by the beijing municipal commission of transport (bmct) using the remote traffic microwave sensor (rtms) (fig. s1 ). the historical monitoring data included the hourly traffic speed and flow for each day, covering not only four major ring roads but also some non-ring roads, collected from february 2018 to october 2018. the fitting performances among the underwood model, greenshields model, and van aerde model were compared (5) (6) (7) . the results showed that the van aerde model was the best one to reproduce the traffic flow for all selected roads with root mean square error (rmse) in the range of 144-325 pcu·(hour·lane)-1 (fig. s14) . for other road types lack of observations, the fitted underwood models from other local researches were used (2, 8) . the hourly traffic speed data was not available for all roads, especially for less-travelled roads outside the 5th ring road (fig. s1) . therefore, the research domain was divided into 27 traffic zones according to intersecting areas of ring roads and administrative districts, and the averaged traffic data of monitoring roads with the same road type and zones were used for non-monitored roads. (10) . the nox emission was also corrected by the real-world environmental condition collected from the national climate data center (ncdc, ftp://ftp.ncdc.noaa.gov/pub/data/noaa/) integrated surface database, using the method recommended by the motor vehicle emission simulator (moves) model (11) . the vehicle fleet composition data was different for various road types, which was collected by bmct based on the video data from typical roads in beijing (fig. s15) . we changed the proportion of hdv to zero during the covid-19 outbreak and spring festival due to two reasons: (1) all inter-provincial passenger transport into or out of beijing was prohibited to prevent the spread of virus (2) the hdv owned by private companies almost disappeared because of the spring festival holiday. the vehicle registration information of each administrative district in beijing was also used to estimate the proportion of every vehicle category classified by fuel and emission standards. in the end, the co, hc, nox, pm2.5, and pm10 emissions from vehicle exhausts for each street segments were calculated using the following formula: , , = × , × ∑( , × , ) where , , was the emission of pollutant for street link in hour, units in g/h; was the length of street link , units in km; , was the total traffic flow for street link in hour, units in veh/h; , was the proportion of vehicle category in total traffic flow for street link ; , was the emission factor of pollutant for vehicle category , which was influenced by the travel speed and meteorological condition, units in g/km. also, since vehicular evaporation emissions (vees) accounted for about 39% total hc emissions from vehicles, we quantified hc emissions from the running loss, namely the largest contributor among vees (12): where , was the evaporation emissions for street link in hour, units in g/h; was the length of street link , units in km; , was the total traffic flow for street link in hour, units in veh/h; , was the proportion of gasoline vehicles in total traffic flow for street link ; was the hc emission factor for the running loss process, which was from previous tunnel studies (13) (14) (15) , units in g/h; , was the average speed of traffic flow for street link in hour, units in km/h. we assembled emissions from multiple recent studies to ensure the accuracy of the emission inventory for the air quality simulation (table s2 ). the emissions of all hdt travels which happened in beijing was calculated using the full-sample enumeration model called trackatruck, developed in our recent study (16) . the hdt positioning data was provided by the governmental hdt monitoring platform. for each travel, trackatruck mapped the positioning data to multiple trajectories containing information about the operating modes and calculated the emissions along the trajectories. the trackatruck model can produce extremely high temporal-spatial resolution of day-to-day, hour-to-hour hdt emission maps (0.01˚ × 0.01˚). based on the hdt emissions in 2018 in our previous study, the year-on-year change in beijing's road freight volume from 2018 to 2020 was used to estimate the hdt emissions changes in periods that covered the covid-19 outbreak. the base emissions from other sources were obtained primarily from the "2+26" plan, which included urban anthropogenic emissions of 28 cities in the bth region and its surrounding areas for the year 2018. this emission inventory was developed with the unified method standard and the exhausted source categories at the district or county level. to estimate the emission reduction during the covid-19 outbreak and spring festival period, the activity level for each major sector were investigated respectively, in accordance with the period defined in fig. 1 . considering the distinctive industrial structure of beijing compared to the surrounding areas as well as its large proportion of the migrant population, we also estimated the emission reduction separately for beijing and other regions. it should be noted that, since the emissions of slove model covered only within and surrounding the 6th ring road, the non-hdt vehicle emissions outside this domain were derived from "2+26 plan" and further applied the reduction rates calculated from the slove model for the transition and lockdown periods, respectively. for emissions from the industry, the reduction rates were estimated based on the power load curve during the spring festival (19) . the historical year-round power load curve for beijing in 2018 was collected from national development and reform commission, where a drastic decline appeared during the spring festival holiday, beginning from a week before the holiday. according to the monthly electricity consumption data from the national bureau of statistics (http://data.stats.gov.cn/), the tertiary industry and residential activities showed constant electricity consumption throughout the winter, while that for the secondary industry showed a significant decrease in february. it could be thus assumed that the drastic decline of the power load curve was mostly related to the reduction of secondary industry activities. given the statistical data that the electricity consumption from secondary industry accounted for 29% of the whole society (beijing statistical yearbook, 2019), combined with the load curve before and the during the spring festival, the reduction rates of electricity consumption from the secondary industry could be estimated -39% for transition and 71% for lockdown period. as the industry of beijing is dominated by the light industry, we assumed these factors could be used as the proxies for emission reduction. residential emissions were mainly composed of cooking smokes from residents and restaurants. notably, beijing is a city with a large migrant population, different from the surrounding cities. during the spring festival, emissions from residential combustions in beijing would be reduced as people return to their home towns. thus, we used the proportion of the migrant population to the local population from the 2019 beijing statistical yearbook (35%) as the reduction rates for residential combustion emissions during the lockdown period. their emissions during the transition period were reduced by half to reflect the gradually decreased population. the cooking emissions from restaurants were assumed as usual until the covid-19 outbreak. according to the investigation of the china cuisine association, 78% of catering enterprises lost more than 100% of their business income during the epidemic period, compared to the last spring festival. in this study, we thus assumed their emissions decreased by 70% during the lockdown period. emissions from power industry were assumed constant since there are only a few local power plants in beijing and a large proportion of its electricity consumption are supported by the surrounding areas. other sectors were also considered not affected by the covid-19 and spring festival. for other regions outside beijing, the control factors from wang et al. (20) were mainly referenced, except the transportation sector, which was obtained from the slove model. table s1 summarized the emission reduction rates for all anthropogenic sources in beijing and other regions. the models used in this study for basic air quality simulation included the weather research and forecasting model (wrf, https://www.mmm.ucar.edu/weather-research-andforecasting-model) with version 3.8.1 and the community multiscale air quality (cmaq, https://www.epa.gov/cmaq) model with version 5.2, which were developed by us ncar (national center for atmospheric research) and us epa (environmental protection agency), respectively. we applied the wrf-cmaq modeling system to simulate the air quality in beijing from january 10th to february 15th in 2020, with 3 days of spin-up time to avoid the influence of the initial condition. as shown in fig. s16 , four nested domains with a horizontal resolution of 36 km×36 km, 12 km×12 km, 4 km×4 km, and 1.33 km×1.33 km, respectively, were used to improve the accuracy of simulated boundary conditions. the first guess field and boundary conditions for wrf were generated from the 6-h ncep fnl operational model global tropospheric analyses dataset. the four-dimensional data assimilation (fdda) was enabled using the ncep adp global surface and upper air observational weather data the cmaq model was configured to use the cb05 gas-phase mechanism and the aero6 aerosol module with aqueous chemistry. the single-layer urban canopy model (slucm) coupled with the noah land surface model was applied in the wrf model to improve the prediction of the meteorological fields. since slucm can represent the urban geometry by assuming infinitely-long street canyons, the thermal and dynamic effects of urban areas are considered, including the shadowing, trapping, and multiple reflections of solar radiation, canopy flows and anthropogenic heat (27) (28) (29) . many previous researchers have successfully used it to improve the fine-gridded simulated surface temperature and wind in a big city (30) (31) (32) (33) (34) . furthermore, a high-resolution land use data (250m) for 2014 from finer resolution observation and monitoring-global land cover (from-glc, http://data.ess.tsinghua.edu.cn) was provided for the urban canyon model to define a more realistic underlying surface, especially for urban areas which were reclassified based on the fraction of impervious surface (fig. s17-a) . considering the heterogeneous spatial distribution of urban morphological characteristics, the gridded urban canyon parameters (ucps) were also applied in slucm, as shown in fig. s17 -b. the basic building data obtained from amap, including the number of floors and footprint outline for each building, were used to create the ucps database following the formulations in he, et al. (34) and burian, et al. (35) . considering the non-linear relationship between air pollutant concentrations and emissions, the cmaq version 5.0.2 with the integrated source apportionment model (isam) was applied to determine the source contribution to the ambient pm2.5 and its species concentrations before and during the covid-19 outbreak (36) . the same meteorological field, emissions and configurations were used as described in last paragraphs. in addition, due to limited computational ability, we divided the emissions into five groups to trace them separately in the isam model, including emissions from local sources of beijing (mobiles, industry, domestic and other local sources) and emissions from regional sources outside beijing. besides the local mobile source referred to the on-road emissions calculated in this study, other local sectors were classified according to the meic emission model. the "other" local sources mainly included emissions from agriculture, off-road traffic and open burning. all chemical components available in the isam were tagged, including primary organic matter (pom), element carbon (ec), sulfate, nitrate, ammonium and other nonreactive components. however, due to the limitation of the existing isam model, the secondary organic matter (som) were not considered in the source apportionment. furthermore, the heterogeneous reaction of so2 was incorporate into cmaq models with both version 5.2 and 5.0.1 to enhance the sulfate formation, respectively, considering the significant contribution of sia (secondary inorganic aerosols) formation to fine particles during the heavy pm2.5 pollution in northern china, especially for beijing (37) (38) (39) (40) (41) (42) (43) . in this study, the chemistry was parameterized using a pseudo-first-order rate constant as following (44, 45) : where was the heterogeneous rate constant for species (s-1), was the total aerosol surface area in the aitken and accumulation mode (m2), was the reactive uptake coefficient, was the thermal velocity (m s-1). the uptake coefficients for so2 heterogeneous reaction, a key parameter to determine the reaction rate, were from fu, et al. (46) , which increased rapidly with the growth of ambient relative humidity (rh), especially when rh was higher than 0.5: where (6×10-5) was the uptake coefficient under the dry condition. the wrf model performance was evaluated against ground-level observations in four major meteorological parameters including surface wind speed (ws), surface wind direction (wd), surface temperature (t) and relative humidity (rh). the meteorological observations in beijing at every 1-h were obtained from the national climate data center (ncdc, ftp://ftp.ncdc.noaa.gov/pub/data/noaa/) integrated surface database. the benchmarks suggested by previous research was used to judge the meteorological performance (the mean biases (mb) ≤ ±0.5 k for t，mb ≤ ±0.5 m/s for ws and mb ≤ ±10° for wd) (47) . as shown in fig. s18 and table s3 , the simulated surface winds are correlated well with observations with mb in the range of related recommendations. high correlation coefficients (r, 0.88, and 0.94) for surface temperature and humidity proved that the model performances were acceptable, although the mb of t was a litter higher than the suggested goal. we also estimated the model performance of the revised cmaqv5.2 in predicting the no2, o3, pm2.5, and its chemical compositions' concentrations by comparing the modeling results with observations at 34 air quality monitors and three pm2.5 composition monitors of beijing (locations were shown in fig. s16 -b), as described in table s4 . the real-time hourly air quality data (including no2, o3, and pm2.5) based on thermo scientific samplers and analyzers, were obtained from the beijing municipal environmental monitoring center (bjmemc, http://zx.bjmemc.com.cn/). in addition, five dominant chemical components of pm2.5 including element carbon ( ), organic matter ( ), sulfate ( 4 2− ), nitrate ( 3 − ) and ammonium ( 4 + ) were discussed in detail, and observations were collected from the national research program for key issues in air pollution control. in general, the model can capture the temporal and spatial variations of the air quality with correlation coefficients (r) higher than 0.5 of all species. the simulated hourly pm2.5 was well agreed with observations, with the overall model performance within the performance criteria suggested by boylan and russell (48) (mean fractional bias (mfb) ≤ ±60% and mean fractional error (mfe) ≤ ±75%). the o3 concentrations were slightly underestimated, mainly due to uncertainties in emission inventory and unavoidable deficiencies during meteorological and air quality simulation. table s4 showed their detailed performance statistics. the simulation errors for each species were relatively small, with mbs ranging from 0.3 μg/m3 to 4.3 μg/m3. however, all species were a little overestimated with nmbs (normalized mean bias) ranging from 7.3% to 32.2%, especially for clean days (a day with 24-hour averaged pm2.5 concentration lower than 75 μg/m3). since ec was mainly from the primary emissions, the uncertainties in emissions inventory and meteorological field were probably responsible for this overestimation. during the heavily polluted days, the revised model slightly overestimated the moreover, we quantified the differences in pm2.5 and its species concentration between modeling results from cmaqv5.0.1-isam model with the observations (table s4 ). in general, the simulation error of pm2.5 was also within the recommended criteria (mfb ≤ ±60% and mfe ≤ ±75%), and the performances in predicting ec and sna (sulfate, nitrate and ammonium) were both similar with those of cmaqv5.2, with a small mbs ranging from 0.3 μg/m3 to 3.6 μg/m3. however, due to the lack of additional som formation pathways included in cmaqv5.2 (e.g. aging of s/ivoc and primary organic aerosol), the om was underpredicted with a nmb of -62.6%, leading to a lower estimated pm2.5 concentration (49, 50) . as mentioned in section s3, since the som formation was not traced in the isam model due to the existing limitation, the underestimated som would not affect the source apportionment results. in sum, the predicted pm2.5 and its chemical components concentrations in both cmaqv5.2 and cmaqv5.0.1-isam showed a acceptable agreement with observations, which provided confidence in the source contribution as described in the results section of main text. the uncertainties of this research are mainly from the estimation of emission inventories and the wrf-cmaq modeling system, as discussed below. 1 ) vehicle emission inventories. the on-road emission inventory estimated by the slove model was subject to a few inherent uncertainties and limitations, such as the systematic error of local speed-volume models and the emission factors chosen. in addition, hc emissions from other vees processes, including refueling loss, hot soak loss and diurnal loss, were not calculated in this study due to lack of related activity data. however, the traditional uncertainty estimation method (e.g. monte-carlo method) was not available for methods based on big data in this study due to large burden on computing time. therefore, we compared our results with vehicle emissions from other previous studies using a similar bottom-up method (fig. s6) (51), probably resulting in an overestimation of emissions. however, these uncertainties were relatively minor during the covid-19 lockdown period, as most of the industrial activities are largely limited and the emission was reduced to a low level. 3) wrf-cmaq model. although we have applied slucm in the wrf model and incorporated the heterogeneous reactions of so2 into the cmaq model to improve the predictions during haze pollution, there were still inevitable uncertainties due to the chemical processes and physical parameters. in addition, the sensitivity analysis by setting emission reduction scenarios in the wrf-cmaq model may result in uncertainties of the o3 concentration due to its nonlinear relationship with its gaseous precursors. however, the model performances have been proved to be reliable in section s4, which was in line with other recent modeling studies in beijing (52, 53) . 4) isam model. due to the limitation of the existing isam model, voc emissions could not be tagged to trace the formation of som in the source apportionment. while the som was one of the major pm2.5 components in winter of beijing, especially with a rapid growth during heavy pollution days ( fig. 2-a) . therefore, we probably underestimated the contribution of vehicle emissions to pm2.5 concentrations since (a) aromatics from gasoline vehicle exhausts is a critical determinant of urban secondary organic aerosol formation (54) and (b) synergetic oxidation of vehicular exhaust leads to efficient formation of ultrafine particles (ufps) under urban conditions which was also not included in this model (55) . quantification of vehiclar contribution to soa and ufps is necessary and suggested for future investigation. s19 . evaluation of simulated hourly pm2.5 chemical composition concentrations against ground-based observations. the dashed lines correspond to the 1:1, 1:2, and 2:1 lines, respectively. heavily polluted days stand for a day with 24-hour averaged pm2.5 concentration higher than 75 μg/m3, and clean days refer to other periods. note: the comparison for wind direction was corrected with the consideration of the periodic nature of wind. traffic operation of the whole city was stable and orderly during the spring festival holiday of an investigation of transmission control measures during the first 50 days of the covid-19 epidemic in china a preliminary assessment of the impact of covid-19 on environment -a case study of china air pollution reduction and mortality benefit during the covid-19 outbreak in china. the lancet planetary health covid-19 restrictions and ease in global air pollution dominant role of emission reduction in pm2.5 air quality improvement in beijing during 2013-2017: a model-based decomposition analysis beijing (2020) beijing ecology and environment statement comparison of air pollutant emissions among mega-cities on-road vehicle emission control in beijing: past, present, and future on-road vehicle emissions and their control in china: a review and outlook amap (series report: china urban traffic analysis report high-resolution mapping of vehicle emissions of atmospheric pollutants based on large-scale, real-world traffic datasets development of a vehicle emission inventory with high temporal-spatial resolution based on nrt traffic data and its impact on air pollution in beijing -part 1: development and evaluation of vehicle emission inventory the impacts of firework burning at the chinese spring festival on air quality: insights of tracers, source evolution and aging processes a novel technique for quantifying the regional component of urban aerosol solely from its sawtooth cycles exploring the severe winter haze in beijing: the impact of synoptic weather, regional transport and heterogeneous reactions the nitrate radical: physics, chemistry, and the atmosphere variability in nocturnal nitrogen oxide processing and its role in regional air quality the formation, properties and impact of secondary organic aerosol: current and emerging issues a wrf-chem model study of the impact of vocs emission of a huge petro-climate change cmaq: the community multiscale air quality modeling system establishment of urban morphology database and its effect on meteorology and air quality simulation impacts of stabilized criegee intermediates, surface uptake processes and higher aromatic secondary organic aerosol yields on predicted pm2.5 concentrations in the mexico city metropolitan zone modeling analysis of secondary inorganic aerosols over china: pollution characteristics, and meteorological and dust impacts contributions of trans-boundary transport to summertime air quality in beijing, china impacts of improved modeling resolution on the simulation of meteorology, air quality, and human exposure to pm2.5, o3 in beijing gasoline aromatics: a critical determinant of urban secondary organic aerosol formation remarkable nucleation and growth of ultrafine particles from vehicular exhaust traffic-related air pollution near busy roads development of the real-time on-road emission (roe v1.0) model for streetscale air quality modeling based on dynamic traffic big data high-resolution mapping of vehicle emissions of atmospheric pollutants based on large-scale, real-world traffic datasets road traffic noise mapping in guangzhou using gis and gps development of a vehicle emission inventory with high temporal-spatial resolution based on nrt traffic data and its impact on air pollution in beijing -part 1: development and evaluation of vehicle emission inventory speed, volume, and density relationships a study of traffic capacity. highway research board proceedings single regime speed-flow-density relationship for congested and uncongested highways multi-level fundamental diagram of traffic flow for beijing based on multi-source data (in chinese) technical guide of air pollutant emission inventory for on road vehicles (trial) copert: a european road transport emission inventory model emission adjustments for temperature, humidity, air conditioning, and inspection and maintenance for on-road vehicles in moves2014 an updated emission inventory of vehicular vocs and ivocs in china heavy-duty diesel vehicles dominate vehicle emissions in a tunnel study in northern china emission factors of volatile organic compounds (vocs) based on the detailed vehicle classification in a tunnel study decadal changes in emissions of volatile organic compounds (vocs) from onroad vehicles with intensified automobile pollution control: case study in a busy urban tunnel in south china a big data approach to improving the vehicle emission inventory in china health and climate impacts of ocean-going vessels in east asia the impact of the "air pollution prevention and control action plan" on pm2.5 concentrations in jing-jin-ji region during 2012-2020 hourly disaggregation of industrial co2 emissions from shenzhen severe air pollution events not avoided by reduced anthropogenic activities during covid-19 outbreak. resources, conservation and recycling impact of cloud microphysics on the development of trailing stratiform precipitation in a simulated squall line: comparison of one-and two-moment schemes the kain-fritsch convective parameterization: an update radiative forcing by long-lived greenhouse gases: calculations with the aer radiative transfer models coupling an advanced land surface-hydrology model with the penn state-ncar mm5 modeling system. part i: model implementation and sensitivity parameterization of orography-induced turbulence in a mesobeta--scale model the step-mountain eta coordinate model: further developments of the convection, viscous sublayer, and turbulence closure schemes a simple single-layer urban canopy model for atmospheric models: comparison with multi-layer and slab models coupling a single-layer urban canopy model with a simple atmospheric model: impact on urban heat island simulation for an idealized case the integrated wrf/urban modelling system: development, evaluation, and applications to urban environmental problems a numeric study of regional climate change induced by urban expansion in the pearl river delta air quality modelling in the berlin-brandenburg region using wrf-chem v3.7.1: sensitivity to resolution of model grid and input data impact of an improved wrf urban canopy model on diurnal air temperature simulation over northern taiwan modeling impacts of urbanization and urban heat island mitigation on boundary layer meteorology and air quality in beijing under different weather conditions high-resolution dataset of urban canopy parameters for beijing and its application to the integrated wrf/urban modelling system development and assessment of the second generation national building statistics database implementation and evaluation of pm2.5 source contribution analysis in a photochemical model implementation of dust emission and chemistry into the community multiscale air quality modeling system and initial application to an asian dust storm episode heterogeneous chemistry: a mechanism missing in current models to explain secondary inorganic aerosol formation during the january 2013 haze episode in north china trends of pm2.5 and chemical composition in beijing roles of regional transport and heterogeneous reactions in the pm2.5 increase during winter haze episodes in beijing characteristics of pm2.5 and sna components and meteorological factors impact on air pollution through 2013-2017 in beijing nitrate dominates the chemical composition of pm2.5 during haze event in beijing parameterization of heterogeneous reaction of so2 to sulfate on dust with coexistence of nh3 and no2 under different humidity conditions secondary particle formation and evidence of heterogeneous chemistry during a wood smoke episode in texas impacts of stabilized criegee intermediates, surface uptake processes and higher aromatic secondary organic aerosol yields on predicted pm2.5 concentrations in the mexico city metropolitan zone modeling analysis of secondary inorganic aerosols over china: pollution characteristics, and meteorological and dust impacts enhanced meteorological modeling and performance evaluation for two texas ozone episodes. prepared for the texas natural resource conservation commission pm and light extinction model performance metrics, goals, and criteria for three-dimensional air quality models understanding sources of organic aerosol during calnex-2010 using the cmaq-vbs semivolatile poa and parameterized total combustion soa in cmaqv5.2: impacts on source strength and partitioning notice of the state council on issuing the three-year action plan for winning the blue sky defense battle contributions of trans-boundary transport to summertime air quality in beijing, china impacts of improved modeling resolution on the simulation of meteorology, air quality, and human exposure to pm2.5, o3 in beijing gasoline aromatics: a critical determinant of urban secondary organic aerosol formation remarkable nucleation and growth of ultrafine particles from vehicular exhaust mix: a mosaic asian anthropogenic emission inventory under the international collaboration framework of the mics-asia and htap model of emissions of gases and aerosol from nature version 3 (megan3) for estimating biogenic emissions h.l and z.l. designed this research and performed analysis. z.l. and x.w wrote the paper. all authors took contributions to discussing and improving this research. key: cord-298953-9aifql2f authors: day, brett h. title: the value of greenspace under pandemic lockdown date: 2020-08-04 journal: environ resour econ (dordr) doi: 10.1007/s10640-020-00489-y sha: doc_id: 298953 cord_uid: 9aifql2f the covid-19 outbreak resulted in unprecedented restrictions on citizen’s freedom of movement as governments moved to institute lockdowns designed to reduce the spread of the virus. while most out-of-home leisure activities were prohibited, in england the lockdown rules allowed for restricted use of outdoor greenspace for the purposes of exercise and recreation. in this paper, we use data recorded by google from location-enabled mobile devices coupled with a detailed recreation demand model to explore the welfare impacts of those constraints on leisure activities. our analyses reveals evidence of large-scale substitution of leisure time towards recreation in available greenspaces. indeed, despite the restrictions the economic value of greenspace to the citizens of england fell by only £150 million over lockdown. examining the outcomes of counterfactual policies we find that the imposition of stricter lockdown rules would have reduced welfare from greenspace by £1.14 billion. in contrast, more relaxed lockdown rules would have delivered an aggregate increase in the economic value of greenspace equal to £1.47 billion. as the covid-19 pandemic swept across the planet, national governments instituted various rules designed to reduce human contact and slow rates of infection. the severity of these lockdown rules differed from nation to nation, largely mirroring the severity of the virus outbreak. this paper focuses on england, whose own lockdown experience began on 23rd march, 2020. the lockdown in england placed unprecedented restrictions on citizen's freedom of movement. as well as not being able to go to their places of work, citizens were deprived of access to most shops, food and drink outlets, entertainment establishments and leisure facilities. one of the few privileges that remained was the opportunity to spend time outdoors walking and exercising, activity often undertaken in greenspace. this paper presents an empirical exploration of the levels of engagement with greenspace over the lockdown in england. it focuses on the question of how greatly the lockdown rules impacted on the value flows realised by english citizens from their greenspace and explores how those impacts might have differed had stricter or more relaxed restrictions been imposed. a priori, it is not self-evident whether the value derived from greenspace as a focus for outdoor recreation was diminished or amplified by the rules of lockdown and the conditions of the covid-19 outbreak. on the one hand, citizens may have reduced their use of greenspaces in an effort to minimise their risks of exposure to the virus. likewise, lockdown rules prevented citizens from visiting all but highly local greenspaces. limiting citizens' options to a small set of potentially less-desirable destinations will again have acted to dampen demand. on the other hand, under lockdown, citizens were unable to participate in nearly all other forms of out-of-home leisure activity, demand for greenspace may have increased as citizens substituted away from those unavailable alternative uses of their leisure time. in addition, under lockdown many citizens were unable to work. releasing the usual-leisure time constraints on those individuals will also have acted to increase demand for outdoor recreation. as the lockdown unfolded, localised evidence of changing behaviour arose. newspaper reports described normally busy beaches as all but deserted (betts 2020; crane 2020; ikonen 2020) . in contrast, incidents of overcrowding in city greenspaces resulted in temporary closures of several large urban parks (including london's brockwell park and victoria park as well as middlesbrough's stewart park). in this paper we make use of data collected by google from location-enabled mobile devices which provides systematic evidence on the rates of visitation to greenspace across the regions of england over the course of the lockdown (google 2020). as described in sect. 3, this google mobility data reveals that demand for greenspace changed over the course of the lockdown in ways which mirror the evolving rules on outdoor activity. the second key resource used in this paper is the outdoor recreation valuation (orval) model (day and smith 2017) , which we use not only to predict demand for visits to greenspace under the restrictive rules of the lockdown but also to estimate the changes in economic value experienced by residents of england as a consequence of those rules. developed in partnership with the uk government, 1 orval is underpinned by an econometric model estimated in the random utility framework. as such, orval follows in a tradition stretching back at least as far as kocur et al. (1979) and feenberg and mills (1980) , what distinguishes orval from other such models is that it is, as far as we are aware, the first to consider the entire range of publically-accessible greenspace sites including parks, gardens, playing fields, church yards, cemeteries, allotments, nature reserves, woodlands, wetlands, river and lakeside walks, beaches and the network of coastal and countryside paths. we briefly review the orval model in sect. 4. of course, orval was estimated on data in which individuals were not concerned with risking exposure to a deadly virus, in which their pursuit of alternative leisure activities was unrestricted and where they faced the leisure-time constraints of normal working conditions. in this paper, we assume that differences between the orval predictions of recreation behaviour under the lockdown rules and those observed in the google mobility data are the net result of those, and possibly other, factors. as described in sect. 5, we undertake a novel statistical exercise in model calibration using techniques of latent class regression to estimate parameters for the orval model which capture the net effect of those factors on recreation behaviour. those estimates allow us to construct a times series of orval predictions for recreation activity under the rules of the lockdown that can be contrasted to a counterfactual in which covid-19 had not come to pass. we present the findings from that comparison in sect. 6. in brief, we find that while the lockdown imposed very significant restrictions on outdoor recreation activities, citizens engaged in substantial compensating substitution behaviour. the mitigating effect of that substitution behaviour meant that over the lockdown, citizens of england experienced only a 2.1% fall in the welfare they might otherwise have enjoyed from greenspace, an amount equating to a loss in aggregate economic value of £150 million. our calibration of the orval model allows us to explore other counterfactuals; namely, how engagement with the outdoors might have proceeded through the covid-19 outbreak under stricter or under more relaxed lockdown rules. not surprisingly, we find that in the strict-rules counterfactual welfare from greenspace is £1.14 billion lower than under the actual lockdown rules. in contrast, applying less strict lockdown rules on outdoor recreation allows for even greater use of the outdoors and delivers an aggregate welfare benefit of £1.47 billion. this paper's contribution is primarily empirical. it attempts to quantify the impact of the covid-19 pandemic and its associated lockdown on one aspect of a nation's everyday life; outdoor recreation in greenspace. not surprisingly, given the recency of the events, little exists in the published literature with a similar intent. an unpublished manuscript by venter et al. (2020) examines changes in outdoor activity in oslo, norway during the virus outbreak. using data on the route choices of runners and cyclists, they find that spatial patterns of exercise activity changed over lockdown to favour greener and more remote locations. through a calibration exercise, venter et al. estimate that outdoor recreation activity in oslo increased by291%. in another yet to be published manuscript, rice and pan (2020) explore data made publically available by google on the use of greenspace during the covid-19 pandemic, data that we also exploit in our study (google 2020) . focusing on 111 counties in the western united states, they identify an average 2.5% increase in greenspace visitation and find that differences across counties are chiefly explained through differences in weather. 2 our study differs from these other contributions in a number of ways. the focus of our study is england, where lockdown rules on recreation were not dissimilar to those in the western us but significantly stricter than in oslo. 3 rather than routes used for exercise we explore visits to greenspace. and unlike both venter et al. (2020) and rice and pan (2020) , our focus is not primarily on how recreation patterns changed over space, but how they responded to changes in lockdown rules. perhaps the clearest point of separation is that we are the first to attempt to attribute an economic value to the changes in greenspace use that arose over the lockdown. the english lockdown began on march 23rd, 2020 with non-essential workers asked to work from home. shops and entertainment outlets were forced to close unless selling essential items and travel was only allowed if absolutely necessary. our particular interest concerns the rules on outdoor recreation for which specific guidelines were issued people were expected to use open spaces near to their homes and encouraged to limit themselves to one trip a day. driving to open spaces for the purposes of outdoor recreation was not allowed (hc deb 24th march 2020). requirement to abide by these measures was passed into law under the uk coronavirus act (2020) giving police the authority to issue fines of up to £960 to those that did not comply. after seven weeks of strict lockdown rules in the uk, outdoor recreation was amongst the first areas of daily life to experience a loosening of restrictions. in his televised speech to the british public on 10th may 2020, the british prime minister stated that, "we want to encourage people to take more and even unlimited amounts of outdoor exercise. you can sit in the sun in your local park, you can drive to other destinations, you can even play sports" (johnson 2020) . it was not until the middle of june that restrictions began to be lifted more generally. our analysis runs through to 15th june when many retail shops and public-facing businesses were allowed to re-open to the public. evidence regarding the impact of the lockdown rules on the use of greenspace is provided by community mobility reports (google 2020) . using data from mobile devices running google software enabled for location reporting, the mobility reports record changes in engagement in different activities over the lockdown period. the data is presented as a daily time series by region and records the percentage change in visits to numerous types of destination. our focus is on the data provided on trips to parks which google describe as including locations such as national parks, marinas, public beaches, dog parks, plazas and public gardens. google also comment that the parks data does not include visits to "the general outdoors found in rural areas" (google 2020) . this paper uses the google time series for 86 regions in england spanning the period 15th february to 7th june 2020. 4 , 5 each data point in a time series indicates the park 4 the regions used by google are level 2 administrative areas identified by the gadm (release 3.6) datab a se (global administrative areas 2020), which aligns very closely to counties and unitary authorities. 5 regional time series are not always complete. data points are missing when the numbers engaged in an activity on that day fall below google's privacy threshold such that there is insufficient data to ensure anonymity. no further information is provided by google on this censoring process and in the analyses that follow, we do not attempt to correct for the absence of these data points. visitation observed on that day relative to activity levels observed in that region over a baseline period. the baseline period used by google is the five weeks from 3rd january to 6th february 2020. in particular, a data point shows the percentage difference in visitations on that day relative to the median visitation observed for that same day of the week over the baseline period. throughout this paper we refer to that measure as one of park visitation change. the time series for england as a whole is shown in the top panel of fig. 1 , overlain with a smooth plot showing the central trend of the time series over the period. observe that the visitation change data initially oscillates around an average value of 6.1%. in other words, the park visitation measured by google over the period before the lockdown was around 6.1% higher than that measured over the baseline period. the impact of the enforcement of a strict lockdown on 23rd march appears to leave a clear signal. over the seven weeks from 23rd march through to 10th may visitation change falls to around 17% of baseline levels. likewise the relaxing of lockdown measures around 10th may, including the sanctioning of driving for engagement with the outdoors, coincides with a sharp upswing in parks visitation. on average over that last period of the time series visitation change is around 37% above baseline levels. on first examination, the google data appear to support the notion that outdoor recreation patterns in england were significantly affected by the lockdown rules. google, however, caution against over-interpretation of the raw data (google 2020) . the baseline for the data (3rd january to 6th february 2020) was chosen as a period before widespread disruption from covid-19. even without the disruption of covid-19 and the lockdown, we would expect outdoor recreation patterns to change from the winter months of the baseline to the spring and summer months of the lockdown. the central and bottom panels of fig. 1 over-plot the park visitation change time series with temperature and rainfall data for england. 6 on both panels, a smooth of the weather data is provided to identify the central trend. figure 1 reveals that the beginning of lockdown on march 23rd coincided with a well-defined change in the weather in england. after a very wet february and early march, the uk entered a prolonged dry spell. temperatures also began to increase, starting in the low tens at the beginning of lockdown and climbing to the low twenties by the end of may. a reasonable expectation might be that outdoor recreation would increase with that warmer and dryer weather, an expectation that runs contrary to the sharp fall observed in the park visitation change time series at the beginning of the lockdown. after an initial sharp fall, the visitation change data assumes a general rising trend that mirrors the rising temperature across england. it would be reasonable to assume that at least part of the differences in visitation seen over this period are attributable to the improving weather. in a similar vein, it is evident that visits respond to particular weather events. down spikes in the google data can be seen to coincide with significant rain events. likewise some of the peaks in the visitation data appear to correlate with spells of warm weather. in this paper, we take the patterns of change as suggesting that the story of greenspace use under lockdown in england can be broadly characterised as consisting of two distinct periods; • strict lockdown rules (23rd march to 10th may) over the first period of lockdown the restrictions on the use of greenspace will have exerted downward pressure on recreation activity. we expect also that behavioural adjustments to avoid infection over this period will have further reduced demand relative to normal activity levels. the upward trend in visitation change after the initial sharp fall, may reflect improving weather conditions. • relaxed lockdown rules (11th may to 15th june) entering this second period of lockdown, two things changed. first the rate of new cases had begun to fall, suggesting that england was past the peak of the virus and that the risk of infection was now falling, perhaps more significantly restrictions on outdoor recreation were lifted. both those factors will have acted to increase visitation to outdoor greenspace. that these increases in visitation are so substantial suggests that demand for greenspace may also have been inflated by the lack of alternative uses of leisure time coupled with a large segment of the population being freed from the time constraints of their normal working conditions 7 the orval model is underpinned by the orval greenspace map, a detailed spatial dataset that describes the location and characteristics of accessible greenspace across england (day and smith 2016) . the orval greenspace map identifies some 128,295 greenspace sites in england that could form the focus of a recreational trip. each recreation site is described by its physical characteristics including its dimensions, landcovers, designations and points of interest. data to estimate the orval model was provided by the monitor of engagement with the natural environment (mene) survey (natural england 2017). collected for the purposes of uk government national statistics, the mene survey provides a large, representative and random-location sample of adult (over 16 years of age) residents of england. the survey records trips to greenspace taken by each respondent over the seven days prior to the interview. for one randomly selected trip, the focus trip, the survey elicits detailed information including the location of the site visited and the mode of travel used to reach that destination. the mene survey runs throughout the year, sampling at least 800 respondents each week ensuring the data is temporally representative. orval was estimated from seven waves of data from 2009/10 through to 2015/16. in estimating orval, the destinations of focus trips in the mene data were matched to the orval greenspace map and choicebased sampling used to draw 78,154 observations for the purposes of model estimation. our econometric estimation corrects subsequently for the nature of the sample selection rule (manski and lerman 1977). 8 given the nature of the mene data, the orval model progresses from the assumption that each day represents a recreation choice occasion on which individuals can select from a choice set comprising (1) not taking an outdoor trip, and then (2) an option for traveling to each site by car and (3) an option for each site visited on foot. as such, our econometric model takes the form of a repeated discrete-choice recreation demand model (morey et al. 1993; breffle and morey 2000) where the repetition is over recreation decisions each day and the discrete choice is the decision over which of the options to select from the choice set. one significant complication in estimating a recreation demand model for all recreation possibilities across an entire nation is the size of the choice set. in estimating the orval model we make use of techniques of importance sampling to select a choice set for each individual that provides us with reasonable power in identifying the parameters of the model (guevara and ben-akiva 2013). our subsequent estimating procedures make corrections for choice-set sampling (daly et al. 2014) . following standard practice the orval model is constructed from a linear specification of conditional indirect utility functions (mcfadden 1973) . for the option of not taking a trip to an outdoor recreation area (alternatively, to choose the outside good) utility is assumed to be a function of an individual's characteristics (e.g., age, ethnicity, dog ownership, gender) the features of the particular day (e.g., the weather, time of the year, day of the week) and a set of spatial fixed effects defined by administrative regions at the level counties, unitary authorities and london boroughs. more formally, the utility of the outside good, labelled option 0, for person i on day t , is given by; where v i0t is the modelled part of utility which is taken to be a linear function of the factors assumed to influence choice of the outside good, labelled x it , and a set of parameters, 0 . finally, i0t is an econometric error term. a similar formulation is used to characterise options where recreation is chosen. these options are two-dimensional; they comprise both the choice of a greenspace destination and a mode of transport. in the orval model we assume that the utility from a site-mode combination is driven by two main factors; that site's characteristics including its landcover (e.g., woodland, natural grass, saltmarsh), designations (e.g., national park, country park, nature reserve), points of interest (e.g., archaeological remains, historic buildings, playgrounds, car parking facilities) and, second the costs that the individual incurs in travelling to that site by a particular transport mode. in orval those calculations are expressed as a monetary travel cost, tc ijq ; that is to say the combined costs in time and money that footnote 8 (continued) number of observations in each category. following manski, and lerman (1977) we correct for choicebased sampling through reweighting observation in the log-likelihood where the weight for observations in a category are simply the ratio of the population share making that choice to the same share in the sample. individual i incurs in traveling to site j using mode q (i.e. car or walk). 9 accordingly our model of site-mode utility is given by; where v ijqt is modelled utility for a site-mode option which is a linear function of a vector of site characteristics, labelled z j , associated with a set of parameters, 1 . utility is also determined by the travel costs of that site-mode option, tc ijq with associated parameter interpretable as the marginal utility of income. again, ijqt is an econometric error term. our estimating equations follow from the choice of distribution for the error terms, i0t (∀i, t) and ijqt (∀i, j, q, t) . in the orval model we assume those errors are draws from a distribution in the generalised extreme value (gev) family (mcfadden 1978) . more specifically, we assume that the errors are independent over individuals (i) and time (t) while allowing for the possibility of correlation in error terms across site-mode options belonging to the same, pre-defined similarity group. in orval, those similarity groups are identified by mode of transport (i.e. car, walk), the type of recreation site (i.e. park, path, beach) and the landcovers and land uses characterising a site (i.e. agriculture, allotment, church yard, moors and heath, natural grass, coastal, woods, wetlands, managed grass and fresh water). site-mode options can be members of more than one group, with the degree of membership of an option in a landcover group being determined by the proportion of a site's area under that landcover. a final, single-member group contains the outside option. those particular assumptions lead us to the cross-nested logit model specification (bierlaire 2006) in which the probability of a particular mode-site option is given by; here p ijqt represents the probability that person i , chooses to visit site j using mode q in time period t . in eq. (3) similarity groups are indexed by n = 1, 2, … , n , jqn identifies the pre-determined membership of site-mode option j, q to similarity group n and n (n = 1, 2, … , n) are parameters that capture the level of correlation in error terms for members of group n. equation (3) can be developed into a likelihood function for the observed choices and the model parameters, 0 , 1 , and estimated through methods of maximum likelihood. a full description of the development of the orval model, the parameter estimates and robustness testing is available in day and smith (2017) . 10 (2) travel costs for driving and walking are calculated from each respondent's home address to each site through the ordnance survey's detailed road and path network for the uk using state-of-the-art optimal routing algorithms (dibbelt et al. 2016) . fuel consumption while driving was estimated using formulae provided by the uk department of transport (2014) for an average family car and converted to a cost by multiplying by the price of fuel current in the respondent's region in the month in which they were surveyed. driving and walking times were converted into costs following guidelines on the valuation of travel time provided by the department of transport (2015). 10 note that while day and smith (2017) is a report to the uk government, the orval model was developed under expert oversight and subjected to academic review (see research project website: defra 2018). given it is based on a spatially and socioeconomically representative sample, orval can be used in exercises predicting recreation activity for the population of england. estimating visits is relatively straightforward. given an individual's characteristics and their travel costs for each site-mode option, eq. (3) can be used to predict the probability of them visiting some particular site using a particular transport mode on a particular day. in the analyses we present later, our focus is on predicting the number of visits to a region over a particular period of time. to estimate that for an individual using the orval model, one would simply sum the daily probabilities of visiting a site in that region where the probabilities would differ from day to day over that period on account of changing weather, day of the week and month of the year. to estimate total visits to the region over that period one would sum the result of that calculation for all adult residents of england. the predictions reported in this paper make a number of simplifications to that calculation both to account for the availability of data and to manage the magnitude of the calculation task. first, our predictions are based on the populations of small-area statistical areas named lower super output areas (lsoas) in england. the socioeconomic characteristics of lsoa residents was taken from the 2011 census and augmented with 2016 population estimates. we identify the population in each lsoa falling into 8 discrete groups defined by two key drivers of recreation engagement; socioeconomic segment and dog ownership. 11 taking averages of other sociodemographics, allows us to calculate daily visitation probabilities by group and lsoa. to enable comparison with the observed google mobility data, we require orval visitation predictions not only for the period of lockdown under both strict and relaxed rules but also for the period used as a baseline for the google data; a total of 120 days. a second simplification we adopt in our analyses is to group days into categories and only estimate visitation probabilities for each category. in particular, we categorise days according to month and whether they fall on a weekday or a weekend. our prediction period spans 6 months giving a total of 12 such day-month categories. in making visitation predictions we then use the met office daily weather data (see sect. 3.2) to calculate the average weather experienced in each lsoa for every day-month category. our most disaggregate visitation probabilities, therefore, constitute predictions for each day-month category from a socioeconomic group in an lsoa to a recreation site. aggregation to regional visit estimates on a particular day-month combination proceeds through a number of steps. first, for each socioeconomic group in an lsoa, we sum the visitation probabilities for that day-month combination across all sites in a region. multiplying up by each group's population in that lsoa and summing provides an estimate of visitation from that lsoa to the region. repeating those calculations across each of the 32,844 lsoas in england and summing the results provides orval's estimate of visits to a region. since we will have cause to refer to this calculation later, a more formal presentation is given by; where v gmd is the orval estimate of visits to region g on the particular day-month combination given by the index md where m indexes months and d ∈ {weekday, weekend} ; r indexes lsoas while s indexes the set of socioeconomic groups, such that n s r is the number of individuals in group s living in lsoa r ; c g is the set of site-mode options in region g and p s r jqmd is the orval estimate of the group-day-month probability of visiting site j by transport mode q. one useful property of gev models is that there exists a simple closed-form expression for the expectation of the maximum utility a respondent might expect to derive from being able to choose an option from their choice set. in the case of the cross-nested logit model that expression amounts to; where w it (c) is the expectation of maximum utility realised by individual i in time period t given the opportunity to choose from the set of site-mode options in the choice set c , and is the euler-mascheroni constant (that takes a value of approximately 0.5772). it follows that the expected level of welfare change that an individual would experience if the nature of their choice set were to change can be estimated from (small and rosen 1981) ; where c is the original choice set and c ′ is the changed choice set. in simple terms, eq. (6) describes the analyst's best estimate of how an individuals' utility will change as a result of changes in the choice set with that quantity translated into money terms by dividing through by the marginal utility of income, . in this paper, the choice set restriction explored is the one created by the strict lockdown rules where individuals were prohibited from travelling to outdoor recreation sites by car. as with our visit calculations, arriving at welfare estimates for such changes for the whole of england requires aggregating up from group-day-month welfare estimates calculated at the lsoa scale. using eq. 4, we generate daily predictions of recreation activity over the lockdown, simulating the lockdown rules by removing the option of driving to greenspace from each individual's choice set over the period of strict lockdown rules and returning those options to the choice set over the period of relaxed lockdown rules. in order to draw comparison with the google mobility data, these orval predictions must be expressed in terms of visitation levels relative to the baseline period (3rd january to 6th february 2020). accordingly, we also estimate visitation to each region during the baseline period, quantites we denote v gd 0. 12 daily orval predictions of relative regional visitation, compatible to those in the google data can then be calculated according to v gmd ∕v gd 0 . figure 2 plots out these orval prediction of visitation change over the lockdown period comparing them to those in the google mobility data. in interpreting fig. 2 , it is worth noting some caveats regarding the validity of a straight comparison of the two data series. first, there is not perfect congruence in the set of locations considered as outdoor recreation destinations. google's estimates, for example, ignore recreational use of countryside paths, trips that are included in orval estimates. second google's data reports on visitors to regions irrespective of their home location while orval is restricted to visits from residents of england. third, orval predicts day trips to greenspace locations but the google data does not distinguish between day trips and trips made while staying overnight away from home. fourth, the google data records visits by individuals carrying mobile devices enabled for location reporting, a group which does not necessarily represent the adult population of england whose behaviour is modelled by orval. as a final comment, we note the fact that the google data is reported in relative terms. accordingly, our comparisons are predicated on the assumption that changes in recreation behaviour in the areas of incongruence between the two data series experience the same relative changes as those where they overlap. observe that a sharp step down in the orval predictions is evident as the strict lockdown rules are brought into force and the option of driving is removed from choice sets. the predicted time series steps up again when the recreation activity rules are relaxed and continues on to the date at which the general lockdown began to be lifted on 15th june. unfortunately, at the time of writing google had not released its mobility data for the period between 7th june and that date. 12 since the baseline spans 2 months we acquire four estimates of these region-visitation quantities; weekdays and weekends in january and february. given the baseline period comprises 29 days in january and only 6 in february, we reach an estimate of weekday and weekend visitation in the baseline, v gd 0 , as a weighted sum of the estimates from those 2 months. the parameters of the orval model are estimated from the observed recreation behaviour of the english population under normal conditions. the fact that over the strict lockdown period the orval predictions are relatively lower than the baseline arises, therefore, purely on account of the removal of the option to travel by car. the predictions do not make adjustment for the other possible drivers of visitation change under lockdown. all the same, the orval time series does a reasonable job at defining the central trend of the google data over this period. in the period of relaxed lockdown rules, the orval predictions rise to a level of around 10% above the baseline. again these predictions simply reflect normal recreation in may and june which tends to exceed that in the winter months of the baseline. notice, however, that over this second period of lockdown the orval predictions lie well below the central trend of the google data. clearly, the recreation behaviour recorded in the google mobility data over this period cannot be explained solely on account of normal variation in recreation activity across the year. a further clear pattern of difference between the google time series and the orval time series concerns recreation activity over weekends. in fig. 2 the saturday of each weekend is marked by a light grey vertical line. recall that both time series are expressed in measures of visitation relative to the baseline. accordingly, while orval predicts weekend rates of visitation to be substantially higher than midweek visitation, it does so both in the baseline period and in the periods of lockdown. indeed, for orval, the ratio of the weekday and weekend predictions to their counterparts in the baseline remain relatively constant for both lockdown periods. the same is not true of the google time series. following the commencement of lockdown, that data series is characterised by a regular pattern of down spikes coinciding with weekend periods. since those same down spikes are not evident in early march, they are suggestive of a systematic change in behaviour during the lockdown. in particular, lockdown appears to have resulted in a relative redistribution of visits across the week with comparatively more trips being taken on weekdays when compared to weekends. such changes are compatible with a relaxing of leisure time constraints amongst workers normally limited to weekend periods for their outdoor recreation. figure 2 makes clear that the use of greenspace over the lockdown was not simply normal patterns of recreation behaviour constrained by the lockdown rules. indeed, differences between the orval and the google time series provide insights into the scale of the demand shifts precipitated by the various other factors impacting on greenspace use over this period. accordingly, the next step in our analysis is to use those observed differences to estimate parameters for the orval model that capture the demand shifts experienced under lockdown. within the orval model, a demand shift parameter, ̃ , can be specified as a fixed factor entering the utility function for the outside good. adding that parameter to eq. 1 we get; if ̃ takes a negative (positive) value then the utility of the outside good falls (increases) and visiting greenspace is relatively more (less) attractive. of course our comparison of the google and orval time series suggests that the level of demand shift differs from the first period of lockdown to the second and, during each of those periods, from weekdays to weekends. accordingly, we seek to estimate four demand (7) v i0t = x it 0 +̃ shift parameters, ̃t d where d ∈ {weekday, weekend} and t indexes periods of the lockdown; that is, t ∈ t 1 , t 2 . we build our estimating equations from the basic assumption that, augmented by the true shift parameters, the orval model provides unbiased estimates of the daily visits to a region's greenspaces. recall from eq. 4 that to reduce computational burden, predictions of visitation on day t are approximated by an estimate specific to the month of t m t and whether t is midweek or on a weekend d t . the calibrated orval estimate of visitation to region g on day t , therefore, can be denoted v gm t d t ̃t t d t where t t indicates the period of lockdown in which day t falls. the actual number of visits, v gt , differs from the orval estimate on account of myriad factors that we relegate to a mean-zero error term. according to this model, the google and orval estimates of relative visitation to region g on day t of lockdown period t are related according to the equation; where y gt is the visitation change observed by google, v gt 0 is the median level of visitation to region g on the same day of the week as t during the baseline and v gd 0 t is orval's prediction of visits during the baseline on a day equivalent to that identified by d t . we progress by assuming that the error terms in eq. (8) are independent draws 13 from a mean-zero normal distribution with variance 2 . it follows that the right-hand-side of eq. 8 amounts to a ratio of normal variates with identical variance but different means. such a ratio is a cauchy distributed variate with probability density function p y y; 1 , 2 , 2 , where 1 is the mean of the normal variate in the numerator and 2 the mean of the normal variate in the denominator (see hinkley 1969 for the exact functional form of this probability). given values for the demand shift parameters and the variance parameter, 2 , therefore, we can calculate the probability of observing each data point in the google time series according to prob y gt |̃t the demand shift parameters can then be estimated by solving the maximum likelihood problem; where y is the vector of google parks visitation observations for each region over each day of the lockdown period and ̃ is the vector of demand shift parameters to be estimated. the possibility exists that behavioural responses to lockdown may have differed across england. to explore that possibility we expand eq. 9 into a latent class regression analysis (wedel and desarbo 1994) . in this analysis we assume that the english population consists of a finite set of unobserved sub-populations or classes, indexed by h = 1, … , h with each class characterised by different demand-shift parameters, ̃ h . the unobserved size of the population in each class is given by a group membership proportion h (with ∑ h h = 1). the log likelihood for the latent class regression is given by; where class membership probability, h , is specified as a function of a parameter h according to exp . the parameters to be estimated include the demand shift parameters for each class, ̃ h , the class membership parameters = 1 , … , h and the class variance parameters 2 = 2 1 , … , 2 h . following standard practice (nylund-gibson and choi 2018), the log likelihood in eq. 10 was maximised over a series of different assumptions regarding the number of classes, with a four-class model being chosen as the model delivering the best fit according to the bayes information criterion (bic). parameter estimates from that model are reported in table 1 . the a priori class membership probabilities, h , suggest a fairly even distribution of membership over the four classes ranging from 19.2% in class 3 up to 36.5% in class 2. to help in the interpretation of the demand-shift parameters, fig. 3 plots out the implied park visitation change time series associated with each different class. in that figure, comparison is made to the uncalibrated orval predictions; a time series which assumes that the only change experienced during the lockdown was the imposition of restrictions on recreation activity. the shaded areas show how demand for trips to the outdoors for each class differs from that reference level. areas shaded in green show periods where demand for trips to the outdoors exceeded the reference, those in red where demand fell below the reference. the first thing to note from fig. 3 is that for each class the demand-shift parameters distinguish a change in relative preferences for recreation on weekdays as compared to weekends. compared to the reference (orval's uncalibrated time series), over the lockdown relatively more trips are taken during the week and relatively less on weekends; possibly a result of an easing of leisure-time constraints on furloughed workers. also observe from fig. 3 that when the lockdown rules were relaxed, levels of demand for all four classes substantially exceed reference levels. that pattern possibly reflects a substitution effect as people turned to outdoor recreation in lieu of access to other prohibited leisure activities. it might also reflect an increasing propensity to engage in outdoor activities as the risks of infection diminished. considering the class 1 predictions, notice that over both periods of lockdown the time series exceeds that of the uncalibrated reference; the net effect of the demand shifters for this class is to increase use of the outdoors. indeed, class 1 represents the sub-population whose demand for the outdoors increased most substantially under lockdown. the patterns of recreation activity expressed by populations in class 2 and class 3 are reasonably similar. in both, over the period of strict lockdown rules, recreation activity tracks reference behaviour, differing primarily in the redistribution of visits from weekends to weekdays. that redistribution effect is somewhat more substantial for class 2 populations. over this first period of lockdown, it appears that for classes 2 and 3 the demandreducing effect of virus-exposure risk and the demand-increasing effect of restrictions on alternative leisure options are either small or act to cancel each other out. after the relaxation of lockdown rules, both classes exhibit a similar and substantial upward shift in demand for recreation, though the redistribution of trips from weekends to weekdays remains more pronounced in class 2. class 4 are the only population to exhibit levels of recreation activity than are consistently lower than the reference. for these populations the period of strict lockdown saw engagement with the outdoors fall below that which might be expected just from the restrictions on driving to recreation locations. after the relaxation of that rule, class 4 populations expanded their demand for outdoor recreation above reference behaviour, but considerably less so than the other populations. while the group membership probabilities of table 1 provide an indication of the mix of different behavioural classes across england, it is also possible to derive an estimate of the specific mix characterising visits to each region of the google data. using bayes theorem, the posterior probability that the observed visitation data for region g results from populations expressing the class h recreation pattern of recreation activity is; calculating such posterior probabilities for each class, we arrive at a set of estimates g h ; h = 1, … , h that we interpret as representing the proportions of visitors from each class contributing to recreation activity in region g. accordingly, we refer to those quantities as the class shares for a region's visits. our objective is to use these class shares to determine the class most likely to represent the recreation behaviour of the population of each lsoa. knowing those classes allows us to calibrate the orval model by assigning the appropriate demand-shift parameters to the choice equations for residents of each lsoa. orval can then be used to derive estimates of recreation activity and welfare changes under lockdown condiations. one approach to assigning classes to lsoas would be to identify the region in which an lsoa is located and ascribe it the class for that region with the highest visit share. the intuition here is that the majority of visits from an lsoa, r , will be to the region in which it is located, g r , such that our best guess of the behaviour class of an lsoa's population will be that most frequently observed in visits to g r . of course, that calculation ignores the fact that residents of an lsoa may also visit other regions, such that information about the behaviour class of an lsoa is also contained in the class shares of visits to those other regions. to make use of that information, we make an initial guess at the trips taken by residents of lsoa, r , to each region, g, 14 and use those to calculate the proportion of visits from r that choose g as a destination. using these proportions as weights, we calculate the weighted sum of the class shares for each region's visits, to arrive at our best guess of the class shares characterising r . we assign r the class exhibiting the highest class share. figure 4 maps out the classification of lsoas in england to different classes. to simplify presentation and reflect their similarity, areas in class 2 and 3 are presented in the same shade. while the data is plotted at the lsoa scale, as might be expected, the pattern in particular, we calculate the visits from each lsoa to each region assuming class 1 behaviour, then repeating those calculations for each of the three remaining behavioural classes. our estimate of visits from lsoa r to region g are calculated as the weighted sum of those four visit estimates where the weights are given by that region's class-visit shares, of class membership broadly follows the regions upon which the data analysis is based. those regions are outlined in white and close inspection reveals that our classification procedure allots some lsoas along region borders to a different behaviour class to lsoas in the region interior. there exists some interpretable spatial pattern in the distribution of class membership described in fig. 4 . for instance, all the major metropolitan areas of england exhibit class 2 and 3 behaviour changes (expected activity under strict lockdown, much increased activity under relaxed lockdown). in addition, class 1 behaviour changes (increased activity under strict lockdown, greatly increased activity under relaxed lockdown), show clear patterns of regional clustering most notably along the south coast and central-south region of england. we suspect that these patterns reflect regional differences in the perceived and actual risks of exposure to the virus. areas exhibiting class 4 behaviour changes (reduced activity under strict lockdown, increased activity under relaxed lockdown) are largely located in relatively remote and rural areas of england. that pattern would be commensurate with locations whose workforces are primarily engaged in the food production sector; an occupation classed as essential in the lockdown and not subject to restriction under the lockdown rules. the top left panel of fig. 5 presents orval's predictions of visitation change for england once the recreational choices of residents of each lsoa have been adjusted with the demand shifters for their estimated behaviour class. applying the methods described in sect. 4.2, we can now use this calibrated version of the orval model to estimate levels of table 2 . the estimates in table 2 are for visits and values aggregated over all english residents over (1) the seven weeks of strict lockdown rules, (2) the five weeks of relaxed lockdown consider first the changes in estimated visits. where normally we would expect some 168.8 million trips to the outdoors taken by car, such trips were prohibited over the 7 weeks of strict lockdown rules. what the orval estimates reveal is that individuals responded to those restrictions through substituting to trips taken on foot. in the period of strict lockdown, the calibrated model estimates that 275.0 million trips were taken to greenspaces on foot, an almost 40% increase over the 198.8 million expected under normal conditions. figure 6 illustrates how recreation behaviour changed across england in this period. the left-hand panel plots out orval estimates of the spatial distribution of weekly visits taken by residents of the major metropolitan areas of england under normal conditions. the right hand panel contrasts that with the distribution of visits under the strict lockdown rules. prohibited from driving, outdoor recreation activity refocused on local greenspaces. once the restrictions on driving were lifted, the effects of the demand-shifts evident in the google data become clear. visits by both car and on foot increase, resulting in levels of recreation visits that are some 27.5% above those expected under normal conditions. the story of outdoor recreation under lockdown is one in which people offset the restrictions on driving to recreation sites by switching to walking to greenspaces local to their homes. that behaviour along with an upward shift in demand for recreation resulted in the overall number of visits to the outdoors over the lockdown period being little changed from that under normal restrictions. our calculations of welfare change suggest that the cost of lockdown on welfare derived from greenspaces was negligible, dropping by £150 million or some 2.7% of that realised under normal conditions. a second set of analyses that are possible with the orval model use the calibration parameters to explore the visit changes and welfare consequences that might have arisen should alternative rules on recreation have been instituted in the lockdown. here we consider two such counterfactuals. the first is a counterfactual where the strict lockdown rules prohibiting driving to greenspaces were extended over the whole period from march 23rd to june 15th. the second is a counterfactual in which no restrictions were imposed on recreation activity over lockdown. time series describing recreation activity under those two counterfactuals are presented in the bottom panels of fig. 5 . observe that in the strict lockdown counterfactual we are projecting behaviour out over the second period of the lockdown under rules for which we do not have observations from the google mobility data against which to calibrate. two assumptions are possible. first that the demand shift parameters characterising behaviour under the strict lockdown period continue to characterise behaviour under the extension of those rules into the second period. alternatively, that the second period might be characterised by the demand shift parameters charactering recreation behaviour during that second period under the relaxed rules. since the demand shift parameters of the second period are universally more positive than those for the first period, those two assumptions suggest lower and upper bound estimates of possible behaviour under the strict lockdown counterfactual. those bounds are traced out in the plot of fig. 5 with the grey shaded area demarking the paths lying between those bounds. similar arguments lead to bounds on the recreation activity over the first period in the relaxed lockdown rules counterfactual. these too are shown in fig. 5 . summary details of recreation visits and values under the strict lockdown counterfactual are presented in table 3 . in that table, we present estimates that are averages of those for the lower and upper bounds and contrast those with estimates of visits under the lockdown under the actual lockdown rules. not surprisingly, maintaining the rule prohibiting driving to outdoor recreation locations has the effect of suppressing engagement with greenspaces. the orval model predicts some expansion of walking in the second period of lockdown to compensate for the continuing restrictions on driving opportunities. all the same, maintaining strict rules on recreation over the whole lockdown results in an estimated 18.4% reduction in visits to the outdoors compared to those estimated under the actual lockdown rules. in terms of welfare, the stricter rules impose a welfare cost on english residents; the value flow realised from greenspace access falls by some £1.14 billion. viewed the other way, the government's decision to relax the rules on outdoor recreation activity delivered a £1.14 billion welfare boost to residents of england. table 4 provides an identical analysis for outcomes under the relaxed rules counterfactual in which the lockdown proceeded without restrictions on outdoor recreation activity. under the relaxed-rules counterfactual orval predicts an expansion of recreation activity. visits to the outdoors are some 24.2% greater than those estimated under the actual lockdown rules. again that translates into changes in the economic value of greenspace. a lockdown with no restrictions on recreation activity increases the estimates of the welfare benefits of greenspace access by some £1.47 billion. viewed the other way, english residents suffered a welfare cost of £1.47 billion as a consequence of the government's decision to restrict recreation activity over the first period of the covid-19 lockdown. using analytical methods that leverage google mobility data and the predictive powers of the orval model, this paper explores how the covid-19 lockdown in england changed how people engaged with greenspace and impacted on the economic value they derived from those interactions. we find strong evidence to support the contention that greenspace became a significant source of welfare for citizens at a time when opportunities for alternative uses of leisure time were even more seriously curtailed. one key change identified by our analysis is that the lockdown rules forced citizens to get out of their cars and walk. trips to greenspaces by car fell by 47% over the whole lockdown period with an attendant 34% rise in trips taken on foot. increased engagement in outdoor recreation (particularly in the second period of lockdown) coupled with this substitution behaviour meant that, despite the restrictions citizens maintained value flows from greenspace over the lockdown comparable to those they would have enjoyed over that same period under normal conditions. our analysis also explores how the welfare derived from greenspace might have differed under alternative lockdown rules. we discover that the adoption of more relaxed rules on the use of greenspace during the first period of the lockdown would have delivered increased welfare flow from greenspace of £1.47 billion. a retrospective interpretation of the decision to impose limitations on engagement with greenspace, therefore, would be that the government judged that the health costs associated with the increased risk of infection from adopting less strict rules over that period were in excess of £1.47 billion. a second counterfactual policy considered the maintenance of the rules limiting engagement with greenspace into the second part of the lockdown. our analysis reveals that such a policy would have reduced the value flow from greenspace by £1.14 billion. the retrospective interpretation of that figure is that by the time the rules on outdoor recreation were relaxed the government judged that the societal costs of the increased infections that might arise as a consequence, to be less than £1.14 billion. several important research questions remain to be answered and the analytical framework developed in this paper stands well placed to address them. as our analysis reveals, behavioural responses to the lockdown differed across the country. in this paper we offer only tentative speculations as to why those differences arose. a more detailed analysis relating the observed changes in outdoor recreation activity to factors including regional differences in the risk of exposure to covid-19, profiles of occupations, sociodemographics and the local availability and quality of greenspaces might reveal important information as to the key drivers of outdoor recreation behaviour under lockdown. likewise that detailed exploration of spatial differences in outdoor recreation activity, might help identify those communities that were most seriously disadvantaged by the lockdown restrictions perhaps on account of the lack of availability of high quality local greenspace. our analysis also reveals that in the second period of lockdown, use of the outdoors expanded very substantially, far exceeding that expected under normal conditions. the google mobility data accessed for the purposes of this analysis provided observations only as far as 7th june 2020. more recent data releases suggest that this increased demand has been maintained even as other areas of everyday life gradually return to normal. that trend has led to speculation that the covid-19 lockdown has precipitated widespread "re-engagement" with outdoor recreation and is perhaps evidence of a structural shift in preferences for greenspaces (royal society for the protection of birds 2020). revisiting the google mobility data in a few months' time and extending the analyses of this paper should help establish the degree of persistence of that shift. if covid-19 has indeed led citizens of england to discover the delights of the outdoors then perhaps that offers a faint glimmer of positive news in a period so scarred by suffering. open access this article is licensed under a creative commons attribution 4.0 international license, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the creative commons licence, and indicate if changes were made. the images or other third party material in this article are included in the article's creative commons licence, unless indicated otherwise in a credit line to the material. if material is not included in the article's creative commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. to view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. haunting pictures show weston-super-mare in lockdown a theoretical analysis of the cross-nested logit model investigating preference heterogeneity in a repeated discrete-choice recreation demand model of atlantic salmon fishing seaside resorts at risk of becoming ghost towns due to coronavirus. the telegraph practical solutions for sampling alternatives in large-scale models the outdoor recreation valuation (orval) tool: data set construction expanding and improving the outdoor recreation valuation tool-nr0168 values of time and vehicle operating costs: tag unit 3.5.6. department for transport, london department for transport (2015) provision of market research for value of travel time savings and reliability customizable contraction hierarchies measuring the benefits of water pollution abatement covid-19 community mobility reports sampling of alternatives in multivariate extreme value (mev) models 85/covid -19upd ate covid-19) statistics: hmrc data about the coronavirus job retention scheme, the self-employment income support scheme, and the vat payments deferral scheme the green book: central government guidance on appraisal and evaluation on the ratio of two correlated normal random variables coronavirus: empty streets in brighton during lockdown. the argus prime minister's statement on coronavirus (covid-19) a model of weekend recreation travel demand here-are-the-new-rules -as-newso m-order s-all-calif ornia ns-to-stay-at-home the estimation of choice probabilities from choice based samples modelling the choice of residential location. spatial interaction theory and planning models midas: uk daily temperature data. ncas british atmospheric data centre. https :// catal ogue.ceda.ac.uk/uuid/1bb47 9d3b1 e38c3 39adb 9c82c 15579 d8 midas: uk daily rainfall data a repeated nested-logit model of atlantic salmon fishing monitor of engagement with the natural environment: the national survey on people and the natural environment ten frequently asked questions about latent class analysis understanding drivers of change in park visitation during the covid-19 pandemic: a spatial application of big data recovering together a report of public opinion on the role and importance of nature during and in our recovery from the coronavirus crisis in england applied welfare economics with discrete choice models urban nature in a time of crisis: recreational use of green space increases during the covid-19 outbreak in oslo a review of recent developments in latent class regression models. in: bagozzi r (ed) advanced methods of marketing research key: cord-335272-jypxi99z authors: sharma, anupam joya; subramanyam, malavika a. title: a cross-sectional study of psychological wellbeing of indian adults during the covid-19 lockdown: different strokes for different folks date: 2020-09-03 journal: plos one doi: 10.1371/journal.pone.0238761 sha: doc_id: 335272 cord_uid: jypxi99z the psychological impacts of the lockdown due to the covid-19 pandemic are widely documented. in india, a family-centric society with a high population density and extreme social stratification, the impact of the lockdown might vary across diverse social groups. however, the patterning in the psychological impact of the lockdown among lgbt adults and persons known to be at higher risk of the complications of covid-19 (such as persons with comorbidities or a history of mental illness) is not known in the indian context. we used mixed methods (online survey, n = 282 and in-depth interviews, n = 14) to investigate whether the psychological influence of the lockdown was different across these groups of indian adults. we fitted linear and logistic regression models adjusted for sociodemographic covariates. thematic analysis helped us identify emergent themes in our qualitative narratives. anxiety was found to be higher among lgbt adults (β = 2.44, ci: 0.58, 4.31), the high-risk group (persons with comorbidities) (β = 2.20, ci:0.36, 4.05), and those with a history of depression/loneliness (β = 3.89, ci:2.34, 5.44). persons belonging to the lgbt group reported a greater usage of pornography than the heterosexuals (β = 2.72, ci: 0.09, 5.36) during the lockdown. qualitative findings suggested that lgbt adults likely used pornography and masturbation to cope with the lockdown, given the limited physical access to sexual partners in a society that stigmatizes homosexuality. moreover, both qualitative and quantitative study findings suggested that greater frequency of calling family members during lockdown could strengthen social relationships and increase social empathy. the study thereby urgently calls for the attention of policymakers to take sensitive and inclusive health-related decisions for the marginalized and the vulnerable, both during and after the crisis. the coronavirus disease , caused by the novel coronavirus sars-cov-2, first emerged in wuhan, china during late 2019 and was labeled a public health emergency by the world health organization [1] . the recent and rapid increase in the number of covid-19 cases, 16 ,523,815 as on 29 th july 2020 globally [2] , has increased panic across countries [3] . every country affected by the virus adopted several measures in order to curb its spread. india, plos one | https://doi.org/10.1371/journal.pone.0238761 september 3, 2020 1 / 23 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 home to 1.3 billion people, announced a nationwide "lockdown" on 25 th march 2020 [4] . the lockdown restricted citizens' physical mobility, advocated social distancing norms, and limited a majority of public services while allowing the essential ones. however, these measures of sheltering-in-place, equivalent to an extended quarantine, likely created a stressful environment for the citizens, given the sudden disruption in their daily routines [5] , [6] . these disruptions could contribute towards adverse psychological outcomes such as post-traumatic stress symptoms [7] and aggressive behaviors [8] . for instance, one indian study by gautam & sharma [9] , highlighted that the lockdown could increase the psychological toll on the indian academic fraternity because of the disruption in their work, which additionally brings financial instability to the contractual staff. however, the impact of a lockdown might vary across diverse social groups. individuals who are living alone or away from family (or loved ones), those suffering from economic losses, or having a history of negative psychological states, could be at a higher risk of depression, loneliness, and anxiety disorders during the lockdown. in addition to these sources of stress that could get exacerbated during the lockdown, certain social groups may have to deal with stress due to identifying as a member of a minority community in their society. meyer's minority stress model [10] explains this minority stress as arising from the risk of social shunning, discrimination based on identity, or the consequent efforts to hide minority status. one such minority group that could be vulnerable to the effects of the lockdown is the lgbt community. according to meyer's minority stress model, lgbt individuals are exposed to unique stressors rooted in societal structure and related to their minority identity, which could combine with other stressors to impact their psychological wellbeing [10] . therefore, we posit that persons belonging to the lgbt community could suffer from increased stress during lockdown. further, the restrictions on physical mobility might have not only disrupted the social lives of many individuals but also paused their sexual lives [11, 12] . to cope with these disruptions in sexual lives, in addition to the on-going stress and boredom due to the lockdown, individuals could also rely on pornography [11, 13] . reports revealed a 33% average increase in pornwatching during the lockdown period (march 2020) in india [14] . moreover, the time spent indoors during the lockdown could also enable adults to explore their body and experience pleasure through masturbation [15] . pornography use and masturbation are recommended ways of meeting sexual needs without the risk of contracting sars-cov-2 infection during the pandemic [12] , although prolonged and habitual usage of pornography and masturbation could have negative consequences on sexual satisfaction [16] . the psychological impact of the lockdown could also be higher among individuals with a known higher, versus lower, risk of experiencing complications of covid-19, such as the elderly or those with co-morbid conditions (for example, persons with chronic respiratory illnesses and diabetes). the extent of daily exposure to the pandemic may also matter: a study from wuhan, china, found a high prevalence of depressive symptoms among frontline healthcare workers [17] . similarly, another study from australia showed that people living in high risk infection zones reported greater psychological distress than those living in uninfected areas [18] . these findings suggest that a higher perceived risk of covid-19 could increase anticipatory fear and anxiety. this fear, depression, loneliness, and anxiety during the time of crisis not only could affect mental health but also adversely affect one's lifestyle and diet, ultimately impacting physical health [19] . previous studies have shown that depression (or anxiety) worsens sleep disorders [20] and eating disorders [21] . despite these risks, several personal and social resources could be available for individuals to cope with the adverse effects of the crisis. in a family-centric country such as india, family is regarded as a vital social support [22] , especially during a crisis. living with family/relatives (or regular virtual interactions through phone or online media) could act as social support which could result in lowering stress during the lockdown. of note, an opportunity to spend extended time with family members could strengthen family bonds and enhance work-family balance, leading to a better quality of life [23] . while the lockdown period may be spent in fulfilling varied responsibilities, it could also have created opportunities for many to spend time with family (and loved ones) and potentially improve the quality of family relationships through physical or virtual proximity. nevertheless, these opportunities might act as a situational coercion for a few individuals (who have a history of family maladjustment or family conflict) and induce additional stressors, further increasing their vulnerability to adverse psychological outcomes during a lockdown. in addition to these social resources, several individual-level characteristics such as the nature of employment, access to material resources; and, psychological resources such as resilience-coping and optimism might be beneficial in minimizing the effect of this crisis [24, 25] . findings of a recent study from china suggest that positive personal-level characteristics such as emotional-control and optimism could also help minimizing the negative effects of the covid-19 crisis [26] . although these concerns warrant attention in the indian context, we could locate only a few studies reporting the prevalence of depression and anxiety during the covid-19 crisis in india, including a comparison across age and gender groups [27, 28] . moreover, these studies did not examine the prevalence of these outcomes across other social groups, including the vulnerable and the hidden group of lgbt adults. the rapid increase in the number of covid-19 cases in india and the disruptions due to the lockdown, warrant investigating the processes explaining any social patterning in the psychosocial wellbeing of indian citizens during this crisis. in response, our study of indian adults unpacks how social factors such as sexual orientation, relationship status, and residence in high-infection areas, could be linked with several psychological outcomes during the lockdown. we also investigated whether a higher risk of covid-19 complications and a history of depression or loneliness worsens the mental health impact of the lockdown. we further explored the complex processes explaining if and how anxiety or depressive symptoms were related to sleeping and eating habits during the lockdown. we also investigated the role individual-level resources played in coping with the effects of the crisis. because the lockdown likely changed the nature of social interactions, we additionally examined if this brought any change in how individuals viewed the world and in their social empathy, which could be an important psychological resource for overall wellbeing and quality of life. the primary research questions explored in this study were: in addition to the above primary questions, inspired by the initial two qualitative interviews, we also addressed the following in our study: 1. how are sharing vulnerabilities (stress and depression) with loved ones, and the frequency of interaction with family related to strengthening of social bonds and social empathy during the lockdown? we followed a convergent mixed methods approach [29, 30] in our study. first, two exploratory qualitative in-depth interviews were conducted to refine our research questions. the narratives of the two participants (one male and one female participant) guided us in identifying key factors affecting their mental wellbeing during the time of the lockdown. these participants shared their frustrations related to the lockdown, the disruption in their routine work, the chaos around them regarding the increase in number of cases, their challenges in general, and their overall feeling about the entire crisis situation. based on the data from these two interviews, we constructed our online quantitative survey. data for the study were simultaneously collected through the online survey and qualitative interviews. however, whenever the data from the quantitative survey revealed an interesting picture, we dug deeper about it in our qualitative interviews to understand its context and complexity. we carried out an online survey from 9 th may to 15 th may 2020. our survey questionnaire in the form of an anonymous google form was circulated through several facebook groups as well as whatsapp and instagram contacts. we further used a snowball sampling procedure to increase the number of responses. the authors requested their family members, friends, colleagues, and professional networks to further spread the form among their networks. the introductory passage in the google form briefed the participants about the broad objective of the study and requested their voluntary participation. further, the passage also promised anonymity and confidentiality to the participants. once the participants read the introductory passage, they were requested to proceed to fill up the survey. the participants' agreement to take part in the survey after they were given an opportunity to carefully and unhurriedly review the information about all relevant information about the study, the voluntary nature of participation, their right to withdraw at any time, and the confidentiality of their data, was considered to be implied consent. due to the online nature of the survey, we could not limit its spread to a specific geography. however, we specified our eligibility criteria (indian citizen, presently residing in india, aged 18 years or above, and willing to fill the form in english) in the introductory passage to maximize the chance that we only got responses from india. we received responses from 282 participants. response variables. anxiety. we measured anxiety using the general anxiety disorder (gad-7) scale [31] . this widely used scale includes items such as "over the past 2 weeks how often have you been bothered by the following problems: feeling nervous, anxious or on edge?" the responses were recorded on a 4-point likert scale ranging from "not at all (0)" to "nearly every day (3) ." we found a good internal consistency of the gad-7 scale in our sample (cronbach's alpha = 0.91). the aggregate of the item scores reflected the total anxiety score. depressive symptoms. we assessed depressive symptoms of our participants using the short version of the cesd-d scale, a 10-item scale [32] . the scale includes items such as "in the past week how often have you felt any of these: i had trouble keeping my mind on what i was doing." two items were reverse scored. the responses to all the items varied from "less than a day" (0) to "5-7 days" (3) . we later discovered that responses to one item (i was bothered by things that usually do not bother me) did not get recorded possibly due to some technical error in the google form. however, following siddiqui [33] and hawthorne et al. [34] , we imputed the personmean score for the missing item. we found good internal consistency of the scale (including the imputed score) in our sample (cronbach's alpha = 0.86). the item total was used as the depressive symptom score. symptoms of the internet addiction. we used the internet addiction test (iat) scale [35] to measure symptoms of addiction of the internet. the scale includes 12 items such as "over the past 2 weeks, how often have your felt: find yourself saying "just a few more minutes" when online?" responses varied from rarely (1) to always (5) . the cronbach's alpha was found to be 0.92. the total score of all 12 items yielded the internet addiction score. compulsive consumption of pornography. the compulsive pornography consumption (cpc) scale [36] was used to assess the symptoms of usage of pornography. the 6-item scale included items such as "please indicate how these statements described you during the past 2 weeks: i thought of pornography (porn) when i was trying to focus on other things." the responses were recorded on a 7-point likert scale ranging from never (1) to very frequently (7) . the cronbach's alpha was 0.87 in our sample. the sum total of the scores of all items resulted in the pornography consumption score. experiences of hostility. a single item was used to measure experiences of hostility during the lockdown: have you been facing the following problems in the last 2 weeks? you faced a hostile situation (including emotional, physical, and mental violence) from anyone in the place you are currently in. responses were dichotomized to yes and no. change in food habits (time and consumption). we assessed any change in the participants' food habits using a single item: have you been facing the following problems in the last 2 weeks? food patterns (type of foods consumed/timings) have changed. responses varied from not at all (1) to always (5) . sleeping problems. sleeping problems were measured using a combination of two items, have you been facing the following problems in the last 2 weeks? your sleep cycle has changed drastically, and you have difficulty in falling asleep. the responses varied from not at all (1) to always (5) . the additive score yielded the level of sleeping problem with scores ranging from 2 to 10. frequency of masturbation. frequency of masturbation was assessed using a single item, how often are you engaging yourself in masturbation activities in the last 4 weeks? the responses varied from never (1) to multiple times a day (6) . social empathy and quality of social relationships. social empathy was operationalized based on the participants' choice of several options offered. the selection of any of the following options: you have become more socially responsible; you have become more active in neighborhood associations/groups or other social groups near your residence; you have been thinking about the vulnerable in our society and tried to do at least something for them (donating or helping in other ways) indicated increased social empathy coded as 1 (otherwise 0). similarly, the quality of social relationships was recorded as 1 (improved), if the participants selected even one of the following options: you have started liking to spend time with your closed ones more than before; you have strengthened your relationship with your friends; and you have strengthened your relationship with your family/partner, otherwise as 0. predictors. sharing stress and anxiety with loved ones. if participants selected "yes" to the question have you shared your stress and vulnerability with loved ones during the lockdown, it was coded as 1, else as 0. resilience coping. we assessed resilience coping of the participants using the 4-item brief resilience coping scale (brcs) [37] . the scale included items such as i look for creative ways to alter difficult situations, with responses varying from does not describe me at all (1) to describes me very well (5) . the aggregated score of all items reflected the participants' resilience. the cronbach's alpha was 0.82. optimism. optimism was measured using the 10-item revised life orientation test (lot-r) scale [38] . it included items such as in uncertain times, i usually expect the best while responses ranged from, i disagree a lot (1) to i agree a lot (5) . four items were fillers and were removed from the analysis. the aggregate of all item scores resulted in the optimism score. we found moderate internal consistency of the scale in our sample (cronbach's alpha = 0.55). change in frequency of calling family members. we compared the frequency of calling family members during the lockdown with that during october 2019-march 2020. we treated this as an indicator of the change in frequency of calling family members during the lockdown. we coded it as 1 if the frequency increased, otherwise as 0. high-risk group. individuals who reported having any of the following: chronic respiratory illnesses, diabetes, heart disease, hypertension, or a weakened immune system, were categorized as belonging to the "high-risk group (1)", else as "low-risk group (0)." history of depression/loneliness. we grouped the participants who reported having a history of depression or loneliness as "group with history of depression/loneliness (1)", otherwise "group with no history of depression/loneliness (0)." categories of state exposed to covid-19. we referred to data from ministry of health and family welfare, india [39] for categorizing the states as per the counts of covid-19 cases. we coded maharashtra (with cases more than 25000 during the data collection) as "highest exposure;" tamil nadu, gujarat and new delhi (with around 10,000 cases) as "high exposure;" rajasthan, madhya pradesh, and uttar pradesh (near to 5000 cases) as "moderate exposure"; and rest of the states as "low exposure." sociodemographic characteristics. we also collected information on age (18-29/30-44/45-59/and above 60 years), gender (male/female/others), sexual orientation (straight/queer), relationship status (opposite-sex relationship/same-sex relationship/single/complicated), place of residence (rural/urban), educational qualification (postgraduate/graduate or diploma/12 th or lower), and annual income in indian rupees (0-3,00,000 (approximately, $0-$4000)/ 3,00,000-10,00,000 ($4000-$13500)/10,00,000-20,00,000($13500-$27000)/above 20,00,000 ($27000 and above)), and the state of residence. we conducted 14 in-depth interviews from 10 th may through 17 th may 2020. we circulated an advertisement inviting participants for telephonic interviews through social media (facebook, instagram, and twitter) and personal contacts of the authors. the advertisement included a brief introduction about the study, contact information of the first author (also the interviewer) the nature of the interviews, and about the approximate length of the interview. the introduction also informed the participants about the sensitive nature of the topic (which included questions on their personal/intimate lives) and asked their preferences of the gender of the interviewer. however, none of the participants shared concerns being interviewed by ajs (a man). interested participants contacted ajs through email/facebook/whatsapp showing their willingness to participate. at that time, ajs briefed the participants about the study; about their anonymity and confidentiality of data; and that the interview would be terminated at any point the participant showed discomfort. ajs and the participants mutually agreed on a time for the telephonic interview. before beginning the interview, ajs once again sought informed verbal consent. in addition to verbal consent, ajs also sought consent to audiotape the interviews. four participants were reluctant to get the interviews audiotaped. detailed notes (including several quotes) were taken during these interviews. all interviews began with broader questions such as "how do you feel about the entire situation (of covid-19 and lockdown)?" ajs was cautious while asking personal questions, especially about romantic and sexual lives of the participants. ajs ensured participants' comfort, not only while asking sensitive questions, but throughout the interview process, by taking a pause and asking, "should we proceed?" however, there were no instances where there arose the need to terminate any interview. all participants shared their emotions, vulnerabilities, moods, challenges, change of lifestyle, and perceived wellbeing during the lockdown (and the covid-19 crisis). specific comments regarding the mood and context were noted by ajs to give rigor to the analysis. at the end of the interviews, the participants were requested to share about the study with their peers and network, seeking their participation. the length of the interviews ranged from 28 minutes to 1 hour 40 minutes. eight participants were recruited through the advertisement while 2 participants were recruited through snowball sampling. additionally, we allowed the participants of the online quantitative survey to express interest for a follow-up telephone call. we recruited 4 participants through this method. different methods of recruitment helped us get a socially diverse sample in a short time. in addition to the interviews, we collected data through an open-ended question in the online quantitative survey. this allowed the participants to share their concerns related to the pandemic situation (and lockdown). extracted quotes were used in our qualitative analysis. quantitative analysis. the distribution of all continuous variables was checked for normality [40] . next, we fitted separate multivariable linear regression models to estimate the association of the independent variables (sexual orientation, relationship status, high-risk group, and living in a state with high number of cases) with psychological outcomes (anxiety, depressive symptoms, internet addiction and pornography consumption) adjusted for the sociodemographic covariates-age, gender, annual income, educational qualification, place of residence-and for individual personal resources (optimism and resilience). we fitted separate logistic regression models to estimate the associations of sexual orientation and relationship status with the binary variable indicating the experience of hostility, adjusting for all sociodemographic variables. we also fitted multivariable linear regression models to estimate the association of anxiety and depressive symptoms with changes in sleep and food cycles (separate models), adjusted for the sociodemographic covariates and personal resources. additionally, we fitted separate logistic regression models to estimate the association of increased frequency of calling family members with social empathy and the quality of social relationships adjusted for sociodemographic covariates. for all our analyses, alpha was set at 0.05. all statistical models were run in stata version 12 [41] . qualitative analysis. we chose a thematic analysis approach [42] to analyze the qualitative data. the analysis began with ajs (who also conducted all the interviews) familiarizing himself with the data by spending prolonged time in re-listening to the audiotaped interviews and reviewing the transcript excerpts. the participants' narratives about their emotional responses to the situation; their description of how the lockdown affected their routine, relationships, and social responsibilities; their sense of self (including their body); their perspective on life; their coping mechanisms; and, views towards a "new world" guided the coding process. four themes emerged from these indexed codes [43] and the detailed comments. additionally, nvt (an external researcher) categorized the themes emerging from the codes. the coding scheme was discussed among the two authors (and nvt), and after critical analysis, the themes were confirmed with a high inter-coder reliability [44] . follow-up interviews with two participants were carried out separately for respondent validation [45] . additionally, several quotes from the open-ended section of the online survey were included in the themes that emerged from the qualitative interviews allowing better representation of all the voices heard. both quantitative and qualitative findings carried equal weight in this study. the qualitative themes that emerged gave richer context to the quantitative results during the interpretation phase. the study was motivated by previous work of ajs and mas on the queer community, and their understanding of the community's unique vulnerabilities. apart from this, the lockdown has severely restricted the ability of ajs and mas, not only in terms of physical mobility but also in terms of distance from loved ones and has affected their productivity. interviews by ajs were conducted with this frame of reference. the study was approved by the institutional ethics committee, iit gandhinagar, india. utmost precaution was taken by ajs while conducting the telephonic interviews. the participants were informed about the sensitive nature of the questions and were informed that they could skip any question. ajs constantly monitored the mood of the conversation and frequently asked the participants about their willingness to continue. names of all the participants have been changed in this study to protect anonymity. we analyzed a sample of 282 indian adults who responded to the online survey (table 1) . a majority (~75%) of our participants were 30 years or younger. around 60% identified themselves as male, and about 77% reported to be heterosexuals. only a small proportion (~12%) of our participants had education less than 12 th standard (high school). greatest proportion (~81%) of the participants resided in urban areas. anxiety and depressive symptoms across social groups. our fully adjusted models (adjusted for gender, age, educational qualification, income, and place of residence) (see table 2 ) found that gad scores were higher, on average, in lgbt adults (β = 2.44, ci: 0.58, 4.31) versus heterosexuals, high-risk group (β = 2.20, ci:0.36, 4.05) versus low-risk group, and participants with history of depression/loneliness (β = 3.89, ci:2.34, 5.44) versus participants with no history of depression/loneliness. however, gad scores were lower for single participants (β = -2.35, ci: -4.30, -0.39) than those who were in opposite-sex relationships. we could not find statistically significant associations of living in a state reporting a high count of covid-19 cases with anxiety symptoms. unsurprisingly, we found a statistically significant association of a history of depression/ loneliness with increased depressive symptoms during the lockdown (β = 4.34, ci: 2.38, 6.30), independent of other covariates. however, we could not find any evidence linking sexual orientation, relationship status, living in a state reporting a high count of covid-19 cases, and belonging to a high-risk group, with depressive symptoms. addiction to the internet, consumption of pornography, and frequency of masturbation across groups. we found that a history of depression/loneliness was statistically significantly associated with higher internet-addiction symptoms (β = 4.55, ci: 1.47, 7.63), independent of all other covariates. however, we could not find evidence that other predictors were associated with internet addiction. moreover, our fully adjusted models showed greater symptoms of pornography usage, on average, in lgbt adults (β = 2.72, ci: 0.09, 5.36) versus heterosexuals, in the high-risk group (β = 2.80, ci: 0.22, 5.39) versus low-risk group, in participants in same-sex relationships (β = 9.15, ci: 0.93, 17.38) versus opposite-sex relationships, and among those with a history of depression/loneliness (β = 2.63, ci: 0.41, 4.86) versus no such history. additionally, our adjusted models showed that lgbt adults (β = 1.39, ci:0.94, 1.86) and participants in same-sex relationships (β = 2.07, ci = 0.50, 3.63) reported a higher frequency of masturbation during the lockdown compared to their heterosexual peers. experiences of hostility. we did not find statistical evidence that experiences of hostility differed across sexual orientation and relationship status. although statistically not significant, lgbt adults (versus heterosexuals) had higher odds of experiencing hostility (aor = 1.63, ci: 0.82, 3.23) during the lockdown independent of the sociodemographic covariates. association of anxiety and depressive symptoms with food and sleep habits. our fully adjusted models (adjusted for sociodemographic variables and positive resources) (see table 3 ) showed that higher depressive symptoms and anxiety symptoms were associated with greater reports of self-reported sleep disorders (β = 0.16, ci: 0.11, 0.20 and β = 0.19, ci: 0.14, 0.25, respectively) and self-reported changes in food pattern (β = 0.05, ci: 0.03, 0.08 and β = 0.08, ci: 0.05, 0.11). social empathy and quality of relationships. our fully adjusted logistic regression models (see table 3 ) showed that participants who increased the frequency of calling their family members during the lockdown (compared to 6 months earlier) had higher odds of enhancing the quality of their social relationships (aor = 2.56, ci: 1.19, 5.52), and reporting increased social empathy (aor = 2.27, ci: 1.06, 4.88), independent of all sociodemographic covariates. moreover, our models found that sharing vulnerabilities (stress/depression) with loved ones was associated with higher odds of being socially empathetic (aor = 3.99, ci:1.95, 8.14), and enhancing social relationships (aor = 2.96, ci: 1.52, 5.74), after accounting for all sociodemographic covariates. fourteen participants shared with us the slices of their lives during the lockdown. of the 14 participants, 8 were male, 5 were female, and 1 identified themselves as a non-binary transgender. six of the 14 identified themselves as lgbt adults. four were students, 5 worked in private/public sectors (hereafter "service"), and 5 engaged in business/entrepreneurship. the thematic analysis of the narratives of these 14 participants revealed four broad themes. not all participants' narratives highlighted all the themes; however, each participant's narrative was reflected in at least one theme. theme 1: emotional responses to "distance from the real world". all 14 participants expressed their unique concerns about the lockdown situation. words such as "frustrated," "stressed," "angry," and "suffocated," were frequently used to describe their emotions. although the intensity of the negative impact of the lockdown varied across participants, most participants (9/10) shared how the lockdown disrupted their lives causing frustration and agitation. for instance, ashok (male, heterosexual, service, 27 years old) shared, similar responses of frustration were shared by several students who had mostly enjoyed an outdoorsy life, be it spending time on their college campuses or with friends outside. however, a few shared a different reason for their anxiety: living in a place with a high number of covid-19 cases. rajini's (female, heterosexual, homemaker, 39 years old) narrative is an example: we are in a containment (severe movement restriction) zone, and since the last 12 while several respondents shared the fear they felt currently due to a high number of covid-19 cases in their areas, two respondents, tulika (female, belonged to lgbt group, service, 23 years old) and salma (transgender (non-binary), belonged to lgbt group, service, 43 years old) shared how this "worse" time had forced them to revisit their past trauma. tulika, who had gone through a break-up one year earlier and was recovering with the help of therapy elaborated, another instance of past trauma being triggered was in the case of salma (they/their/them), who had always managed situations of discrimination (against their transgender identity) calmly but recently lost their temper during such an event. the policewoman stopped us (them and their partner), and asked me "what do you think you are," [. . .] i have always tried to be calm in such situations, but this time, i just lost all my calmness. it was a mix of so many things, my frustration at work during the crisis, my mother falling ill just a week before the incident, all these acted together. their stories revealed that the stress and anxiety developed during the lockdown had revived old memories of trauma. thus, in response, tulika chose to go through emergency sessions with her therapist, while salma failed to stay calm and burst out when faced with genderbased discrimination. while tulika and salma revisited trauma, anurag (male, belonged to lgbt group, businessman, 37 years old) who moved to his parental home during the lockdown felt distant from his "real" world. he shared, anurag's feeling of distance from his "real" world highlighted how uncomfortable he was at his old home with his parents. he shared how things around his parents' house reminded him of how uneasy he felt while growing up as a gay man. anurag was conflicted by a dilemma: while the lockdown brought him closer to his parents at a time of their need, it also placed on him an additional psychological burden. the suffocation potentially felt by lgbt adults forced due to the lockdown to stay with others to whom they were not out was evident in this quote shared via the online survey: while a majority of the participants (9/14) shared mostly about their negative states of mind, a few took a moment to share the positive impact of the lockdown on their lives. for instance, rumi (female, heterosexual, businessperson, 33 years old) described how she saw the lockdown as an opportunity to introspect about her own life. the narrative of ashish (male, belonged to lgbt group, service, 23 years old, hiv positive) highlighted his equanimity during the crisis. he shared that he did not have any trouble during the lockdown, mostly because he was an introvert who had always loved spending time indoors. however, he mentioned that he and his mother (whom he was living with) had been adhering to the usual precautions to avoid the virus. he said that he had always been protective about his health, as was his mother. theme 2: impact of the lockdown/covid-19 on lifestyle. almost all participants (11/14) described how their daily routines had changed because of the lockdown. while many of them were trying to keep themselves healthy by striving to live a "normal" life, a few mentioned about drastic changes in lifestyle, especially in their eating and sleep patterns. for instance, tulika shared how she lost motivation to stick to a routine during the lockdown. rumi however understood this "bad time" as an opportunity to work on herself, a point shared by three other participants. she believed that she could explore a completely different side of hers because of this long break from her busy work. she narrated, i have been experimenting with my life these days. i wake up early and do yoga and then meditation, and then helped my mother with some household chores. currently, i think i feel physically very light, maybe because of exercise. the free time has helped me a lot to explore these, which otherwise was not really possible given the busy life that i had. theme 3: coping with challenges. each participant shared unique stories of coping with the crisis. while most adapted themselves to the "smaller world," a few struggled with it and found alternate ways to negotiate their challenges. a few other participants were positive about the crisis, spending time relaxing or pursuing long-held passions. for instance, rumi described her introspective exploration and enhanced ability to connect with the society through solitude and meditation. on the other hand, tulika kept her mind distracted from the "outside chaos" by immersing herself in social media. for instance, tulika shared, the number of hours i am awake, i am using social media. even if i am going to sleep, i will mindlessly keep scrolling until i fall asleep. because these are the places currently where you see people. otherwise it is quite just you. i think it is a good place to connect. it keeps me engaged. tulika's narrative indicates that the online activities have connected her with the outside world, which created an avenue for her to share her vulnerabilities with others through online interactions (messaging and commenting on posts). notably, more than 50% of the participants reported perceiving the internet as a way to reduce stress and anxiety during the lockdown. in fact, most of them referred the internet as the easiest way to keep them distracted from thinking (or overthinking). a few of the participants who shared their frustration about disruption in their sex lives, reported finding solace in watching pornography and masturbating. sujoy (male, belonged to lgbt group, student, 22 years) quoted, i think it is the only thing you could right now. i mean i see people online in grindr (a dating app for lgbt adults), and surprisingly people are still looking for sex. and very honestly, i completely understand this desperateness. most of them, and that includes me, were having our fun days. and all of a sudden this happens. initially, even i had the thought, ghar ke paas to jake hookup kar sakte hain (it should be okay to have a hookup close to my place). but i realized immediately that it is not wise enough to meet people during this time, especially when you know that it (covid-19) could also not show any symptoms[. . .] what i do right now is watch porn and jerk off. while sujoy showed high self-control and overcame his behavioral impulse of going out to have sex, his initial inclination to go out of his home for sex highlights the repercussions of forced abstinence due to the lockdown. similarly, a participant (male, heterosexual, student, 18-25 years old) answering the online quantitative survey shared via the open-ended question: my sexual desires are making me feel more anxious to masturbate, as a single (man), very often during this lockdown period. theme 4: new perspectives on self, life, and society. ten of the 14 participants believed that their lives were no longer the same. they believed that they had changed significantly in terms of how they viewed their selves and society in general. for instance, tulika said, earlier i used to be worried about very little things, and now it has changed drastically, maybe after spending so much time with myself. i am clearer about my life. i feel i got clarity and i see a huge change in myself. i feel spending time with myself has been the best thing. similar observation was made by rumi, who pointed out that, [. . .] my mental state is also is very different than before. earlier i used to get irritated, worried, angry frequently, but now i feel that i am quite easy, surprisingly, even with my husband. we used to fight (giggles) but now i see that i have been very much calm even with him. it is more because of a lot of things, like about spending time with myself. the universal vulnerability, extended time to reflect on their lives, and sharing their vulnerabilities with others had increased the participants' level of compassion and social empathy, which further strengthened their relationships with their loved ones. this was reflected in what tulika shared, what i have learnt from this entire situation is that you do not take things for granted anymore, even for like interacting with people. i connected with a lot of friends lately, made a couple of new friends as well, and i feel the conversations are no longer superficial but they seem very real, even though we are not in the same physical space but i feel closer to the people more than when we were closer physically [. . .]people have become more vulnerable and have started sharing. everyone is going through some upheaval right now with this feeling and everyone is trying to connect with others. we are in our mid 20s and everyone is going through this time in pretty much (a) similar way, a huge disruption in our lives i would say. so, we become more vulnerable, now i guess these feelings of vulnerability comes out, when you are sharing. and especially when you know that the other person is going through the same as well, and it actually also allows you to connect to the society at large. . .to a wide range of people. similarly, rajan (male, heterosexual, student, 30 years old) reflected, rumi considered the lockdown situation as an opportunity to reflect upon her own life, tried to connect with people around her, spent longer time in spiritual, motivational, and meditational activities-all of which had helped her find meaning in her life and optimism about her marriage. using quantitative data from 282 indian adults and qualitative narratives of 14 adults, our mixed methods study found that even though the covid-19 crisis indiscriminately affected everyone, its psychological effects were disproportionate among diverse social groups in india. our quantitative and qualitative findings both suggest that lgbt adults, compared to the heterosexuals, are at a higher risk of developing anxiety, depressive symptoms, and using pornography during the lockdown. moreover, higher levels of anxiety and depressive symptoms were associated with greater disruption in sleep and food cycles. lastly, our findings unpacked how sharing vulnerability with loved ones, and frequently talking to family members, strengthened social relationships and social empathy among indian adults during the covid-19 lockdown. the higher risk of anxiety in our survey among the lgbt adults than heterosexuals was corroborated by our qualitative findings. several reasons might explain this. first, previous research suggests that the lgbt community have a higher prevalence of anxiety and depressive symptoms compared to heterosexuals [46] , independent of any crisis. this could be explained using the meyer's minority stress model [10] during the lockdown, their minority stress (such as sexual orientation-based discrimination and internalized homonegativity) could interact with the lockdown-related stress, thereby increasing their anxiety much more than that experienced by heterosexuals. second, the lockdown had likely paused their social as well as sexual lives (which connected them with their own community) which likely restricted their access to a safe space, and limited the social support they received from the community [47] . subhash and sujoy's narratives are good examples that show how abstinence from active sex life could make the lgbt adults more anxious during the lockdown. while sujoy's example reflects the high demand of self-control during the covid-19 lockdown [48] , research suggests that the application of self-control is effortful and aversive [49] . therefore, individuals with low self-control could be at a higher risk of succumbing to their behavioral impulses during the pandemic. this also explains our qualitative finding which suggests a higher likelihood of compulsive consumption of pornography and greater frequency of masturbation among lgbt adults (and people in same-sex relationships) during the lockdown compared to heterosexuals (and people in opposite-sex relationships). this disruption in sexual life could explain our quantitative finding which suggested greater anxiety among heterosexuals who were in (opposite-sex) relationships. the lockdown could have resulted in restrictions in physical interactions and romantic dates with their partners, and reduced the social support received, thus increasing their anxiety. however, the higher consumption of pornography and frequency of masturbation might suggest a healthy sexuality. notably, these practices may enable individuals to stay away from seeking physical intimacy outside the home during the lockdown which could contribute towards curbing the spread of covid-19 [12] . lastly, during the lockdown, it is likely that most adults would move closer to their families for support and to avoid loneliness [50] , especially in a family-centric country such as india. however, for many lgbt adults moving in with their parents, to whom they were not out or who disapproved of their sexuality, could be challenging, increasing their risk of experiencing hostility during the lockdown. previous studies have shown that parental support and familial environment play crucial roles in self-acceptance among lgbt adults [51, 52] . the lack of such familial and/or parental support could hinder self-acceptance among lgbt adults, see for instance, anurag's narrative of how he could accept his sexuality only after he moved out of his parental home. moreover, moving away from his "safe space" to a place which brought memories of discomfort likely increased his anxiety. this could be true for several of the lgbt adults who had gone through interpersonal and familial conflict earlier. our quantitative findings also found that individuals at greater, versus lower, risk of the complications of covid-19 showed higher levels of anxiety. a previous study suggested that patients with existing risk factors to covid-19 such as cardiovascular disease (cvd) were more also more likely have worse health outcomes if infected [53] . irrespective of worse health outcomes, belonging to a group with increased risk of covid-19 complications, given that covid-19 has no known cure and unpredictably causes mortality, could potentially induce additional stress and anxiety. this corroborates findings from previous studies suggesting a higher prevalence of stress among front-line health workers [17] , elderly persons [54] , and people living with hiv [55, 56] during the global public health crisis. however, our qualitative findings are in contrast to this finding. ashish (who was living with hiv) showed no added concern (or anxiety) due to the lockdown. one explanation for this could be that ashish was among those who adopt optimism and, in combination with constant precaution, show stronger resilience to adverse situations. findings from a previous study found that people living with hiv could develop resilience despite their physical and psychological challenges [57] . in fact, our study found that optimism and resilience coping were negatively related to anxiety and depressive symptoms. moreover, ashish enjoyed spending time indoors, which likely reduced any frustration related to not being able to enjoy regular life, in addition to the lower likelihood of contracting covid-19. we did not find quantitative evidence supporting the hypothesis that living in a state with a higher count of covid-19 cases predicted greater anxiety and depressive in indian adults. this is in contrast to previous research in australia which found that respondents living in areas with a high number of influenza cases were at much greater risk of stress than those living in uninfected areas [18] . however, our qualitative results supported our hypothesis. the lack of evidence in our quantitative findings could be because of the operationalization of the concept of area. our study operationalized area-level risk at the state level. it is possible that anxiety was higher among people living in a neighborhood (and not the state) with higher number of covid-19 cases. also, in addition to just the count of covid-19 in the neighborhood (or state) including the infection fatality rate in the operationalization could have given a reliable estimate of the influence of place. our quantitative findings suggest that a past history of depression or loneliness could increase anxiety and depressive symptoms during the lockdown. our qualitative findings corroborate this. tulika and salma's narratives suggest that stress during lockdown could revive past trauma. previous study findings also support this interpretation [58] . our qualitative and quantitative findings also suggested that increased depressive symptoms in this group could also increase their internet consumption leading to internet addiction during the lockdown. depressed individuals could use the internet as way to cope with their negative psychological state during the lockdown, which risks addiction during a restrictive state such as a lockdown and could affect their quality of life even after the lockdown. anxiety and depressive symptoms during lockdown were found to predict disruptions in sleep and food schedules, corroborating findings from a previous study [59] . qualitative data found that an increase in anxiety and a lack of motivation to lead a routine life increased the risk of an unbalanced sleep cycle, which also impacted food consumption and its timings. additionally, the inaccessibility to quality food due to the restrictions in physical mobility during the lockdown could affect the balanced diet among indian adults. beyond any immediate health effects which could further worsen mental health, such prolonged changes in timing and consumption of food could impact their overall food eating patterns even after the lockdown, resulting in poorer physical and mental health in the longer term as well. our quantitative findings suggest that sharing about stress with loved ones and an increase in frequency of interacting with family members likely strengthened social bonds and also increased social empathy among indian adults. our qualitative findings elucidate this. tulika's narrative highlighted that the universal vulnerability due to the global pandemic and her sharing about it with others in a similar situation improved her connectedness with the people thereby strengthening her social relationships. similarly, interacting and knowing the vulnerability of rajan's parents (more vulnerable) made him more empathetic, and increase his connection with the society at large. this fits with the findings from a previous study that highlighted this argument-sharing and expressing emotions (and vulnerabilities) could make people more empathetic [60] . such increased social empathy could also be a positive response to the pandemic (and lockdown). for instance, a recent study from the west found that higher empathy towards the more vulnerable could induce motivation to maintain and promote social distancing [61] . there are several limitations of this study that need to be noted while interpreting the results. we used convenience sampling in our study that limits the generalizability of the findings. generalizability is also limited due to our use of an english language questionnaire, a small sample size, and choosing the online mode of administering it. however, despite the modest sample size and the sampling design, we were able to show several interesting findings with statistical confidence. we were also constrained in our ability to seek non-english speakers and administer the questionnaires using hard copies due the lockdown situation. notably, our use of a qualitative strand strengthened the interpretation of the quantitative findings by allowing us to unpack several complex processes. furthermore, we speculate that the associations under study are stronger among those sub-groups who have a limited access to the internet that plays a crucial role in ensuring social connectedness during the lockdown. we used psychological scales to measure anxiety, depressive symptoms, internet addiction, and compulsive consumption of pornography, instead of clinical interviews which would have yielded medical diagnoses. however, our use of widely cited, reliable scales are informative and could indicate symptoms of the psychological outcomes we explore. additionally, due to scarcity of available validated scales, we could not use scales that were validated for use in the indian population. however, we have used scales that have been frequently administered in the indian context earlier. we also found high internal consistency of the scales in our study suggesting their reliability. another limitation of our quantitative study is the use of several shorter and single-item scales (such as the brief resilience coping scale). longer scales could have yielded robust results. because our pretest suggested that the length of the questionnaire was perceived as "a lot" we chose to use shorter scales. for instance, we used a single item to measure self-reported change in food consumption patterns and used two items to measure self-reports of sleep disturbances. however, the items used in our study captured the perception of the participants about the change in their food and sleep cycles during the lockdown, which adds an informative nuance. additionally, our qualitative data around these measures add more details and increase our confidence in interpreting these findings. lastly, the cross-sectional nature of the study limits our ability to make causal claims. longitudinal studies with frequent follow-ups during the lockdown could have shed light on causal processes. however, we were grateful to be able to recruit a diverse sample for our quantitative and qualitative strands, which allowed us to explore the differences in the psychological outcomes during the lockdown across different groups in india. despite these limitations, our mixed methods findings highlight the additional psychological burden that the lockdown has brought to an invisible group, the lgbt adults. to our knowledge this is the first study to look at the differential psychological impact of the lockdown across different social groups (including sexual orientation) in india. moreover, our use of qualitative narratives allowed us to understand the processes linking several social factors to the psychological outcomes in a nuanced manner. our study also highlights a few positive aspects of the lockdown, underscoring the increase in social empathy and strengthened social bonds among indian adults. our findings echo balagos' argument that the marginalization of lgbt adults would be heightened during disasters, because existing inequalities are magnified at such times [62] . while the indian supreme court decriminalized homosexual acts in 2018, indian policies are not yet inclusive of lgbt adults, who remain socially invisible. our findings call for the attention of counsellors and health professionals in understanding the specific psychological needs of the lgbt adults during such crises and providing services accordingly. this study highlights the need for regular interaction and emotional support from friends, family, partners, and caregivers of lgbt adults, individuals with a history of depression or loneliness, a higher risk of developing complications if they contract covid-19. a recent study [63] highlighted the promise of delivering psychological support through online-and telecounselling. this study warrants the use of such technologies in an inclusive manner. the study also opens avenues for researchers to further investigate the extent and nature of the psychological impact in such marginalized groups during crises or disasters. lastly, our study findings provide evidence for mental health policymakers to begin designing inclusive policies to address the concerns of marginalized groups during and in the aftermath of the covid-19 global crisis. all in all, our study highlights the differential psychological effect of the covid-19 pandemic among lgbt adults, groups with history of depression, and those with high-risk of covid-19 complications. the study thereby urgently calls for the attention of policymakers to take sensitive and inclusive health decisions for the marginalized and the vulnerable, both during and after the crisis. supporting information s1 file. covid-19 data. (xls) statement on the second meeting of the international health regulations (2005) emergency committee regarding the outbreak of novel coronavirus (2019-ncov) coronavirus disease (covid-19) covid-19: real-time dissemination of scientific information to fight a public health emergency of international concern india coronavirus lockdown | day 1 updates march 25, 2020-the hindu understanding the school community's response to school closures during the h1n1 2009 influenza pandemic the public's response to severe acute respiratory syndrome in toronto and the united states sars control and psychological effects of quarantine the relevance of psychosocial variables and working conditions in predicting nurses' coping strategies during the sars crisis: an online questionnaire survey 2019-ncov pandemic: a disruptive and stressful atmosphere for indian academic fraternity prejudice, social stress, and mental health in lesbian, gay, and bisexual populations: conceptual issues and research evidence impact of the covid-19 pandemic on the sexual behavior of the population. the vision of the east and the west sexual health in the sars-cov-2 era less sex, but more sexual diversity: changes in sexual behavior during the covid-19 coronavirus pandemic infographic: indians watching more porn during covid-19 lockdown | india news-times of india covid-19 and sexuality: reinventing intimacy pornography use and loneliness: a bidirectional recursive model and pilot investigation factors associated with mental health outcomes among health care workers exposed to coronavirus disease factors influencing psychological distress during a disease epidemic: data from australia's first outbreak of equine influenza impact of anxiety and depression on physical health condition and disability in an elderly korean population sleep disorders as core symptoms of depression binge eating disorder mediates links between symptoms of depression, anxiety, and caloric intake in overweight and obese women queering india: same-sex love and eroticism in indian culture and society the relation between work-family balance and quality of life resilient coping moderates the effect of trauma exposure on depression optimism and its impact on mental and physical well-being the psychological distress and coping styles in the early stages of the 2019 coronavirus disease (covid-19) epidemic in the general mainland chinese population: a web-based survey covid-19 and lockdwon: a study on the impact on mental health study of knowledge, attitude, anxiety & perceived mental healthcare need in indian population during covid-19 pandemic a framework for design: qualitative, quantitative, and mixed methods approaches designing and conducting mixed methods research a brief measure for assessing generalized anxiety disorder: the gad-7 screening for depression in well older adults: evaluation of a short form of the ces-d methods for computing missing item response in psychometric scale construction imputing cross-sectional missing data: comparison of common techniques internet addiction: the emergence of a new clinical disorder a brief scale to measure problematic sexually explicit media consumption: psychometric properties of the compulsive pornography consumption (cpc) scale among men who have sex with men the brief resilience scale: assessing the ability to bounce back distinguishing optimism from neuroticism (and trait anxiety, self-mastery, and self-esteem): a reevaluation of the life orientation test the robustness of test statistics to nonnormality and specification error in confirmatory factor analysis stata statistical software: release 14 using thematic analysis in psychology case study research in education: a qualitative approach from text to codings member checking: a tool to enhance trustworthiness or merely a nod to validation? prevalence of depression and anxiety among bisexual people compared to gay, lesbian, and heterosexual individuals:a systematic review and meta-analysis gay, lesbian, and bisexual youth and young adults: social support in their own words too bored to bother? boredom as a potential threat to the efficacy of pandemic containment measures task duration and task order do not matter: no effect on self-control performance coronavirus drives some millennials home to their parents-vox the influence of family environment factors on self-acceptance and emotional adjustment among gay, lesbian, and bisexual adolescents gay, lesbian, and bisexual youths coming out to their parents: parental reactions and youths' outcomes cardiovascular disease and covid-19 physical distancing in covid-19 may exacerbate experiences of social isolation among people living with hiv notes from the field symptoms, stress, and hiv-related care among older people living with hiv during the covid-19 pandemic motivation, management, and mastery risk for recurrence in depression sleep and anxiety disorders predictors of empathy in women social workers the emotional path to action: empathy promotes physical distancing during the covid-19 pandemic the warias of indonesia in disaster risk reduction: the case of the 2010 mt merapi eruption in indonesia the need for a mental health technology revolution in the covid-19 pandemic ajs and mas thank nilesh thube (nvt) for his diligent contribution in data management and representation of results. ajs conveys special thanks to dipankar dutta for being extremely patient during the work. ajs and mas thank harvansh dandelia, rakshit verma, and several other friends and colleagues who helped in the circulation of the online survey in a short time. ajs thanks his bula da. ajs is grateful to all the participants for filling up the form and sharing their emotions during this crisis. conceptualization: anupam joya sharma, malavika a. subramanyam.data curation: anupam joya sharma. key: cord-275071-2uiaruhg authors: balmford, ben; annan, james d.; hargreaves, julia c.; altoè, marina; bateman, ian j. title: cross-country comparisons of covid-19: policy, politics and the price of life date: 2020-08-04 journal: environ resour econ (dordr) doi: 10.1007/s10640-020-00466-5 sha: doc_id: 275071 cord_uid: 2uiaruhg coronavirus has claimed the lives of over half a million people world-wide and this death toll continues to rise rapidly each day. in the absence of a vaccine, non-clinical preventative measures have been implemented as the principal means of limiting deaths. however, these measures have caused unprecedented disruption to daily lives and economic activity. given this developing crisis, the potential for a second wave of infections and the near certainty of future pandemics, lessons need to be rapidly gleaned from the available data. we address the challenges of cross-country comparisons by allowing for differences in reporting and variation in underlying socio-economic conditions between countries. our analyses show that, to date, differences in policy interventions have out-weighed socio-economic variation in explaining the range of death rates observed in the data. our epidemiological models show that across 8 countries a further week long delay in imposing lockdown would likely have cost more than half a million lives. furthermore, those countries which acted more promptly saved substantially more lives than those that delayed. linking decisions over the timing of lockdown and consequent deaths to economic data, we reveal the costs that national governments were implicitly prepared to pay to protect their citizens as reflected in the economic activity foregone to save lives. these ‘price of life’ estimates vary enormously between countries, ranging from as low as around $100,000 (e.g. the uk, us and italy) to in excess of $1million (e.g. denmark, germany, new zealand and korea). the lowest estimates are further reduced once we correct for under-reporting of covid-19 deaths. electronic supplementary material: the online version of this article (10.1007/s10640-020-00466-5) contains supplementary material, which is available to authorized users. sars-cov-2, the virus which causes the covid-19 disease, is a zoonotic pathogen which emerged in wuhan in late 2019 (huang et al. 2020) . at the time of writing, in early july 2020, it had already claimed the lives of over half a million people globally (beltekian et al. 2020 ). in the usa covid-19 deaths now exceed the number of us military deaths arising from all conflict since the second world war (statista 2020) while in the uk the four weeks to 24th april saw more londoners lose their lives to covid-19 than during the deadliest four week period of the blitz (morris and barnes 2020) . this death toll is only the extremely saddening tip of the much larger iceberg of disruption that covid-19 has caused and continues to cause. confirmed cases across the world now exceed eleven million (beltekian et al. 2020 ) and the true infection rate is likely far higher. each case imposes a real cost on every infected individual. while symptoms may sound innocuous, including a dry cough, fever, and tiredness (who 2020a; verity et al. 2020) , longer term this morbidity is likely to impose significant costs on sufferers' health, including potentially permanent lung damage or fibrosis associated with impacts upon the heart, kidneys and brain (citroner 2020) , all of which are likely to have negative consequences for future well-being and productivity. moreover, alongside the vast disruption that the virus itself has caused directly, preventative measures have caused further disarray in the economy. at present, there are no known specific treatments or available vaccines to either cure or prevent covid-19 infections (who 2020b). therefore governments world-wide have relied upon preventative measures which aim to reduce the number of people exposed to the virus, and lower the effective reproductive number (the average number of new cases per infection, known as r), ideally suppressing it below a value of 1 at which point the number of active cases decreases over time (ferguson et al. 2020) . while some of these measures impose relatively little personal or economic cost (such as simple hand hygiene and the use of face masks), the failure of such measures to stem the rapid world-wide spread of the virus has necessitated international "stay at home" lockdown requirements, entailing significant impacts across the global economy. the international monetary fund (imf) predicts a contraction in global gdp of three percent in 2020-a decline of 6.4% relative to its october 2019 forecast-and a decrease which it describes as being "much worse than during the 2008-2009 financial crisis" (imf 2020a). short term effects are even more extreme. for example, in the uk, gdp fell by 20.4% in april 2020 (ons 2020a), while those claiming unemployment benefits rose nearly 70% to over 2 million (ons 2020b), although even this is dwarfed by the 200% increase in us unemployment over the same period (aratani 2020). 1 globally sovereign debt is also soaring: predicted to grow nearly 20% to $53 trillion in 2020 (standard and poor 2020) as administrations around the world race to protect cash-strapped companies from going out of business in order to prevent further unemployment. at the human level, lives and livelihoods have been turned upside-down. hence the true economic costs are more diverse and quite possibly more severe than that captured by financial metrics alone. they include negative ramifications for people's mental health (pancani et al. 2020; chaix et al. 2020; branley-bell and talbot 2020) ; increased 1 kurmann et al. (2020) note that small business employment contracted by 60% (over 18.2 million) between mid-february and mid-april 2020 since when over 9 million had been rehired to the end of june 2020. prevalence of domestic violence (mclay 2020); and likely reduce the educational achievement of today's children (pinto and jones 2020; van lancker and parolin 2020) . as with previous financial crises (hoynes et al. 2012 ) and pandemics (nikolopoulos et al. 2011) , the virus and the economic fall-out are disproportionately affecting people from disadvantaged groups and lower-income households. black, asian and minority ethnic people are more likely to be infected and die (bhala et al. 2020; garg 2020; khunti et al. 2020; yancy 2020; public health england 2020) ; and lower-income households are less likely to be able to work from home, so face greater negative income shocks (hanspal et al. 2020; hensvik et al. 2020) , just as poorer countries are likely to suffer more than richer nations (hevia and neumeyer 2020) . as is well known, different countries have had very different death tolls. the usa currently has the highest death toll in the world, already exceeding 130,000 deaths (as of 4th july 2020). 2 in contrast, vietnam-which recorded its first case just 4 days after the usais yet to experience a single death. understanding what drives these differences is clearly crucial, potentially enabling improved responses to the continuing covid-19 outbreak and future pandemics. this paper begins to answer the critical question of why different countries have suffered different death rates, and what we can learn for future policy. the remainder of the paper is set out as follows. in sect. 2 we first compare the numbers of deaths attributed to covid-19 across all oecd countries. the paper briefly focusses upon the uk as an example of a broader pattern; that public reporting of numbers related to the pandemic can be somewhat misleading. next, we control for any within-country under-reporting by analysing the overall increase in all deaths above what would be seasonally expected. assessing these 'excess deaths' data suggests that in most nations for which information is available official reporting of covid-19 tends to explain most of this unexpected mortality. however, analysis also reveals some clear exceptions, such as in the netherlands, spain and the uk where more than 40% of all covid-19 deaths seem likely to have not been counted as such. addressing such reporting problems is an essential element of providing the informational base required for an evidence-based policy response to this and any future pandemics. in sect. 3 we assess the impact of government decisions regarding lockdown, their effectiveness and the policy trade-off between economic activity and health risk that they reveal. accepting that they are a conservative estimate of the total impact of the pandemic, officially attributed covid-19 deaths are used to investigate the price of life implied by lockdown policies. first we use a simple regression analysis to show that differences in mortality rates between countries are not driven by factors which are beyond the short term control of policy makers-such as differences in income and equality which, at least within the time available to fight coronavirus are effectively fixed. this in turn allows us to examine the degree of control which policymakers do have at their disposal, such as the rapidity of lockdown imposition and the duration of such controls. we use country-specific susceptible-exposed-infected-recovered (seir) models, similar to the approach of ferguson et al. (2020) , to ask how changes in the timing of lockdown measures affect the current death toll. our analyses provide good evidence that these policy tools actually determine the majority of variation in covid-19 impacts between countries. finally, we link these estimates to financial data to reveal a huge variation in the implied price of life across countries. section 4 concludes. table 1 presents the number of tests, cases and deaths that are officially recorded as (at least in part) caused by covid-19 across all oecd countries as of 9th june 2020 (data from our world in data; beltekian et al. 2020) . as mentioned, and considered in greater detail subsequently, these official estimates are likely to under-estimate deaths from covid-19. however, the degree of under-reporting is far from constant across countries. for example, while almost all countries only counted deaths which had been confirmed to be linked to covid-19, belgium adopts a much broader approach also including deaths where covid-19 is merely suspected as a contributory factor (chini 2020) . this results in much higher death rates than in other countries. arguably adopting the belgian approach internationally might provide a more accurate picture of covid-19 mortality. it is worth drawing attention to the very substantial variation in tests, recorded covid-19 case numbers and official death tolls across countries. adjusting for population, iceland has undertaken far more testing per capita than any other oecd country, at over 183 k/ million compared to just 2 k/million in mexico. much media attention has been expended upon reporting cumulative covid-19 numbers in each country. in terms of cases the roughly 2 million cases reported in the usa is indeed a prominent result. however, unsurprisingly it is the total numbers of deaths by country which has attracted more attention and again the us total of well over 100,000 deaths is eye-catching. however, this media and policy-maker focus upon totals disguises the true comparison of these figures in failing to make even the most basic of adjustments for variation in population size between countries. once this is done then the death rate per million shown in the final column of table 1 reveals a substantially different story. here we need to rule belgium out of comparison because its addition of suspected covid-19 deaths to the confirmed deaths reported by other countries, upwardly inflates its death rate. given this, the death rate reported in the uk is the highest amongst all of the oecd, exceeding even those of spain and italy which experienced their first major outbreaks much earlier on in the pandemic. it is worth highlighting how reporting elsewhere can be somewhat misleading. we do so by focussing on the uk as this is the country we are most familiar with, but the story is highly likely to be similar elsewhere. figure 1 graphs the development of total recorded deaths (vertical axis) for a selection of 10 countries over roughly the first 100 days since each country recorded its 50th death (horizontal axis). this graph and its selection of countries is dictated by that which the uk government chose to highlight for comparison at its daily coronavirus press briefings. 3 setting aside for the moment the us trend, clear separation can be observed between those countries such as germany and korea, which rapidly entered into lockdown and quickly controlled the growth of the virus, and those countries such as the uk and spain, 3 the figure is a redrawing of one which was displayed daily at the uk press briefing from 30th march 2020 until being left out of daily briefings from 10th may 2020 onwards. speeches by the prime minister on covid-19 had been conducted before then (for example on the 9th and 12th march) but they only became a daily occurrence with a relatively standardised format from 16th march onwards. slides from these briefings are available here: https ://www.gov.uk/gover nment /colle ction s/slide s-and-datas ets-to-accom pany-coron aviru s-press -confe rence s. where lockdown was delayed resulting in a higher plateau. this is the first indication of the positive effects of early lockdown action, which we consider further subsequently. the uk government's decision to only display the total number of deaths in each of the countries shown took no account of even basic differences between countries such as population size; and as table 1 has already shown, this makes fair comparison of death rates difficult. it might seem unusual to fail to make such basic adjustments, however the choice of such a display by the government is one which shows the uk cumulative total initially below that of european neighbours such as italy and spain and consistently dwarfed by that for the us, rising to more than twice the uk level. the fact that the us population is more than five times that of the uk, and that therefore per capita rates were much higher in the uk, is not obvious in this display. during the early days of the coronavirus outbreak, this omission of per capita data and focus upon cumulative totals allowed the uk government to make cross country comparisons which indicated that the country appeared to be faring better than many international counterparts (such sentiments are clear in transcripts of the verbal explanation which accompanied the graph, presented in online appendix 1). for example, on the 1st april, the graph was described by the uk government as showing "it has not been as severe here as in france, and we are just tucked in under the usa and obviously italy on a different trajectory". however, as the pandemic developed so the performance of the uk relative to these other countries worsened. this situation was exacerbated by an outcry against the uk government's use of statistics based only upon deaths within hospitals rather than also including those in the community, ignoring obvious discrepancies such as a clear rise in deaths within care homes into which elderly hospital patients had been moved without testing for coronavirus (discombe 2020; grey and macaskill 2020) . shifting to reporting deaths from all settings revealed that the uk was faring far worse than nearly all other cumulative deaths (vertical axis) plotted for various countries (as selected for comparison in uk government briefings) over approximately the first 100 days since each country recorded its fiftieth death (horizontal axis). note that spain's apparent decrease in cumulative deaths around day 70 is an artefact of their reporting problems countries and indeed in per capita terms was experiencing one of the highest death rates globally (beltekian et al. 2020) . the impact upon the official narrative presented at uk press briefings was swift and noticeable. while initially much emphasis had been placed upon the uk's apparently favourable performance compared to other nations, now government officials started to mention the difficulty of making cross country comparisons, as highlighted by the pink dots at the top of fig. 1 (and data presented in online appendix 2). 4 these caveats increased in both regularity and stridency until, on 10th may 2020, cross country comparisons were removed from government press conferences. we have no reason to suspect that the uk government was unique in attempting to provide a positive representation of trends. however, a failure to provide clear and objective information is a well acknowledged cause of mistrust in authority (kavanagh and rich 2018) and is corrosive to public life at any time, but particularly in a pandemic where trust in institutions is vital. in undertaking cross-country comparisons of the impacts of covid-19 a first issue to be tackled is the difference in national approaches to reporting. this can be seen even in the reporting of testing statistics, differences which some authorities have argued may be politically motivated (norgrove 2020) . likewise, some countries (e.g. belgium) are far more likely than others to ascribe a death as caused by covid-19 (chini 2020) . given these concerns, we complement our comparisons of official covid-19 statistics with analysis of patterns in excess mortality data. here we define excess mortality for a country as the deviation in mortality rate during the period january to april 2020 compared to a baseline of expected deaths from previous years. excess mortality data is therefore not biased by differential rates of covid-19 testing or legislation on ascribing cause of death. there are however important caveats to the excess mortality figures. such numbers do not exclusively capture the increase in mortality that is directly caused by the presence of the novel virus. in addition, people may be less likely to visit hospital and therefore less likely to get treated for what are, in normal times curable diseases, thus tragically dying at a higher rate (thornton 2020) . similarly, first response services may get overwhelmed and therefore be less able to respond to life threatening emergencies such as heart attacks and strokes, again causing higher than expected death rates (oke and heneghan 2020) . acting in the opposite direction, government responses to coronavirus such as lockdown, may reduce the number of deaths from other causes; transmission rates for other communicable diseases are likely to be suppressed while a reduction in travel reduces the mortality associated with traffic accidents (alé-chilet et al. 2020). it is therefore not a priori obvious whether excess mortality is positive or negative. nonetheless, comparison of excess mortality with official covid-19 deaths will provide a more informed picture of the overall impacts of the pandemic within and across countries. table 2 presents excess mortality data for the subset of oecd countries for which it is available. in general, the data are from the economist (2020) but are supplemented for some countries by data from other sources. 5 baseline mortality is typically calculated as the mean number of deaths occurring in january-april 2015-2019. 6 excess deaths are calculated as the difference between the number of deaths observed in january-april 2020 and baseline mortality. the final column is the ratio of excess death to cumulative deaths at the end of april for each country, as reported by our world in data (beltekian et al. 2020 ), calculated as: the heterogeneity that was present in the statistics of officially recorded covid-19 deaths is also present in the excess mortality data. some countries, such as austria, iceland and portugal see only very marginal increases in death rates as compared with background death. there are countries which appear to do even better; denmark, finland, germany, israel and norway all observing fewer deaths than expected. as discussed above, these negative excess death numbers could be the result of measures to combat covid-19 reducing other-cause mortality, or from previous years used to calculate the baseline number of deaths being particularly bad. indeed 2020 does seem to have been a year with relatively few deaths from influenza (center for disease control 2020). at the other extreme, countries which appear worst hit based upon the officially recorded per capita death data are also those experiencing the highest percentage increase in mortality: belgium, spain and the uk all record deaths that are more than 15% higher than expected. note that italy too may well have been in this list, but the data for italy is only available to 30th march, about the time the country experiences its peak daily mortality. turning to the ratio of excess deaths to officially reported deaths, again there appears considerable heterogeneity across countries, suggesting countries are indeed measuring the death toll from the pandemic by very different yard sticks. generally, countries officially reporting high deaths tolls are also those which have the highest ratio of excess deaths to officially reported deaths. indeed, austria, iceland and portugal report more covid-19 deaths than the excess deaths they experience. it is worth noting this is not to say that these (1) ratio = excess deaths∕officially reported deaths 5 the other data sources used for particular countries are: austria -http://www.stati stik.at/web_de/stati stike n/mensc hen_und_gesel lscha ft/bevoe lkeru ng/gesto rbene /index .html); belgium-https ://epist at.wivisp.be/momo/; finland -https ://pxnet 2.stat.fi/pxweb /pxweb /en/kokee llise t_tilas tot/kokee llise t_tilas tot__vamuu _koke/statfi n_vamuu _pxt_12ng.px/; iceland-https ://hagst ofa.is/utgaf ur/tilra unato lfrae di/danir -tt/; ireland (note these are death registrations rather than government figures)-https ://rip.ie/death notic es/all; israel-https ://www.healt h.gov.il/units offic e/hd/ph/epide miolo gy/pages /epide miolo gy_repor t.aspx?wpid=wpq7&pn=6; new zealand-https ://www.newsr oom.co.nz/2020/05/05/1,157,173/arethere -hidde n-covid -19-death s-in-nzs-stati stics ; spain (importantly accessed on 11th june, after there was a major addition to the figures)-https ://www.scb.se/conte ntass ets/edc2b 33f85 ad415 d8e79 09002 253ed 84/2020-04-09%e2%80%94pre limin ar-stati stik-over-doda-inkl-eng.xlsx; usa-https ://data.cdc.gov/ nchs/exces s-death s-assoc iated -with-covid -19/xkkf-xrst. even among the countries for which data is available, mortality data are only available for a few months of the year, generally at least to the end of april, hence the focus january-april 2020 deaths. data tend to be aggregated to the week level, hence the exact endpoint is rarely 30th april 2020. rather, the last day used in 2020 is determined by the data availability, and chosen to be as close as possible to 30th april. in all cases, we compare like-for-like, such that the baseline deaths are recorded over the same time period. likewise, the cumulative death toll we use to calculate the ratio of excess to reported death is that which was officially reported on the last day of the 2020 mortality data we use for each country. 6 for some countries data availability means this is not possible. for austria, belgium and germany it is 2016-2019; iceland and usa use 2017-2019; for spain baseline deaths are modelled by momo. countries are recording deaths as covid-19 when they are not; rather it is entirely plausible the interventions to prevent covid-19 in these countries have suppressed other deaths too. at the other extreme, some countries, notably the netherlands, spain and the uk, have ratios which imply upwards of 40% of covid-19 deaths that are occurring are not being officially recorded. there are of course outliers to the overall pattern. belgium, france and sweden, have ratios below 1 despite having high per capita death tolls. likewise, chile and new zealand have very high ratios, but these are almost certainly an artefact of them having so few covid-19 deaths by the end of april, rather than because of under-reporting in each nation. to recap, there are vast differences in the number of cases and deaths caused by coronavirus in different countries. this heterogeneity does not merely disappear when we account for potentially different reporting guidelines in each country; rather it may even be exacerbated. so what could be driving these patterns? while most countries chose to implement a relatively similar policy response, they did so at different times in their respective pandemics and some have been criticised for only belatedly imposing lockdown. 7 there is some early correlative evidence that differences in current death tolls could be explained by lockdown date (burn-murdoch and giles 2020) and we now move to consider this issue in greater detail. our investigations of the potential impact of different approaches to reporting show the usefulness of an internationally agreed standard for assessing the impact of the pandemic. however, in the absence of such a standard we use national official estimates of covid-19 mortality to understand the impact of lockdown policies. data is supplied by the our world in data programme (beltekian et al. 2020 ). an initial task was to estimate the overall impact which policy responses could plausibly have had on covid-19 mortality. to achieve this we undertook regression analysis examining the extent to which variation in covid-19 deaths across all 37 oecd countries might be explained by socio-economic and demographic differences which no government could reasonably be expected to address during the timescale of a pandemic. a number of such exogenous determinants have already been highlighted in the literature. of these one of the most clearly established mortality risk factors is a positive association with age; all other things considered, older sufferers are more likely to die from contracting covid-19 than are younger people (dowd et al. 2020) . therefore, across countries, populations which include a greater proportion of elderly people are likely to report higher death tolls. similarly, those living in closer proximity to others may be more likely to pass on and contract the respiratory disease, hence variation in population density across nations may be a determinant of covid-19 deaths (rocklöv and sjödin 2020) . beyond simple average population density, the degree to which populations are clustered in large urban centres may influence covid-19-related mortality (stier et al. 2020) . health outcomes might also differ because of within-country variation in wealth (marmot 2005) which we capture in our regression by controlling for the gini coefficient of income inequality for each country. richer nations are likely better placed to limit the spread of pandemics (e.g. hosseini et al. 2020 ), hence we use per capita gdp as a regressor to net-out cross-country differences owing to wealth. finally, previous studies (e.g. fraser et al. 2004) have highlighted that early detection may play a crucial role in halting virus spread, hence it seems plausible that countries which were exposed to covid-19 earlier in the pandemic, and that therefore had less time to prepare, faced worse consequences. to account for this, we use the regressor "warning days"-the length of time (in days) between the who declaring that the covid-19 outbreak was a "public health emergency of international concern" on 30th january 2020 and the country recording its 100th confirmed case (who 2020c). the linear regression we use, details of which are presented alongside full results in online appendix 3, is deliberately simple and we are not claiming that the model necessarily captures causal relationships. however, even after including the list of exogenous factors which have been hypothesised to be major socio-economic and demographic drivers of cross-country variation in mortality rates, over 75% of the cross-country variation in covid-19 mortality differences remains unexplained. covid-19 deaths vary greatly across countries due to factors beyond socio-economics and demographics; the major remaining determinant is the policy responses implemented by national governments of which the most obvious difference is when different countries implemented lockdown. 8 to investigate the impact of lockdown upon cross-country variation in covid-19 mortality we calibrate country-specific seir models. seir models have a long history of development (li and muldowney 1995) with applications across a variety of infectious diseases including measles (bolker 1993), hiv (shaikhet and korobeinikov 2016) and ebola (lekone and finkenstadt 2006) . more recently seir models have also been applied to covid-19 (e.g. annan 2020; flaxman et al. 2020; pei et al. 2020) . however, as far as we are aware, ours is the first study to use the seir modelling framework to examine the effects of lockdown timing across multiple countries in the same study, and the first to combine these results with financial forecasts to obtain cross-country implied price of life estimates. price of life estimates derived in this paper are of critical importance given that government intervention has the ability to save life, yet trades-off against other goods. for example, closing schools is expected to reduce the transmission of infectious disease, hence decreasing the number of lives lost in a pandemic by imposing a human capital cost on today's children (viner et al. 2020) . likewise, there is evidence that the more stringent the government intervention to reduce the spread of coronavirus, the fewer lives that have been lost (stojkoski et al. 2020 ). this too is not free: we all pay with restrictions on our basic freedoms. beyond coronavirus, governments spend money and introduce legislation which imposes significant costs on society in a variety of sectors: healthcare (nice 2012), road safety (dft 2016), and safety at work legislation (hse 2016). governments also often have to consider multiple policy options for issues of environmental concern, be that considering pollution (ackerman and heinzerling 2002) , climate change (stern 2007) or biodiversity loss (ellis et al. 2015) . here too, lives can be saved and lost as a consequences of policy decisions. hence understanding how governments should value life is of critical concern. indeed, a significant section of relevant policy documents is occupied by discussion of the value which a government should place on statistical life when evaluating policy (e.g. the green book; h.m. treasury 2018). in the case of coronavirus, there are already studies which aim to assess the economic value of particular policy interventions by reducing the number of lives lost. hale et al. (2020) ask: how much of one year's consumption would an individual be willing to forgo in order to reduce the mortality associated with covid-19, suggesting the answer lies in the range one-quarter to one-half depending on exact mortality rates. underpinned by assumptions about the rate of transmission and how policies may affect this, greenstone and nigam (2020) show the economic benefit of social distancing measures in the usa to be very substantial-about $8 trillion. similarly, thunström et al. (2020) use initial global estimates for the basic reproductive rate, and assume decreases to transmission from policy intervention from studies on spanish flu, to go further. they conduct a cost-benefit analysis for similar measures, again in the usa, showing that the net benefits exceed $5.2 trillion. gandjour (2020) and holden and preston (2020) conduct similar cost-benefit style analyses for germany and australia, respectively, both highlighting that lockdown comes out net positive. here we ask a different but related question. not whether lockdown makes economic sense, but rather what the timing of interventions reveal about the relative prices different governments place on their citizens' lives. we focus on 9 countries with very different mortality rates and intervention timing-if there are discrepancies between countries for the price of life, they are most likely to be shown in this set of countries. in china, lockdowns were implemented on a province-by-province basis on very different dates. therefore, at the country-level our gdp calculations would be incomparable with other nations. to overcome this challenge, we additionally parameterise an epidemiological model for hubei, the province worst hit by the pandemic. we use the results from hubei in our price of life calculations to maintain comparability across countries. to be clear, the implied price of life should not be regarded as comparable to the value of a statistical life (vsl). 9 specifically, vsl is a concept from normative economicshow much consumption should governments be willing to trade-off for an increase in the number of lives saved. this is a question which can be answered through stated-preference methods as has been done elsewhere (e.g. alberini 2005; carthy et al. 1999; jones-lee 1974) . rather, the implied price of life we calculate can be seen as an answer to the positive economics question of how governments actually do price lives saved in terms of consumption lost when making policy decisions. the key insight is that as the pandemic progressed governments continually had to decide when the moment was right to introduce a lockdown. earlier lockdowns would save more lives, but likely impose greater immediate costs upon the economy. likewise, delaying lockdown also delays the point at which a government becomes either morally or legally responsible for addressing the costs which such restrictions impose upon business. therefore, ex-ante the expectation was that earlier lockdown meant greater financial cost. expost, it seems governments may have been somewhat wrong to make that assumption as longer-term earlier lockdowns actually appear to be associated with shorter overall lockdown length, as is clear in online appendix 4, which in turn result in lower long-term economic costs (balmford et al. 2020). nonetheless, early imposition of lockdown imposed the certainty of cost, while a delay held out the possibility that the epidemic may turn out to be less severe than expected. gambler governments chose to delay rather than act. the chosen date of lockdown reveals a government's preferences regarding the trade-off between avoided deaths and gdp losses. 10 relative to the chosen lockdown date, a later lockdown would have cost more lives, but reduced the financial impact. in its choice of lockdown date a government implicitly accepted the associated gdp loss rather than bear a greater death toll. earlier lockdowns would have had the reverse effect; saving more lives but at a greater cost to the economy. in choosing not to enter lockdown earlier, the government rejected the higher financial cost of earlier lockdown in favour of more deaths. hence, we are able to calculate both accepted and rejected prices for human lives: upper and lower bounds for the implied price of life in each country. 11 a criticism of this method may be that decision makers at the time were unaware of the benefits of lockdown for public health. the evidence, however, points to the contrary. for example, it was reported in the print media at least as early as 7th march that the lockdown in wuhan was showing signs of slowing the spread of coronavirus (qin 2020) . within the uk there is evidence that scientific advisors notified the uk government of the benefits of lockdown two weeks prior to its imposition (barlow 2020) . 12 calculations of the implied price of life for each country require two data points. first, the differential effect on human lives lost from a marginal change in lockdown date. second, the marginal effect on gdp from the same change in lockdown date. we use a compartmental epidemiological model to simulate the epidemic in each country and in particular to predict the outcomes of the counterfactual scenarios in which lockdown dates are changed. in this type of model, at any moment in time the population of a region or country is distributed between compartments according to disease status, and the function of the model is to describe (and predict) how the population flows between these 10 that such a trade-off is inevitable and in principal morally defensible is not questioned, indeed it follows logically from the vsl. increasing economic costs impact upon human welfare. an approach which says that every life is of infinite value would impose infinite costs upon the economy, resulting in far greater losses of human wellbeing (and almost certainly life) than acting in a way which imposes an implicit and non-infinite price on life. it is the cross-country comparison of that implicit price which is examined here. 11 our focus on gdp reflects both the ubiquity of this measure and a lack of available, robust, economic estimates of the wider welfare impacts of lockdown. to better understand some of those wider costs, we direct the interested reader to: branley-bell and talbot, 2020; burki, 2020; cash and patel, 2020; chaix et al., 2020; mclay, 2020; pancani et al., 2020; pinto and jones, 2020; sud et al., 2020; van lancker and parolin, 2020 . while driven out of necessity, we think that a focus solely on gdp is also justified. our interest is in the relative price of life across country. even accounting for the external costs, the relative pattern for price of life would remain; it could only be eroded if these external costs are disproportionately larger for countries with lower gdp-based price-of-life estimates. 12 indeed grant shapps, a uk government minister, was questioned on 16th march 2020, a full week before the uk entered lockdown, regarding why the uk was following the example of other countries in implementing a lockdown given evidence that such a response seemed to work. a summary of the interview is available on the sky website here: https ://news.sky.com/video /coron aviru s-uk-appro ach-entir elyscien ce-led-grant -shapp s-11958 199. there is also a video of the interview on the sky facebook channel here: https ://www.faceb ook.com/watch /?v=23018 17381 09777 . compartments as the epidemic progresses. in the seir model which we are using, there are four compartments corresponding to susceptible (i.e., not infected, but vulnerable to the disease), exposed (a latent stage usually lasting a few days, where the victim has been infected but is not yet infectious), infectious (at which point they can pass the disease on to others), and removed (meaning they are no longer infectious and may be either recovered from the disease and immune, or else dead). in more complex models, the population may also be subdivided according to age and other factors, with each subdivision being compartmentalised according to disease status as previously described. this would allow for a more detailed representation of the structure of society and the progress of the epidemic as it spreads through the population, but such detail would greatly increase computational demands (especially for large ensembles of simulations as we are using here) and is not necessary for this work. for a full description of the model we are using, see annan and hargreaves (2020) and also house (2020) where the underlying model equations were originally presented. the flow of the population between the compartments depends on parameters which we estimate by fitting the model to observational data for each country. this model fitting process follows the standard bayesian paradigm of defining prior distributions for uncertain parameters, running the model numerous times with parameters sampled from these priors, and calculating the likelihood on the basis of how well the model outputs match the specific observational data that we are using. this process (using a markov chain monte carlo approach) is described in detail in annan and hargreaves (2020) . this approach requires around 15,000 model simulations for each experiment (i.e. country) and the results are represented by an ensemble of model simulations that samples our posterior probability distribution. one critical parameter of the model, which has been widely discussed in the literature and media, is the reproductive number or r, which is the number of new cases that each infectious case generates in a fully susceptible population. if r is greater than 1, the epidemic initially exhibits exponential growth until it infects a sufficiently high proportion of the population that the remaining susceptible fraction substantially shrinks. if r is less than 1, the epidemic decays, again exponentially. in our estimation procedure, we assume that all uncertain model parameters are fixed in time apart from r, which is treated as piecewise constant. we consider three discrete periods within which r is constant. first, there is an initial period prior to "lockdown" controls being imposed by governments. a new, lower value for r is then assumed to apply during the period of strict controls, with a third value applying after the controls are significantly relaxed. country specific lockdown dates that we use are detailed in online appendix 4. in reality, r and other model parameters are likely to vary somewhat during these periods but this piecewise constant approach has been widely used and captures the dominant features of the system (e.g. flaxman et al. 2020) . 13 due to serious limitations in the testing and reporting of case numbers, we rely exclusively on daily reported death numbers for the calibration of our model. again, this is a common approach which is justified on the basis that the reporting of deaths is usually far more consistent and reliable than case numbers which depend strongly on testing capacity and policy. an alternative approach would be to use the number of excess death. while this may better reflect the number of deaths caused by covid than reported death statistics, daily excess death data are not available. moreover, the key results in the model are driven by changes in the rate of infection, hence even if death numbers in a particular country are underestimated due to systematic biases, this will not usually bias the estimates of model parameters. therefore to calibrate the models we use daily reported deaths from our world in data up to 9th june (beltekian et al. 2020) , and later suggest how accounting for excess mortality would alter our estimates. the prior estimate for r after the release of lockdown is taken to be n(1,0.2 2 ) which represents our assumption that the policies are intended to be as open as possible while keeping the epidemic controlled. in many cases, there are insufficient data to constrain this prior estimate strongly, and therefore it plays a greater role in our results than the priors used in earlier phases of the epidemic. estimates of all the r values, as well as our priors, are detailed in online appendix 5. lockdown clearly reduces the infection rate across the board. easing lockdown allows the infection rates to increase again. figure 2 compares observed and modelled deaths in the uk, showing deaths on the (exponential) vertical axis over time. modelled mortality (the solid line) closely matches the actually observed deaths (circles), illustrating that the modelling framework is flexible enough and the methodology sufficiently rigorous that the epidemiological model well replicates the observed patterns in the uk. indeed, only on 3 days do observed deaths fall outside the 95% confidence interval (shaded area), and all such occurrences are in the postlockdown period when the number of daily deaths is comparatively low. similarly, close relationships are displayed for the other countries in the equivalent plots (online appendix 6), highlighting that the model well captures the country specific pandemic pathways. in order to calculate the effects of changing the dates of lockdown, we use the fitted parameter values, and perform simulations in which the date of imposing lockdown is changed-either delayed or advanced by 3 days. we also explore advancing or delaying lockdown by 7 or 12 days, results of which are presented in online appendix 7. this approach is similar to that of others (e.g. flaxman et al. 2020) in which the effects of policies have been analysed. since we are using a single date to represent the net effect of multiple policies which were introduced across a period of several days, it would be more precise to interpret these scenarios as representing a change in the timing of all such policies by the given number of days. likewise, we identify the impact of lockdown using within-country variation in the rate of infection. therefore, to the extent that the stringency of policy interventions vary between countries, our simulations reflect the same countryspecific set of policy interventions of the same stringency being implemented either earlier or later. that said, the lockdown is widely believed to be the most important of these measures (flaxman et al. 2020 ) and so we consider our interpretation to be a reasonable approximation of the impacts of lockdown and variation therein. differences in total mortality for each country dependent on date of lockdown are calculated to 24th june 2020. we also calculate the number of deaths that likely would have occurred were no lockdown implemented, again to the 24th june 2020. for illustrative purposes, the graph of predicted daily deaths for the uk under such a scenario is in online appendix 8. 14 in all cases, no correction is made for the possibility that hospitals got overwhelmed, causing an increase in infection-fatality ratios. to the extent that such an outcome would have occurred, yet more lives would have been lost under the delayed-and no-lockdown scenarios. 14 the graphs are similar for all other countries, and hence not displayed here. table 3 highlights the likely impacts of lockdown policy. it is clear that the imposition of lockdown likely saved in excess of 14 million lives across the countries we examine. this overall analysis of lockdown is similar to that of flaxman et al. (2020) and comparison of overlapping results shows that they are in most cases strikingly similar. 15 however, we caution against over-interpreting the result: it is likely that even without a formal lockdown, people would have socially distanced and engaged in other behaviours to limit covid-19 deaths. nevertheless, earlier governmental action would have saved a large numbers of lives, particularly in countries such as the uk and us who acted relatively late. prelockdown reproduction rates are substantially greater than one, hence across all countries, longer delays result in exponentially greater losses of life. the previous sub-section presented clear evidence that the choice of when to impose lockdown drastically affects the likely number of deaths. moreover, there is significant heterogeneity across countries in the number of lives that would have been saved had lockdown been implemented just 3 days earlier or later. how does this heterogeneity translate into the implied price of life across countries? to assess the price of life we require estimates of the financial cost of lockdown on gdp. we first assume that the full cost of any extension to the length of lockdown is felt in the year 2020. therefore, we estimates the cost to gdp by comparing the last imf forecasts of national gdp in 2020 prior to the pandemic (from october 2019; imf 2019) with their most recent forecast for 2020 (april 2020, imf 2020b). 16 further assumptions are needed to understand the cost of a marginal extension to lockdown. the first is the relationship between lockdown length and cost to gdp. in line with the best available evidence, from studies in the us (walmsley et al. 2020) and thirty panglobal countries (with a focus on european nations, fernandes 2020), length of lockdown appears to be directly proportional to the percentage gdp loss. of course, not all of the gdp loss associated with an extended lockdown is the result of the policy decision alone: progression of the pandemic sufficient to warrant a lockdown (extension) would reduce gdp outlook anyway and there is good evidence that people were changing their behaviours to enact social distancing in advance of direct regulations (gupta et al. 2020) . moreover, it is not just the domestic pandemic which causes gdp losses-some is also driven by the state of the virus in other nations owing to trade (mandel and veetil 2020) . hence we must also make an assumption about how much of the loss in gdp in any given country is the result of the lockdown policy, rather than other factors associated with the ongoing pandemic. andersen et al. (2020) , chronopoulos et al. (2020) and goldsztejn et al. (2020) have all teased apart the effects of lockdown policy from the wider pandemic. all three suggest that the gdp loss caused by lockdown policy is approximately 15% of the total gdp loss experienced by each country. 17 we note of course that there are reasons to believe this figure could be an over-or under-estimate of the proportion of cost attributable to the lockdown policy, and that this could also vary somewhat by country given that lockdown policy may have different impacts on different industries. 18 nonetheless, we see the 0.15 estimate as offering a reasonable ball-park figure, and so adjust predicted gdp losses as per eq. 2: equation 2 states that the gdp loss caused by changing the length of lockdown by some amount (either 3, 7 or 12 days; denoted i ), in country j , is calculated as the relative change in lockdown length, multiplied by the predicted change in gdp as forecast by the imf, and (2) δgdp ij = δlockdown length i actual lockdown length j × imf forecast gdp loss j × 0.15 17 andersen et al use data from individual-level transaction data either side of the border between denmark, which imposed a lockdown, and sweden, which did not. denmark saw transactions reduce 29% in the immediate aftermath of lockdown imposition compared to sweden's 25% reduction. this suggests that 13.8% of the gdp loss denmark experienced is caused by the lockdown rather than mere pandemic progression. chronopolous et al present evidence from either side of the uk lockdown, again using individualconsumer-level transaction data. this suggests a similar proportion of the overall cost is attributable to the lockdown policy: spending drops by 15.2% in the week following lockdown (week beginning 23rd march 2020) relative to the previous period. goldsztejn et al conduct a modelling exercise linking economic data to an seir model for the uk again. this suggests that lockdown accounts for 17% of the overall economic downturn. 18 as more accurate estimates of this key parameter become available, we would encourage the interested reader to replicate our calculations but with an updated estimate of the proportion of gdp loss attributable to lockdown policy to provide more accurate estimates of the price of life. the proportion of the loss attributable to the policy decision ( 0.15 ). we adopt the imf metric for measuring gdp in terms of purchasing power parity international dollars (ppp$) which is held constant such that it is equal to the us dollar. for hubei, we use the same formula as above, however the imf only publishes estimates gdp forecasts at the national level. therefore we partition the effect for hubei alone by multiplying by the proportion of china's gdp which hubei makes up (0.04, 651) . 19 the necessary data, and calculated gdp outcomes, are presented in online appendix 8. it is worth highlighting two further implicit assumptions. first, we assume all of the gdp loss a country experiences occurs during the lockdown period. clearly, countries' economies were already contracting pre-lockdown, and likely will take a long time to return to normal functioning post-easement. however, our assumption ensures that the implied price of life we calculate is an upper bound. second, we assume that the date on which lockdown is eased is independent of the date on which lockdown was imposed. this is an open empirical question as it may be that earlier lockdowns halt the spread of the virus quicker, allowing an earlier end to lockdown. if earlier lockdowns result in earlier release this would lower the overall financial burden of lockdown. hence, again our assumption tends towards an upper bound estimate on the price of life. the additional assumption made for hubei may underestimate the price of life there: the contraction in china's gdp is likely most keenly felt in hubei, the worst hit province. our estimates of price of life would increase if we adjusted for this. aside from the caveat with respect to china, while our assumptions influence absolute estimates of the price of life, the only variables affecting the relative prices across countries are: (1) the number of lives a change in the length of lockdown would save; (2) the original length of lockdown in a country; and (3) a country's gdp. these key variables are not assumed. to underscore the point, our assumptions cannot substantially influence the implied relative price of life across countries. to calculate the implied price of life from a change in the length of lockdown of a set number of days, i , for country, j , we link the predicted change in gdp to the change in number of lives lost as in eq. 3: our primary focus is for the most marginal change in length of lockdown we calculate: imposing lockdown either 3 days earlier or later than its actual date. results for different changes in lockdown date, of 7 and 12 days, are presented in appendices 9 and 10. these show that relative patterns remain unchanged. table 3 showed that the exponential growth in infections means more lives are lost from a delay, than would be saved by shifting lockdown earlier by the same number of days. in contrast the modelled impact on gdp from moving the lockdown date by a fixed number of days is exactly the same; the only difference is in the sign (earlier lockdowns are a cost to gdp, later lockdowns a benefit). hence, the implied price of life is higher for moving lockdown earlier as opposed to later. (3) implied price of life ij = δgdp ij ∕δlives lost ij moreover, as explained previously, by choosing not to impose lockdowns 3 days earlier governments rejected saving more lives when the price was relatively high. similar logic reveals them to have accepted the implied price of life from a delay; they would rather bear the cost in terms of gdp than as further human lives lost. results from these analysis are presented in table 4 . obviously, estimates for prices countries were willing to pay (accepted) are lower than estimates for the prices countries rejected. in almost all cases the estimates of the price of life are below thresholds typically used to estimate the vsl in cost-benefit analyses. hence, ex-post, it is highly likely lockdown enhance social welfare. 20 as with progression of the pandemic, there is huge heterogeneity in the price of life across countries. comparing across countries those who pursued an early lockdown strategy reveal they are willing to pay a high price to save their citizen's lives, only rejecting prices above $1,000,000. the highest implied prices are in korea (> $11,000,000) and new zealand (> $6,000,000), both countries who acted swiftly to suppress the pandemic. 21 however, those countries which imposed lockdown relatively late-on in their respective pandemics were clearly only willing to pay far less to protect lives. belgium, italy and the uk reject prices of life around $100,000. clearly, delayed action in the face of exponential growth cost lives, and implied low price of life in those countries imposed lockdowns relatively late in the pandemic. two comparisons make this cross-country variation in the implied price of life particularly clear. first, the accepted price of life in china ($108,000) is about 25% higher than that for an american ($87,000). this is despite our methods meaning the calculated price of life for china is likely an underestimate. 22 second, compare the acceptable price of life in germany ($525,000) with that in the uk ($67,000). the price of life for a german is nearly an order of magnitude greater than that for a british citizen. that vast difference is despite the two countries being very similar in terms of gdp per capita. these relative implied price of life comparisons are particularly pertinent. our methodology uses ex-post estimates of the number of lives saved to infer what government policy implies for the price of life. yet, these governments were clearly making the decisions ex-ante. nonetheless, these governments were making lockdown decisions at around the same time (except hubei which was far earlier), with nearly identical information sets. thus any differences in relative estimates would hold true even if the pandemic had proved to be far less deadly than it actually is. moreover, this heterogeneity in the price of life is not explained by different values for life. indeed, the implied prices are often far lower than official vsl estimates-seemingly, cash flowing through the market is worth much more than value passing through wellbeing, at least to some countries. the low rejected prices also imply that very few quality 20 ideally we would assess all of the consequences of interventions (e.g. the mental health costs of lockdown) before making such an assertion. however, the difference between vsl values and our price of life estimates suggest that our statement is defensible (certainly for those countries where the latter measures are particularly low). 21 moreover, vietnam would have been included in the modelling exercise, but we were unable to robustly parameterise our epidemiological models as so few cases (let alone deaths, of which there have been none) have occurred. 22 this is true to the extent to which officially reported covid-19 deaths in china are accurate. if officially reported deaths are far lower than the number of deaths which have actually occurred, this figure may well be an overestimate of the price of life in china. we have not found data from china on excess mortality and so cannot speculate on the degree to which mortality data are accurate. adjusted life years (qalys) are assumed to be saved by governments in reducing covid-19-related mortality; otherwise delays to lockdown seem nonsensical. for reference, in the uk the national institute for health and clinical excellence views a qaly costing between £20,000 and £30,000 as good value (nice 2012). as we mentioned when discussing table 2 , those countries with high reported covid deaths, tend to be countries with high ratios of excess mortality to reported death, i.e. there is substantial under-reporting. to examine the extent to which our estimates change when we account for this under-reporting, we focus on the set of countries for which we have reliable estimates of that ratio, and where under-reporting appears prevalent. these countries are: italy, the uk and the usa. the estimates reported in table 5 are calculated by dividing the estimates of the price of life by the ratio of excess mortality to reported deaths (from table 2 ). the intuition behind this is that our estimates of lives saved by lockdowns (used in table 4 ) are based upon reported death data, and hence should be scaled upwards by the degree of under-reporting of deaths. implicit in this correction is the assumption that the ratio of excess death to reported death is constant within a country throughout the pandemic. it is possible that the ratio declines during the tail of the pandemic when covid cases and deaths are less common, and tests more available. nonetheless, our correction offers what is currently the most comparable cross-country figure. table 5 shows that for those countries which under-report covid-19 deaths, implied price of life is substantially reduced, highlighting once again that earlier lockdowns would have increased social welfare tremendously. for example, in the uk, the country for which we estimate a relatively high rate of under-reporting of covid-19 deaths, the adjusted rejected price of life is just $65,000 (equivalent to just over £50,000). the accepted price of life is lower still, at $40,000 (£32,000). this study has begun to disentangle the extent to which cross-country comparisons of responses to covid-19 are valid despite difficulties caused by both exogenous factors and differences in testing rates and the recording of cases and deaths. the results presented in this paper suggest that policy interventions may well explain the majority of cross-country variation in officially reported covid-19 deaths. for some countries, deficiencies in official approaches to the recording of covid-19 mortality mean that estimates based upon deviation of overall deaths away from the seasonally expected norm may provide a more accurate depiction of fatalities caused by the pandemic. such 'excess death' estimates suggest that in some, highly impacted, countries the actual number of covid-19 deaths may considerably higher than indicated in official statistics. for example, within the uk it seems that more than a third of covid-19 deaths may have gone unrecorded. where under-recording is prevalent, then the number of lives lost by delayed intervention (as well as those saved relative to even further delay) is likely to be substantially higher than estimated in this paper. any such under (over) estimation of true deaths would result in an over (under) estimation of the price of life implicit in lockdown decisions. careful consideration of cross-country differences is required if we are to glean the important natural experiment evidence afforded by countries implementing different policy approaches to the pandemic. the results presented in this paper highlight that welldesigned policy can save life. while the economic burden of lockdown is large, comparison with prior decision criteria suggest that such policies generate net benefits for society. pricing the priceless: cost-benefit analysis of environmental protection what is a life worth? robustness of vsl values from contingent valuation surveys activity and the incidence of emergencies: evidence from daily data at the onset of a pandemic pandemic, shutdown and consumer spending: lessons from scandinavian policy responses to covid-19 tweet posted 11th model calibration, nowcasting, and operational prediction of the covid-19 pandemic antconc (version 3.5.8) [computer software covid-19: analogues and lessons for tackling the extinction and climate crises tweet posted 10th coronavirus pandemic (covid-19) (online) sharpening the global focus on ethnicity and race in the time of covid-19. the lancet bolker b (1993) chaos and complexity in measles models: a comparative numerical study exploring the impact of the covid-19 pandemic and uk lockdown on individuals with experience of eating disorders burki tk (2020) cancer care in the time of covid-19 uk suffers second-highest death rate from coronavirus (online) the contingent valuation of safety and the safety of contingent valuation, part 2: the cv/sg 'chained' approach has covid-19 subverted global health? psychological distress during the covid-19 pandemic in france: a national assessment of at-risk populations why does belgium have so many coronavirus deaths? (online) consumer spending responses to the covid-19 pandemic: an assessment of great britain what we know about the long-term effects of covid-19 (online) transport analysis guidance data book, department for transport (online) government has misled public over uk deaths being lower than france (online) demographic science aids in understanding the spread and fatality rates of covid-19 do pollinators contribute to nutritional health? impact of non-pharmaceutical interventions (npis) to reduce covid19 mortality and healthcare demand fernandes n (2020) economic effects of coronavirus outbreak (covid-19) on the world economy estimating the effects of non-pharmaceutical interventions on covid-19 in europe factors that make an infectious disease outbreak controllable the clinical and economic value of a successful shutdown during the sars-cov-2 pandemic in germany hospitalization rates and characteristics of patients hospitalized with laboratory-confirmed coronavirus disease 2019-covid-net, 14 states public policy and economic dynamics of covid-19 spread: a mathematical modeling study does social distancing matter? britain left many of the weakest exposed (online) tracking public and private response to the covid-19 epidemic: evidence from state and local government actions (no. w27027) oxford covid-19 government response tracker, blavatnik school of government. data use policy: creative commons attribution cc by standard exposure to the covid-19 stock market crash and its effect on household expectations which jobs are done from home? evidence from the american time use survey a perfect storm: covid-19 in emerging economies (online) the costs of the shutdown are overestimated-they're outweighed by its $1 trillion benefit (online) predictive power of air travel and socio-economic data for early pandemic spread modelling herd immunity (online) who suffers during recessions? appraisal values or 'unit costs', health and safety executive (online) clinical features of patients infected with 2019 novel coronavirus in wuhan global manufacturing downturn, rising trade barriers (online) imf (2020a) world economic outlook reports (online) the great lockdown (online) the value of changes in the probability of death or injury is ethnicity linked to incidence or outcomes of covid-19? bmj kurmann a, lalé e, ta l (2020) the impact of covid-19 on small business employment and hours: real-time estimates with homebase data statistical inference in a stochastic epidemic seir model with control intervention: ebola as a case study global stability for the seir model in epidemiology the economic cost of covid lockdowns: an out-of-equilibrium analysis social determinants of health inequalities shelter-in-place" isn't shelter that's safe: a rapid analysis of domestic violence case differences during the covid-19 pandemic and stay-at-home orders coronavirus: which regions have been worst hit? (online) assessing cost effectiveness, national institute for health and care excellence (online) an ecological study of the determinants of differences in 2009 pandemic influenza mortality rates between countries in europe sir david norgrove response to matt hancock regarding the government's covid-19 testing data (online) covid-19-collateral damage in scotland (online) ons (2020a) gdp monthly estimate office for national statistics (online) forced social isolation and mental health: a study on 1006 italians under covid-19 quarantine differential effects of intervention timing on covid-19 spread in the united states the long-term effects of educational disruptions (online) covid-19: review of disparities in risks and outcomes (online) china may be beating the coronavirus the economic consequences of covid-19 lock-down in the uk. an input-output analysis using consensus scenarios high population densities catalyse the spread of covid-19 the uk's public health response to covid-19 stability of a stochastic model for hiv-1 dynamics within a host sovereign debt 2020: global borrowing to increase to $8.1 trillion amid favorable financing conditions (online) number of killed soldiers in u.s. wars since world war i as of covid-19 attack rate increases with city size. mansueto institute for urban innovation research paper forthcoming the socio-economic determinants of the coronavirus disease (covid-19) pandemic the economist's tracker for covid-19 excess deaths (online) covid-19: a&e visits in england fall by 25% in week after lockdown the benefits and costs of using social distancing to flatten the curve for covid-19 the green book: central government guidance on appraisal and evaluation covid-19, school closures, and child poverty: a social crisis in the making estimates of the severity of coronavirus disease 2019: a model-based analysis. the lancet infectious diseases school closure and management practices during coronavirus outbreaks including covid-19: a rapid systematic review. the lancet child & adolescent health walmsley t, rose a, wei d (2020) impacts on the us macroeconomy of mandatory business closures in response to the covid-19 who (2020a) coronavirus: symptoms (online who (2020b) coronavirus: overview (online who (2020c) coronavirus: overview (online) covid-19 and african americans (online) the authors are responsible for all views expressed in this paper, and any errors which may remain. that said, we are extremely grateful to a number of people and institutions for making this research possible. joseph billingsley contributed enormously to us being able to successfully gather excess death data for a number of countries. discussions with andrew balmford significantly improved the analyses we conducted. as sources of crucial data, and for their commitment to facilitating open science, we are extremely grateful to: our world in data, the oxford government response tracker, johns hopkins university's coronavirus resource center, the economist (and particularly james tozer and martín gonzález), and the financial times.