key: cord-298433-tev33cjh
authors: Mardimae, Alexandra; Slessarev, Marat; Han, Jay; Sasano, Hiroshi; Sasano, Nobuko; Azami, Takafumi; Fedorko, Ludwik; Savage, Tim; Fowler, Rob; Fisher, Joseph A.
title: Modified N95 Mask Delivers High Inspired Oxygen Concentrations While Effectively Filtering Aerosolized Microparticles
date: 2006-09-22
journal: Ann Emerg Med
DOI: 10.1016/j.annemergmed.2006.06.039
sha: 
doc_id: 298433
cord_uid: tev33cjh

STUDY OBJECTIVE: In a pandemic, hypoxic patients will require an effective oxygen (O(2)) delivery mask that protects them from inhaling aerosolized particles produced by others, as well as protecting the health care provider from exposure from the patient. We modified an existing N95 mask to optimize O(2) supplementation while maintaining respiratory isolation. METHODS: An N95 mask was modified to deliver O(2) by inserting a plastic manifold consisting of a 1-way inspiratory valve, an O(2) inlet and a gas reservoir. In a prospective repeated-measures study, we studied 10 healthy volunteers in each of 3 phases, investigating (1) the fractional inspiratory concentrations of O(2) (F(I)O(2)) delivered by the N95 O(2) mask, the Hi-Ox(80) O(2) mask, and the nonrebreathing mask during resting ventilation and hyperventilation, each at 3 O(2) flow rates; (2) the ability of the N95 mask, the N95 O(2) mask, and the nonrebreathing mask to filter microparticles from ambient air; and (3) to contain microparticles generated inside the mask. RESULTS: The F(I)O(2)s (median [range]) delivered by the Hi-Ox(80) O(2) mask, the N95 O(2) mask, and the nonrebreathing mask during resting ventilation, at 8 L/minute O(2) flow, were 0.90 (0.79 to 0.96), 0.68 (0.60 to 0.85), and 0.59 (0.52 to 0.68), respectively. During hyperventilation, the FiO(2)s of all 3 masks were clinically equivalent. The N95 O(2) mask, but not the nonrebreathing mask, provided the same efficiency of filtration of internal and external particles as the original N95, regardless of O(2) flow into the mask. CONCLUSION: An N95 mask can be modified to administer a clinically equivalent FiO(2) to a nonrebreathing mask while maintaining its filtration and isolation capabilities.

There is worldwide concern about an imminent influenza pandemic. In this event, affected patients will back up in emergency departments (EDs) 1 and ambulances, 2 placing their staff at increased risk of infection. These considerations highlight the importance of transmission prevention strategies in the protection of ambulance and ED personnel. Currently, their personal protective devices include gloves, gowns, shoe covers, appropriate filter masks, and positive-pressure isolation hoods. 3 These barriers are sometimes ineffective because they are overwhelmed by prolonged exposure and large viral loads 4 or because they are not used properly, in time, or at all. [5] [6] [7] Reducing the shedding of infectious particles from contagious patients may provide additional protection for ambulance or ED personnel. Although it is commonly recommended that surgical masks be used on potentially infectious patients during periods of patient triage or transportation, it is readily acknowledged that such masks do not provide full protection from infection transmission. 3 Standard N95 masks may be more effective in containing potentially infectious particles but will not be suitable for most influenza patients ill enough to seek medical care, because these patients will require oxygen (O 2 ) therapy. O 2 therapy is commonly administered with the nonrebreathing O 2 mask, which is not designed to contain any respiratory droplets. In fact, these masks may actually increase the dispersal of respiratory droplets by jetting them through the mask's open side vents. 8, 9 Somogyi et al 8 reported that the use of the Hi-Ox 80 mask, which is designed specifically to deliver high fractional inspired O 2 concentrations (FIO 2 ), can provide respiratory isolation if a bacterial-viral filter is placed on its outlet port. However, the mask is expensive, unfamiliar to many, and not likely to be widely available in numbers sufficient for use as a first-line O 2 mask in the case of a pandemic.

The implementation and routine use of an entry-level isolation O 2 mask during initial patient contact could provide an extra measure of protection to front-line health care workers from potentially contagious patients. Our approach in designing such a mask was to add a nonrebreathing O 2 manifold to the familiar N95 mask, which provides excellent viral and bacterial filtration and is already widely distributed. We tested the effectiveness of this "N95 O 2 mask" in providing supplemental O 2 and compared it to the nonrebreathing mask and to the Hi-Ox 80 mask. We then tested the effectiveness of the N95 O 2 mask in protecting the wearer from microparticles in the environment and compared it to the criterion standard in this regard, the N95 mask. We also tested how well the N95 and the N95 O 2 masks retained microparticles originating from inside the mask. Finally, although the nonrebreathing mask is not designed to provide respiratory protection or isolation, we included it in particle testing because it is common, at least in our institutions, to find health care workers who believe it confers some level of protection to the wearer or the health care provider.

This study is a prospective repeated-measures design. We divided the study into 3 phases. In phase 1, we tested the O 2 delivery characteristics of the N95 O 2 mask, the Hi-Ox 80 mask, and the nonrebreathing mask at resting ventilation and hyperventilation, each at 3 O 2 flow rates (to view photographs of the masks, see Figure E1 , available online at http://www.annemergmed.com). In phase 2, we tested the protection function (ie, the ability of the mask to filter microparticles from ambient air) of the N95 mask, the N95 O 2 mask, and the nonrebreathing mask. In phase 3, we tested the masks' isolation function (ie, the effectiveness of each mask in containing microparticles generated inside the mask). We did not include the Hi-Ox 80 in particle testing, because its protective and isolation function would be that of the particular filter placed on its expiratory port, and this information is already well documented (eg, see Dellamonica et al 10 ) . Neither subjects nor investigators were blinded to the type of mask being used for any of the test phases.

All tests were conducted in a well-ventilated patient room at a university teaching hospital, with the door closed.

After receiving institutional ethics research board approval, we obtained signed informed consent from 10 volunteers for each of the 3 phases of the study. Subjects were recruited by posted advertisement. Inclusion criteria were healthy nonsmoking men (without facial hair) or women between the ages of 18 and 60 years, with no active respiratory disease.

We tested the O 2 delivery characteristics of 3 masks as characterized by the FIO 2 that they supplied. The 

We protect health care workers from communicable disease by giving them protective equipment and using devices that decrease emission of infectious agents from the source patient.

This study shows that an N95 mask can be modified to deliver oxygen to a person in a manner similar to a nonrebreathing mask while providing significant emissions reduction.

What this study adds to our knowledge It is possible to practically increase protection for clinicians treating infectious patients with airbornetransmissible diseases, without compromising oxygen therapy for the patient.

If faced with an influenza pandemic, another severe acute respiratory syndrome-like outbreak, or patients with a dangerous communicable illness, clinicians may be afforded another level of protection through use of a similar device without compromising patient care.

HealthCare, Yorba Linda, CA) and the nonrebreathing mask (Airlife Adult Oxygen Mask; Cardinal Health, McGaw Park, IL).

We modified the N95 mask to allow for O 2 administration by adding an O 2 delivery manifold similar to that of a standard nonrebreathing mask. The manifold consisted of a 1-way valve, an O 2 inlet port, and an O 2 reservoir ( Figure 1 ). The Hi-Ox 80 mask was modified by placing a bacterial-viral filter (DAR Sterivent Mini; Mallinckrodt DAR, Mirandola, Italy) on its expiratory port. 8 Subjects were fitted with the Hi-Ox 80 and initially asked to breathe normally (resting condition). An O 2 flow of 2, 4, or 8 L/minute was selected randomly by draw. We chose these O 2 flow rates for 2 reasons. First, the Hi-Ox 80 is rated to provide clinically useful FIO 2 at these flows. 11 Second, using lower flows with O 2 masks may be necessary in case O 2 supplies become limited, as when treating mass casualties (eg, during a pandemic). Subjects were not told of the O 2 flow being administered. Once end-tidal PCO 2 (PETCO 2 ) and end-tidal PO 2 (PETO 2 ) reached a steady state (defined as less than a 2 mm Hg change in PETCO 2 during 2 minutes), minute ventilation, PETCO 2 , and PETO 2 were recorded for 2 minutes. Subjects were then asked to increase their minute ventilation sufficiently to reduce their PETCO 2 by 10 mm Hg below their resting values (hyperventilation) and maintain that level of PETCO 2 . Once steady state was achieved at the target PETCO 2 , minute ventilation, PETCO 2 , and PETO 2 were again recorded for 2 minutes. Subjects then reestablished steady-state resting ventilation, after which the test was repeated for the other 2 O 2 flow rates. After testing the Hi-Ox 80 at all 3 flow rates, the other 2 masks were tested (in random order) using the same protocol. The Hi-Ox 80 mask was tested first because it is the only mask that permits measurement of minute ventilation. We were then able to use the PETCO 2 to match the minute ventilation when testing the other masks.

We tested the protective function of the masks by measuring the extent to which particles originating from ambient air leaked into the mask during normal breathing. We tested the N95 mask (3M model 1870, 3M, for 9 subjects; Aero mask, Aero Co, Southbridge, MA, for 1 subject), the N95 O 2 mask (constructed from the same models of N95 mask stated previously), and the nonrebreathing mask. Six of the subjects were hospital employees who had been previously fit-tested with N95 masks. We therefore used the same model of N95 mask for their tests. The remaining 4 subjects who were not previously fit-tested were arbitrarily supplied with the 3M model 1870. Although there is a possibility this may have introduced variability in our data, the arbitrary use of masks simulates cases in which this mask would be applied to a patient as an infectious barrier. Each mask was prepared in advance by inserting a gas sampling port into the mask material ( Figure 2 ). The particle generator was turned on and infused microparticles freely into the closed room for 20 minutes before subject testing to increase the atmospheric particle concentration in the room to more than 200/cm 3 . The test mask was placed on the subject's face, and subjects were instructed to breathe normally. After at least 2 minutes, we recorded particle concentrations outside the mask and then from inside the mask for 30 seconds each.

We tested the isolation function of the same 3 masks by measuring the extent to which particles originating from inside the mask leaked out into the surrounding air. Masks were prepared in advance by inserting sampling and particle infusion ports into the mask. Subjects were instructed to breathe Figure 2 . Particle testing setup. The N95 oxygen mask contains an internal gas sampling port and a separate port used for microparticle infusion.

normally. Background particle concentrations were measured 6 cm in front of the mask during 30 seconds with the particle generator off. The particle generator was then turned on and the particles directed into the mask through a plastic tube attached to one of the ports. After an equilibration period of at least 60 seconds, we recorded particle concentrations for 30-second periods from inside the mask, 6 cm in front of the mask and 50 cm to the side of the mask. The latter location simulates the location of a health care worker's face while attending a patient. Both protective and isolation tests were performed once for the N95 mask. For the N95 O 2 mask and the nonrebreathing mask, the tests were performed with O 2 flow rates of 2 and 10 L/minute.

O 2 flow to the masks was controlled by a calibrated flow meter (Voltek Enterprises, Toronto, Ontario, Canada). Gas was sampled continuously from the oropharynx and analyzed for CO 2 (IR1507; Servomex Fairfax, CA) and O 2 (UFO130-2; Teledyne-AI, City of Industry, CA) partial pressures (PCO 2 and PO 2 , respectively). Tidal gas was sampled from the oropharynx by a catheter inserted through a sealed port in the mask, which allowed accurate sampling of end-tidal gases without dilution from the high flows of O 2 .

A particle generator, model 8026 (TSI Inc, Shoreview, MN), was used to generate a steady flow of ultrafine (approximately 0.2 m) sodium chloride particles that are easily suspended in air. Particle concentrations were measured with Portacount Plus, model 8020 (version 2003 Rev K; TSI Inc), which is rated for detecting the concentration of particles as small as 0.02 m. 12

PCO 2 and PO 2 signals were digitized and recorded continuously by a data acquisition and analysis program (Labview; National Instruments, Austin, TX). PETCO 2 and PETO 2 were identified using a custom peak detection algorithm and converted to fractional concentrations (FETCO 2 and FETO 2 ). For each O 2 mask, ventilation, and flow combination, average FETCO 2 and minute ventilation (where applicable) were calculated during 2-minute steady states for each subject. FIO 2 was calculated for each breath from FETCO 2 and FETO 2 using the alveolar gas equation 13 (FIO 2 is a flowaveraged value of the O 2 concentration during inspiration and is reflected in the exhaled concentrations of CO 2 and O 2 ). For each O 2 mask, ventilation, and flow combination, average FIO 2 s were calculated during 2-minute steady states for each subject.

During particle testing, on average, 15 to 20 discrete measurements were recorded for each 30-second sampling period and entered manually into a spreadsheet for analysis. For each subject, background concentrations recorded outside the mask were averaged.

We used FIO 2 as the marker of effectiveness of O 2 delivery. We assumed the N95 mask to be the criterion standard for particle filtration; clinical equivalence was assumed for masks whose performances were comparable to that of the N95. The protective function of each mask was quantified by measuring the concentrations of particles inside the mask and expressing each discrete measurement as a percentage of average external concentrations. The isolation function of each mask was quantified by measuring the external particle concentrations while infusing particles into the mask and expressing each discrete measurement as a percentage of change from the average background concentrations measured before the particle generator was started.

Descriptive statistics and graphical methods were used to display our results. We used bar graphs to display average FIO 2 values for all subjects. Box plots were used to display data points obtained during particle testing to illustrate the variability in particle concentrations that we observed over time.

All quantitative results are expressed as median (range). With the Hi-Ox 80 mask, subjects' minute ventilation and PETCO 2 (pooled for all O 2 flow rates) were 5.8 L/minute (2.3 to 9.5 L/minute) and 39.9 mm Hg (34.9 to 44.5 mm Hg), respectively, at rest and 14.9 L/minute (9.0 to 23.1 L/minute) and 29.2 mm Hg (24.7 to 33.9 mm Hg), respectively, during hyperventilation. The PETCO 2 (our marker for ventilation) for the N95 O 2 mask and the nonrebreathing mask (also pooled for all O 2 flow rates) was equivalent to that obtained by testing the Hi-Ox 80 (39. 6 The Hi-Ox 80 consistently delivered a higher FIO 2 than the other masks, with greater margins at higher O 2 flow rates ( Figure 3 ). The FIO 2 obtained from the N95 O 2 mask was clinically equivalent to that from the nonrebreathing mask for both resting ventilation and hyperventilation. During resting ventilation and at 8 L/minute O 2 flow, the FIO 2 delivered by the Hi-Ox 80 was 0.90 (0.79 to 0.96), consistent with that predicted by the package insert. Under these conditions, the FIO 2 with the N95 O 2 mask and nonrebreathing mask was 0.68 (0.60 to 0.85) and 0.59 (0.52 to 0.68), respectively.

For protective function, the N95 O 2 mask provided the same efficiency of filtration of outside particles as the N95 mask at O 2 flow rates of 2 and 10 L/minute (Figure 4 ). Median particle concentrations inside the N95 and N95 O 2 masks at 2 and 10 L/minute were less than 1% of those outside the mask (0.0% to 3.0%). In contrast, high particle concentrations were found inside the nonrebreathing mask, with O 2 flow at 2 L/minute (53% [37% to 87%]) and with O 2 flow at 10 L/minute (17% [3% to 37%]).

For isolation function, the N95 O 2 mask provided the same efficiency of filtration of inside particles as the N95 mask. There was no increase in external particle concentrations above background values measured outside either mask, regardless of distance from the mask or O 2 flow into the N95 O 2 mask ( Figure 5A , B). In contrast, particle concentrations measured outside the nonrebreathing mask were markedly higher than background values at 6 cm and 50 cm from the mask: 800% (62% to 3100%) and 450% (100% to 12,000%), respectively, when O 2 flow rates were 2 L/minute and 330% (15% to 1300%) and 95% (33% to 3000%) when flows were 10 L/minute.

We studied the O 2 delivery characteristics of the 3 masks in healthy volunteers, not patients. The justification for this method is that the marker of effectiveness of O 2 delivery we studied-FIO 2 -depends only on the characteristics of the mask, the O 2 flow, and the minute ventilation of the subject. The latter 2 were controlled in this study. Moreover, arterial partial pressure of O 2 and oxyhemoglobin saturation depend not just on the FIO 2 but also on such individual patient factors as the ventilation-to-perfusion ratio and shunt. Therefore, arterial blood gases and oxyhemoglobin saturation measurements in patients with pulmonary and other pathology are not required for evaluation of mask-specific O 2 administration efficiency.

We tested a group of 10 subjects for each phase of the protocol, but several subjects did not participate in all 3 phases of the study. However, for the phase of the study in which an individual subject did participate, he or she tested all 3 masks. Furthermore, the measured parameters of each phase of the study are totally independent of each other.

We studied the ability of the masks to filter microparticles, and we attempt to make inferences on their performance with other sized particles, such as respiratory droplets. Nevertheless, the 0.2-m microparticles generated by the TSI particle generator are widely used to test for effectiveness of mask fit and filtration and are reasonable surrogates for such infectious particles as viruses (which are 20 to 300 nm), bacteria (tubercle bacillus; 0.2 to 0.6 by 1.0 to 10 m), 14 and aerosolized respiratory droplets (Ͼ10 m). In vitro testing has shown that N95 material prevents the penetration of approximately 95% of MS2 viruses (approximately 27.5 nm in diameter). 15 The isolation function of the Hi-Ox 80 mask with filter was not compared directly with that of the N-95 O 2 mask for 2 reasons. First, its isolation function would reflect the type of filter used, of which a large variety is available. Second, if the FIO 2 of the Hi-Ox 80 is clinically necessary, the N95 O 2 mask is not a suitable alternative. We propose that the N95 O 2 mask provides an FIO 2 similar to, and therefore is a suitable alternative for, that of the nonrebreathing mask. Still, under some circumstances when isolation function is the critical issue, a direct comparison may be necessary.

Because of the nature of our study setup, it was not possible to blind subjects or investigators to the type of mask being used.

There are 2 main findings of our study. The first is that an N95 O 2 mask can deliver O 2 at least as effectively as a nonrebreathing mask. The second is that an N95 O 2 mask retains its filtration function for small particles in both directions. Although it is accepted that a properly fitted N95 mask provides the wearer with a high level of protection from inhaling external particles, our study is the first to demonstrate that the N95 filter material is also effective in preventing particles originating inside the mask from escaping into the surroundings in vivo. This could not be assumed from its proven protective function, because the inside of the mask is exposed to moisture, higher gas flows and pressures, and much higher concentrations of particles than those present on the outside.

An implication of these findings is that the N95 O 2 mask may be effective in respiratory "isolation" by restricting the spread of similarly sized infectious particles from an infected patient. In contrast, whereas people are frequently admonished to cover their mouths with their hands when they cough or sneeze to protect others (a practice popularized during the 1918 influenza pandemic), we found that covering the mouth with the nonrebreathing mask provides no reduction in particle concentrations at 6 or 50 cm from the subject breathing at rest. Furthermore, persistence of high particle concentrations inside the nonrebreathing mask casts doubt on any assumption that O 2 flow into a nonrebreathing mask will in any way protect a patient from inhaling others' respiratory droplets.

Because no controlled studies outlining the full mechanisms of person-to-person transmission of influenza have been published, 16 we must make certain inferences from animal studies and observational and inoculation studies in humans. It appears that influenza is transmitted by droplets (Ͼ10 m), aerosolized droplet nuclei (Ͻ10 m), and by direct and indirect contact (see Bridges et al 16 and Salgado et al 17 for reviews). During the severe acute respiratory syndrome (SARS) outbreaks, it was assumed that reducing the dispersal of respiratory droplets from infected patients would help control the spread of infection. To this end, the World Health Organization guidelines state: "suspected cases should wear surgical masks until SARS is excluded." 3 Health Canada guidelines similarly recommend that a surgical mask be placed on the patient "until SARS is excluded or the patient is admitted into the room." 18 However, it is also recognized that a surgical mask "does not provide adequate respiratory protection to the wearer [ie, health care worker] if the infection is airborne." Recent studies in vitro have demonstrated that surgical masks can allow the penetration of more than 80% of virions 27.5 nm in diameter 15 -besides those that escape from the gaps in the mask seal to the face. The same would obviously apply to the isolation function of a surgical mask when worn by a contagious patient. Although quarantine of the patient may be effective in limiting contacts, it does not reduce the infectious load on those who enter the room and approach the patient or who participate in direct patient care. 19 When O 2 is administered to a patient, respiratory isolation is even more problematic. Our study confirms earlier work demonstrating that standard O 2 masks do not provide effective containment of droplet or nuclei-sized particles and, furthermore, may even contribute to the pattern and extent of spread. 8, 9 These considerations would be of greatest concern to such front-line health care workers as emergency medical services (EMS) and ED personnel who are required to attend symptomatic patients before any definitive diagnosis is made. The intensity of exposure is increased for these personnel for 2 reasons. First, they may have to provide high-risk interventions such as O 2 administration 9 and endotracheal intubation. 6 Second, the duration of exposure will likely be prolonged because filling of inpatient and critical care areas will result in Table E1 (available online at http://www.annemergmed.com).

"boarding" of patients in the ED and ambulances. 20 The risk to EMS workers appears to be particularly great when EMS are placed on hospital bypass because it leaves these workers in close quarters with sick patients for prolonged periods.

A number of limitations to the use of the N95 O 2 mask may arise in actual clinical practice. One would still have to rely exclusively on health care worker-worn barrier devices when patient isolation is impossible (as during endotracheal intubation or bronchoscopy 6 ), as well as for patients for whom a good mask seal is difficult to obtain (for example, edentulous patients or patients with full beards). The N95 O 2 mask will also not be suitable for patients requiring an FIO 2 greater than 0.7, nebulized medication, invasive or noninvasive ventilatory assistance, or airway protection.

One would also expect that the limitations of the original N95 mask (eg, deterioration of filtration function with prolonged use, excessive handling, and accumulation of moisture) may also apply to the N95 O 2 mask. Indeed, increased airflow resistance and the perception of increased work of breathing were commonly experienced by health care workers after prolonged N95 mask use during the SARS outbreak in Toronto. The increased resistance to breathing is likely due to moisture obstructing the pores in the mask material. Increased breathing resistance would reduce the ability of breathless patients to tolerate long-term use of the N95 mask. Although no detailed studies evaluating the duration of effectiveness of the N95 mask have been done, 21 these considerations indicate that the N95 O 2 mask may be best suited for short-term use, such as during ambulance transport, hospital triage, and transport from one hospital ward to another (eg from the patient room to the diagnostic imaging department). On the other hand, the inspiratory flow resistance may actually be less than that of the original N95 because the N95 O 2 mask includes an O 2 reservoir that provides the major part of the inspiratory volume, leaving only small volumes at low flows at end inspiration to be inhaled through the mask filter material.

We tested the masks for only a short period because they were intended to provide a short-term solution to O 2 delivery and respiratory isolation at initial patient contact. Extrapolation of these results to more prolonged use should be made with caution.

In conclusion, our study indicates that the N95 O 2 mask we constructed can deliver an FIO 2 clinically equivalent to that provided by a nonrebreathing mask, although neither is as efficient as the Hi-Ox 80 . Furthermore, the modification of the N95 mask for O 2 delivery retains the same filtration function as the N95 mask. This is the first study to confirm that the N95 mask, as well as our modified version of the mask, is effective in retaining microparticles originating inside the mask in vivo. Assuming that the kinetics of microparticle distribution studied in humans is a suitable model to study the risk of transfer of infection in humans, we conclude that for O 2 -requiring patients, the N95 O 2 mask appears to provide protection and isolation performance equivalent to that of the N95 mask, whereas the nonrebreathing mask would provide none at all. Additional clinical studies are needed to confirm that O 2 administration through this N95 O 2 mask is well tolerated by patients and that its use actually reduces the risk of disease transmission from infectious patients to health care workers. NRM, Nonrebreathing mask. *The protection function of each mask was quantified by measuring the concentrations of particles inside the mask and expressing each discrete measurement as a percentage of average external concentrations. *The isolation function of each mask was quantified by measuring the external particle concentrations while infusing particles into the mask and expressing each discrete measurement as a percentage of change from the average background concentrations. 

Once upon a time in the emergency department: a cautionary tale

The effect of emergency department crowding on paramedic ambulance availability

World Health Organization. World Health Organization hospital infection control guidelines

Critically ill patients with severe acute respiratory syndrome

SARS among critical care nurses

Transmission of severe acute respiratory syndrome during intubation and mechanical ventilation

Effectiveness of precautions against droplets and contact in prevention of nosocomial transmission of severe acute respiratory syndrome (SARS)

Dispersal of respiratory droplets with open vs closed oxygen delivery masks: implications for the transmission of severe acute respiratory syndrome

Evidence of airborne transmission of SARS

Comparison of manufacturers' specifications for 44 types of heat and moisture exchanging filters

Efficiency of oxygen administration: sequential gas delivery versus "flow into a cone" methods

Comparison of N95 disposable filtering facepiece fits using Bitrex qualitative and TSI Portacount quantitative fit testing

Evaluation of five oxygen delivery devices in spontaneously breathing subjects by oxygraphy

Manual of Clinical Microbiology

Do N95 respirators provide 95% protection level against airborne viruses, and how adequate are surgical masks?

Transmission of influenza: implications for control in health care settings

Influenza in the acute hospital setting

Public Health Agency of Canada. Severe acute respiratory syndrome (SARS)

Identification and containment of an outbreak of SARS in a community hospital

Medical care capacity for influenza outbreaks

Infection control guidance for respirators (masks) worn by health care workers: frequently asked questions: severe acute respiratory syndrome (SARS)