Probability of fit failure with reuse of N95 mask respirators.

Editor—The 2019 coronavirus pandemic (COVID-19) has created a worldwide shortage of disposable N95 mask respirators that has led to extended use and reuse.1, 2, 3 Multiple healthcare organisations have implemented reuse of disposable N95 respirators designed for 8 h of use, for up to 20 days. However, the durability and fit of respirators after multiple days of clinical reuse are unknown. A seal to the face is necessary to ensure that small aerosolised droplets are filtered. We conducted a cross-sectional pilot study to determine the effects of reuse and hydrogen peroxide vapour decontamination on the effectiveness of N95 respirators by qualitative fit testing.

This study was a voluntary, non-randomised, and low-risk quality improvement project; Institutional Review Board review and formal written consent were not required per institutional policy. From April to May 2020, a convenience sample of anaesthesiology clinical staff at an academic tertiary care centre who had within the past year passed fit testing of the same model of N95 mask respirator were included. Individuals who had worn their respirator for less than 1 day were excluded. All anaesthesiology clinical staff are trained yearly on performance of a self-performed user seal check and appropriate respirator use.

Before the start of the study, department management instructed clinicians to continuously record the number of days worn, times decontaminated with Food and Drug Administration approved Bioquell Z-2 vaporised hydrogen peroxide (Horsham, USA), and times the N95 was donned. Before testing, participants self-reported the same information and if they believed the respirator provided a good fit. The participants were screened for COVID-19 risk by asking if they had any of the known symptoms and if their unprotected respirator was directly exposed to COVID-19 patients without subsequent decontamination. Exposed respirators that were not decontaminated were not tested. All six testing staff members were trained on appropriate fitting and testing directly by the Environmental Health and Safety Department. Qualitative fit testing was performed with denatonium benzoate (Bitrex®) in accordance with the Occupational Safety and Health Administration standard 1910.134 App A (3M, St. Paul, USA). On a subset of participants (based on respirator availability) with fit failures, testing was repeated with a new N95 respirator of the same model. The data were analysed with logistic regression of binary fit failure using the R package cgam (R Foundation for Statistical Computing, Vienna, Austria) for flexible monotone increasing failure probabilities (Online Appendix).

Of 74 anaesthesia providers who participated in repeat fit testing, 46 were females and 28 were males. The females were more likely to fail fit testing (63% vs 29%; P=0.008). Ten participants wore the 1860 and 64 wore the 1804 VFlex™ (3M, St. Paul, USA). Figure 1 displays the estimated failure probability by number of days worn. The failure rate was 46% after 4 days (95% confidence interval [CI]: 31–63%), 50% after 10 days (95% CI: 36–63%), and 55% after 15 days (95% CI: 37–71%). Of respirators that passed fit testing, the median numbers of days worn were 7 (n=37; inter-quartile range [IQR]: 5–12) and 8 (n=37; IQR: 5–12) in the group that failed fit testing. The number of sterilizations had a modest association with modest precision (odds ratio [OR]: 1.5; 95% CI: 0.9–2.8) relative to fit failure, adjusting for the number of days worn. Each respirator was sterilised a median of one time in both groups (pass IQR: 0–1; fail IQR: 0–2; P=0.12). The number of days worn and number of uses were correlated (r=0.63; P<0.001). The number of times used per day had a negligible association with modest precision (OR: 0.8; 95% CI: 0.6–1.0) with fit failure. Of respirators that passed fit testing, the median numbers of use were 20 (IQR: 15–40) and 18 (IQR: 12–35) in the group that failed fit testing. Only 7% (one/15) of participants who repeated testing with a new N95 failed. Users with failed fit tests were more likely to state their mask fit poorly (OR: 6.5; P=0.02); however, 73% (95% CI: 57–85%) of users with N95 masks that failed testing believed their respirator fit well. Testers believed 89% (95% CI: 75–96%) of N95 with failed fit tests were of good or like new quality.

Fig 1

Risk of N95 mask respirator fit failure relative to number of days worn (n=74). The black line represents a monotone semi-parametric logistic regression with dashed lines representing the 95% confidence interval. Blue bars represent a histogram of respirators tested. Each respirator was tested only once.

Our findings suggest that risk of N95 mask fit failure is high after 4 days of clinical use. We do not have adequate data to make predictions of the failure rate before 4 days of clinical use. Degesys and colleagues reported 23% of respirators failed fit testing after three shifts (days), and this increased to 67% after more than three shifts. A strength of both of our studies relative to older studies is that reuse of the respirators was not simulated; our subjects were clinicians caring for real patients, and thus, the results we obtained are representative of actual use scenarios.

Potential limitations include qualitative testing, respirator donning, and cross-sectional design. Quantitative testing was not possible in this study as it necessitates irreparable damage to the respirator to allow sampling of air within the respirator. Our design of allowing repositioning after an initial failure should alleviate bias attributable to improper donning. As all participants passed fit testing in the past, it is unlikely that our results reflect respirators that never worked or individuals who were unable to don masks correctly. The markedly higher rate of passing with a new respirator also suggests that improper donning or lack of test reproducibility is not driving these results.

Longitudinal data collection with repeated measures would address some of the potential sources of bias in this study. Based on these and other results, we recommend that organisations follow the Centers for Disease Control and Prevention recommendations and limit reuse to five individual uses as supply allows. If the local supply of disposable N95 mask respirators is insufficient to allow replacement more frequently, then alternative more durable elastomeric solutions or measures to restrict need should be considered. Despite training on user seal checks, the participants were unable to reliably identify if their respirator fit poorly before testing.

Declarations of interest

The authors declare that they have no conflicts of interest.