key: cord-314963-sk8pqjrh
authors: O’Hearn, Katie; Gertsman, Shira; Webster, Richard; Tsampalieros, Anne; Ng, Rhiannon; Gibson, Jess; Sampson, Margaret; Sikora, Lindsey; McNally, James Dayre
title: Efficacy and Safety of Disinfectants for Decontamination of N95 and SN95 Filtering Facepiece Respirators: A Systematic Review
date: 2020-08-13
journal: J Hosp Infect
DOI: 10.1016/j.jhin.2020.08.005
sha: 
doc_id: 314963
cord_uid: sk8pqjrh

BACKGROUND: Decontaminating and re-using filtering facepiece respirators (FFRs) for healthcare workers is a potential solution to address inadequate FFR supply during a global pandemic. AIM: The objective of this review was to synthesize existing data on the effectiveness and safety of using chemical disinfectants to decontaminate N95 FFRs. METHODS: We conducted a systematic review on disinfectants to decontaminate N95 FFRs using Embase, Medline, Global Health, Google Scholar, WHO feed, and MedRxiv. Two reviewers independently determined study eligibility and extracted predefined data fields. Original research reporting on N95 FFR function, decontamination, safety, or FFR fit following decontamination with a disinfectant was included. FINDINGS AND CONCLUSIONS: A single cycle of vaporized hydrogen peroxide (H(2)O(2)) successfully removes viral pathogens without affecting airflow resistance or fit, and maintains an initial filter penetration of <5%, with little change in FFR appearance. Residual hydrogen peroxide levels following decontamination were within safe limits. More than one decontamination cycle of vaporized H(2)O(2) may be possible but further information is required on how multiple cycles would affect FFR fit in a real world setting before the upper limit can be established. Although immersion in liquid H(2)O(2) does not appear to adversely affect FFR function, there is no available data on its ability to remove infectious pathogens from FFRs or its impact on FFR fit. Sodium hypochlorite, ethanol, isopropyl alcohol and ethylene oxide are not recommended due to safety concerns or negative effects on FFR function.

Shortages of personal protective equipment (PPE), including N95 filtering facepiece respirators (FFRs), are common during pandemics. The Centres for Disease Control and Prevention (CDC) recommends N95 FFRs, which filter 95% of airborne particles [1] , as the preferred PPE when entering the room of a patient with suspected or confirmed COVID-19, and that an N95 FFR should be worn during all aerosol generating procedures [2] . Unfortunately, during the ongoing COVID-19 pandemic, some hospitals and healthcare workers are faced with an inadequate supply of N95 FFRs [3, 4] while also dealing with an increase in SARS-CoV-2 positive patients. Project N95, a COVID-19 initiative that transfers PPE from manufacturers and other disciplines to healthcare institutions in need, received requests for over 253 million units of equipment from 6,962 health centres globally since March 20 th 2020 [5] . Consequently, addressing the N95 FFR shortage has become a matter of increasing urgency as cases of COVID-19 continue to rise.

A potential solution to FFR shortages would be to decontaminate and re-use FFRs.

However, prior to utilizing this strategy, it is essential to demonstrate that decontamination does not compromise structural integrity, fit, filter efficiency (aerosol penetration), and airflow resistance of the FFR [6] . Several decontamination methods have been previously investigated, including energetic (e.g. microwave irradiation, ultraviolet germicidal irradiation (UVGI)) [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] , gaseous (e.g. ethylene oxide, vaporized hydrogen peroxide) [7, 9, 11, 19, 20] , and liquid protocols (e.g. hydrogen peroxide, sodium hypochlorite) [7, 9, [11] [12] [13] 18, 19, 21, 22] . The CDC recently released crisis standards of care decontamination recommendations, with a brief summary of evidence for several of these approaches [23] . However, detailed information on the safety and efficacy of a variety of decontamination methods is essential to allow hospital decision makers to J o u r n a l P r e -p r o o f evaluate the evidence and determine the feasibility of rapidly implementing different protocols at their institutions.

To help inform FFR-reuse policies and procedures, our team has conducted three systematic reviews to synthesize existing published data regarding the effectiveness of UVGI [24] , heat and microwave irradiation [25] , and chemical disinfectants for decontamination of National Institute for Occupational Safety and Health (NIOSH)-approved N95 FFRs. This review will focus on chemical disinfectants, with the following objectives: (1) to assess the impact of the each disinfectant method on FFR performance, with a specific focus on aerosol penetration and airflow resistance; (2) to determine the effectiveness of each disinfectant method at reducing viral or bacterial load; (3) to describe observations related to changes in physical traits following decontamination with a disinfectant; (4) to determine the impact of each disinfectant on FFR fit; and (5) to describe findings or observations related to potential health risks or irritation from residual disinfectant remaining on FFRs following decontamination.

The study protocol and objectives were established a priori and registered on PROSPERO on April 5 th , 2020 (CRD42020178440), and reported here according to the PRISMA guidelines for systematic reviews (Appendix A) [26] . The protocol was also uploaded as a pre-print to OSF on April 5 th , 2020 (https://osf.io/8usx6/).

Studies were eligible for inclusion in this systematic review if they satisfied all of the following criteria: (1) Original publication or systematic review; (2) Study reported on decontamination procedures for NIOSH-approved N95 (including SN95) FFRs or their components; (3) At least one of the decontamination procedures evaluated used one of the J o u r n a l P r e -p r o o f following chemical disinfectants: sodium hypochlorite; liquid hydrogen peroxide, vaporized hydrogen peroxide, hydrogen peroxide gas plasma, or ionized hydrogen peroxide; ethanol or isopropyl alcohol; (4) The study reported on at least one of the following outcomes of interest: (i) impact of the disinfectant on FFR performance, with a specific focus on aerosol penetration and airflow resistance (pressure drop); (ii) effectiveness of the disinfectant at removing viral or bacterial load; (iii) observations related to changes in physical traits following decontamination with a disinfectant; (iv) impact of each disinfectant on FFR fit; or (v) findings or observations related to user safety or irritation. Only studies published in English or French were included.

Studies published prior to 1972, the year that the N95 FFR was invented [27] , were excluded. We included peer-reviewed literature and pre-prints. Editorials, narrative reviews, book chapters, patents, and non-peer-reviewed commissioned reports were excluded.

The following databases were searched by two health sciences librarians ( Health, Medrxiv and OSF Preprints were searched March 31, 2020 for the term "N95" and records pertaining to decontamination were selected and downloaded.

Citations were uploaded to InsightScope (www.insightscope.ca) for title and abstract screening and full text review. At both title/abstract and full text screening levels, citations were assessed in duplicate and independently. Before citation screening was initiated, each reviewer was asked to read the published protocol for this systematic review to familiarize themselves with the review objectives and citation screening process. Next, to ensure that the reviewers understood the citation eligibility criteria, the study lead (KO) created a test set of 30 citations.

The test set included 5 true positives (i.e. citations that met the eligibility criteria to be included in this systematic review) and 25 true negatives (i.e. citations that did not meet eligibility criteria to be included in this systematic review) [28] . Each reviewer (JG, RN) was then required to complete the test set by assessing the same 30 citations. Reviewers had to achieve a sensitivity in excess of 80% on the test set before they were given access to title/abstract screening. At both title/abstract and full text review, records were removed only if both reviewers agreed to exclude.

Cases with screening conflicts were resolved by review by the study lead (KO). At the completion of full text review, the study lead (KO) reviewed the eligible citations to identify potential duplicates and confirm eligibility. The reference lists of included studies were reviewed to identify any potentially relevant studies not included in the screening set.

J o u r n a l P r e -p r o o f A data extraction tool was developed in REDCap [29, 30] by the study lead (KO) and piloted on five eligible studies. Eligible studies were divided equally among the reviewers for duplicate, independent data extraction into REDCap and Microsoft Excel, followed by conflict resolution by the study lead. When necessary, data was extracted from figures using SourceForge Plot Digitizer (http://plotdigitizer.sourceforge.net/).

Outcome data is reported for NIOSH-approved N95 (particulate, including surgical) FFRs or their components only. Other respirator types (e.g. R or P-filter type) were not included in the analysis. FFR "component" was defined as a piece of an N95 FFR that had been cut out with all layers still intact. Intervention arms described by authors as vaporized hydrogen peroxide, hydrogen peroxide gas plasma, and ionized hydrogen peroxide were analyzed together and, for convenience, are collectively referred to as vaporized hydrogen peroxide for the remainder of this review.

Risk of bias was assessed for each study by outcome using a predetermined evaluation matrix which included evaluation based on study design, methodological consistency, population heterogeneity, sampling bias, outcome evaluation, and selective reporting (Appendix C).

All statistical analysis were performed using the R statistical programming language [31] .

Data was meta-analyzed using a random effects model with the R package 'metafor' [32] .

Random effects meta-analyses were employed to present either the pooled absolute value pre / post chemical disinfectant intervention or relative change (from control or no treatment arm).

For both the aerosol penetration and airflow resistance outcomes, the data was presented as an absolute value. For the germicidal outcome, the data was presented as a relative log change J o u r n a l P r e -p r o o f in viral load. The majority of studies evaluated the germicidal effect of chemical disinfectants using viruses; to improve comparability studies on bacteria or bacteriophage decontamination were removed from the germicidal analysis and are instead presented descriptively. Random effects meta-analysis was used to calculate the effect size for each type of chemical disinfectant.

Heterogeneity was assessed by calculating an I 2 statistic from a fixed effect model. The standard deviation for each control and treatment arm within a chemical disinfectant class was calculated from the pooled absolute values post intervention. The sample size represents the total number of replicates for all N95 FFR models included.

A total of 454 records were identified through the initial database search. After removing duplicate citations in Endnote, there were 417 citations remaining. Both reviewers correctly identified all true positives and true negatives in the test set. Title and abstract screening excluded 401, with the review team achieving a kappa of 0.9. At the full text level, the reviewers excluded six of the records, with a kappa of 0.86. An additional three pre-print articles were identified following the initial search, screened in duplicate and included in the analysis, resulting in a total of 13 eligible articles [7, 9, [11] [12] [13] [18] [19] [20] [21] [22] [33] [34] [35] . An overview of the search process, results and reason for exclusions are shown in the PRISMA diagram ( Figure 1 ).

Nine studies originated from the United States, three from East Asia, and the remaining one originated from Canada. The studies included a total of 58 intervention arms, including sodium hypochlorite (n = 21), liquid hydrogen peroxide (liquid H 2 O 2 , n = 4), vaporized hydrogen peroxide (vaporized H 2 O 2 , n = 12), ethanol (n = 9), isopropyl alcohol (n = 3), ethylene oxide J o u r n a l P r e -p r o o f (EtO, n = 6), or other (n = 3). Thirty-five N95 models were evaluated across the 13 studies. The most common models studied were the 3M 1870 (n = 7), 3M 1860 (n = 6) and the 3M 8210 (n = 5). The number of articles evaluating the main study outcomes were aerosol penetration (n = 5), airflow resistance (n = 3), germicidal activity (n = 8), fit (n = 2), changes in physical traits (n = 6), and safety/irritation (n=3). A summary of the intervention arms, outcomes, and number of N95 FFR models evaluated is presented in Table I . The majority of studies reported on outcomes following a single cycle of decontamination. Bergman et al. [7] evaluated aerosol penetration and airflow resistance following three decontamination cycles, and Kenney et al. [20] reported on changes in physical traits following five cycles. Fischer et al. [34] and Kumar et al. [35] reported on FFR fit following three decontamination cycles and 1, 3, 5, 10 and 20 cycles respectively.

Three studies were published as pre-prints [20, 34, 35] , the remaining were peer-reviewed publications.

There were five studies identified that evaluated aerosol penetration following decontamination with a disinfectant [7, 9, 11, 13, 21] , including intervention arms that evaluated sodium hypochlorite (n = 6), liquid H 2 O 2 (n = 3), vaporized H 2 O 2 (n = 5), ethanol (n = 1), isopropyl alcohol (n = 3), and EtO (n = 3) (Table II) . The majority of studies measured initial aerosol penetration. This was done using a continuous airflow of 85 L/min and an aerosol of sodium chloride with a count median diameter of 0.075 ± 0.020 µm, geometric size deviation of <1.86 and mass median aerodynamic diameter ~300 nm in accordance with 42 CFR 84.174 for NIOSH certification testing [36] . The exception was Lin et al [21] who used a challenge aerosol with a count median diameter of 101 ±10 and a geometric size deviation of 2.01 ±0.08. Analysis was limited to studies that used a testing aerosol that adhered to NIOSH certification testing J o u r n a l P r e -p r o o f standards. Some of the included studies, such as Viscusi et al [11] and Bergman et al [7] , evaluated different concentrations of sodium hypochlorite or different methodologies for vaporized H 2 O 2 and therefore had more than one arm included. We choose not to provide a pooled estimate of aerosol penetration between chemical sterilization, as this metric would not be relevant for decision makers. Instead, we calculated mean difference from the random effects model and the I 2 from a fixed effects model for each decontamination type ( Figure 2 ). Studies on sodium hypochlorite showed no change in aerosol penetration post-sterilization, with low study heterogeneity. Studies on ethylene oxide and liquid H 2 O 2 provided consistent findings of no change in aerosol penetration post-sterilization, with low between study heterogeneity. Studies on vaporized H 2 O 2 showed no change in aerosol penetration post sterilization, with large between study heterogeneity driven by three decontamination cycles using the STERRAD® 100S H 2 O 2 Gas Plasma Sterilizer in Bergman et al [7] . Studies on isopropyl provided consistent findings that aerosol penetration post-sterilization was impaired and exceeded 5% (pooled aerosol penetration estimate 20.17% (95% CI 17.13, 23.21). A summary of the aerosol penetration results for each chemical disinfectant type is provided in Table III .

Three studies reported on airflow resistance [7, 9, 21] , including intervention arms that evaluated sodium hypochlorite (n = 3), liquid H 2 O 2 (n = 1), vaporized H 2 O 2 (n = 3), ethanol (n = 1) and EtO (n = 2) ( Table II ). The mean difference for sodium hypochlorite was 0.29 mm H 2 O (95% CI -0.52, 1.11) with an I 2 of ~0.%. It was not possible to include the results from Lin et al [21] in the analysis for sodium hypochlorite as there was no standard deviation (SD) for the treatment arm. The mean difference for EtO was 0. Table III .

Eight studies evaluated the germicidal impact of one or more disinfectants, including two studies on bacteria [13, 22] , five studies on viruses [12, 18, [33] [34] [35] , and one study on bacteriophages [20] . Intervention arms included in the eight studies were sodium hypochlorite (n = 15), vaporized H 2 O 2 (n = 6), ethanol (n = 8) and EtO (n = 1) (Table IV) . No studies evaluated germicidal removal following decontamination with liquid H 2 O 2 or isopropyl alcohol. While all of the studies showed a reduction in viral load post decontamination, there were large differences in the magnitude of the effect, ranging from log 2.04 to 6.38 ( Figure 4 ). N95 FFRs that were inoculated with SARS-CoV-2 virus, then sprayed with 70% ethanol until saturation showed viral levels below the limit of detection of the assay (<10 0.5 TCID 50 /mL) at five, ten, 30 and 60 minutes post-sterilization [34] . However, ethanol was not as effective at eradicating bacteria.

Immersion in an ethanol solution for 10 minutes at concentrations of 50, 70, 80 and 95% resulted in relative survival rates of Bacillus subtilis of 89 ± 6%, 72.01 ± 10.69%, 68 ± 3% and 73 ± 7% respectively. By 24 hours post-sterilization, survival rates declined to 33 ± 8%, 22.28 ± 7.88%, 20 ± 2% and 26 ± 7% [21] .

EtO decontamination removed all viable vesicular stomatitis virus (VSV) from four N95 FFR models. In the same study, a single cycle of vaporized H 2 O 2 also removed all viable VSV and SARS-CoV-2 [35] . One cycle of vaporized H 2 O 2 also resulted in bacteriophage levels (<10 PFU) [20] or viral levels (<10 0.5 TCID 50 /ml) [34] below the detectable limit of the assay on one N95 FFR model each, and no growth of H1N1 on two N95 FFR models at seven days following decontamination with vaporized H 2 O 2 [33] .

A summary of the germicidal results for each disinfectant type is presented in Table III .

Six studies were identified that evaluated changes in FFR physical traits following decontamination (Table V) . Sodium hypochlorite and liquid H 2 O 2 consistently resulted in changes in FFR appearance, including tarnished metallic nosepieces [7, 9, 11, 19] , oxidized staples [7, 19] , yellowing of nose pads [7, 9] , bleeding or fading of lettering [7, 11] , stiffening of filter media and elastic straps [11] , or dissolving nosepieces [7] . The majority of studies that evaluated physical appearance following exposure to vaporized H 2 O 2 , isopropyl alcohol, ethanol and EtO reported that FFR appearance was unchanged.

Changes in FFR odor was evaluated in three studies following disinfectant treatment with sodium hypochlorite, EtO, liquid H 2 O 2 and vaporized H 2 O 2 . Sodium hypochlorite left a characteristic bleach odor on the FFRs [7, 9, 19] . None of the other chemical disinfectants assessed resulted in changes in FFR odor.

A summary of the results of evaluations on changes in physical traits for each disinfectant type is presented in Table III . Two studies evaluated FFR fit following decontamination, and included the following intervention arms: EtO (n = 1), vaporized H 2 O 2 (n = 3) and ethanol (n = 1). Kumar et al. evaluated FFR fit using the PORTACOUNT Fit Tester and two exercises (normal and deep breathing) following multiple decontamination bouts [35] on new (unworn) FFRs. Four FFR models were exposed to multiple bouts of decontamination (EtO: 1 and 3 cycles; vaporized H 2 O 2 : 1, 5, and 10 cycles using the VHP ARD System; 1, 5, 10 and 20 cycles using the STERRAD 100NX sterilizer). A single fit test was then performed with each of the four FFR models for each of the decontamination methods and number of cycles described above. All four FFR models passed the fit test (achieved a fit factor >100 as per Occupational Safety and Health Administration (OSHA) standards) following one, three, five and ten decontamination cycles of vaporized H 2 O 2 using the VHP ARD System, but not with the standard cycle of the STERRAD 100NX sterilizer (fit factor >100 after one cycle, but <100 following five, ten and 20 cycles). A passing fit factor score was achieved following both one and three decontamination cycles of EtO.

Fischer et al. evaluated FFR fit following three cycles of two-hour wear and decontamination using a 3M Aura 9211+/37193. After each wear-decontamination cycle, six fit tests were performed using six replicate FFRs. A fit factor >100 was achieved following one, two and three vaporized H 2 O 2 wear-decontamination cycles. The fit factor ranged from 112 to 200 after one wear-decontamination cycle, 167 to 200 after two cycles and 100 to 200 after three cycles. The six replicate FFRs achieved a fit factor >100 following one ethanol weardecontamination cycle (fit factor range 153 to 200), but two out of the six replicate FFRs did not achieve a passing fit factor score following two ethanol wear-decontamination cycles (fit factor J o u r n a l P r e -p r o o f range 54.7 to 200), and four replicate FFRs had a fit factor <100 following three cycles (range 29.5 to 200) [34] .

A summary of the fit test results for each disinfectant type is presented in Table III .

Potential health risks associated with disinfectant decontamination was evaluated in three studies. Viscusi et al. [9] reported that, while letting sodium hypochlorite treated FFRs air-dry inner surface at two hours was 0.6 p.p.m. and undetectable at three hours [33] . A summary of the results of evaluations for potential safety risks is presented in Table III .

A detailed risk of bias assessment can be found in Appendix D. Overall risk of bias for aerosol penetration was low in all studies. Risk was moderate in one study for airflow resistance J o u r n a l P r e -p r o o f due to population heterogeneity (i.e. FFRs not obtained from the same lot) and potential selective reporting. Moderate risk of bias for all germicidal outcomes was primarily due to the use of visual assays and population heterogeneity, while moderate risks for fit evaluations were due to population heterogeneity and missing methodological details. Physical trait assessments had moderate to high risks of bias for various reasons, but unblinded outcome evaluation was common across all studies. For safety/irritation evaluations, no controls were used and two studies lacked enough details to assess all risk categories, resulting in moderate to high risks of bias for this outcome.

This is the first systematic review to synthesize the existing evidence on using chemical disinfectants to decontaminate N95 FFRs. We found that a single cycle of vaporized H 2 O 2 successfully removes viral pathogens without affecting airflow resistance or fit, and maintains an initial filter penetration of <5%, with little change in FFR physical traits. Further research is required before the acceptability of decontamination using liquid H 2 O 2 can be determined.

Sodium hypochlorite, ethanol, isopropyl alcohol and EtO are not recommended due to safety concerns and/or adverse effects on FFR function. This systematic review identified five studies that reported on changes in aerosol penetration following decontamination with a chemical disinfectant. NIOSH has established a 95% filter efficiency standard (i.e. aerosol penetration of <5%) for N95 FFR [37] . Results showed that filter efficiency >95% was generally maintained following decontamination with sodium hypochlorite, liquid H 2 O 2 , vaporized H 2 O 2 and EtO under laboratory test conditions. However, while the majority of studies evaluating sodium hypochlorite and vaporized H 2 O 2 reported a post-decontamination aerosol penetration of <5%, there were conflicting findings in one study J o u r n a l P r e -p r o o f for each method. Four of the five studies that evaluated sodium hypochlorite reported that filter penetration of <5% was maintained following submersion (0.3 to 5.25%, 30 min submersion) [7, 11, 12] or wiping (3 times) with hypochlorite wipes [13] ; however, the study by Lin et al., which used a different aerosol penetration testing protocol, reported penetration values that exceeded 5% for particle sizes >60 nm (0.5%, 10 min submersion) [21] . Vaporized H 2 O 2 maintained filter performance in the majority of the studies where it was evaluated, with the exception of Bergman et al. [7] . In this study, three 55-minute cycles with the STERRAD® 100S [38] . Using NIOSH testing standards, the report showed that filter efficiency of the 3M 1860 FFR was 99.6% ± 0.2 following fifty cycles of vaporized H 2 O 2 , safely exceeding the NIOSH requirement of 95%.

Findings from the included studies showed that submersion in ethanol and isopropyl alcohol resulted in significantly higher filter penetration values that exceeded the levels permissible by NIOSH. This is not surprising, as solvents such as isopropyl alcohol have been shown to J o u r n a l P r e -p r o o f eliminate the electrostatic charges on the FFR filter [39, 40] , and the filter efficiency of an uncharged media is typically ten-fold lower than a charged media [41] .

In addition to standards for aerosol penetration, NIOSH has also established standards for airflow resistance of N95 FFRs (peak average inhalation of 35 A successful decontamination protocol must also remove infectious pathogens from the [38] . The ability of vaporized H 2 O 2 to eradicate virus and bacteria from N95 FFRs could be further enhanced by allowing the FFR to sit for an extended time period between re-use.

It is well established that bacteria and virus levels on surfaces decrease over time [42, 43] . For example, the infectiveness of SARS-CoV-2 declines by at least one log every 24 hours across a variety of surfaces [44] . Therefore, building a "holding period" into the decontamination protocol, where FFRs sit for 5-7 days following exposure to vaporized H 2 O 2 , would further guarantee the absence of viable viral particles.

FFR fit is an important outcome of interest when considering whether a decontamination protocol is acceptable to use. Improper fit results in an inadequate seal of the FFR against the wearer's face, reducing the FFR's ability to prevent particle penetration [45] . Two studies evaluated FFR fit following decontamination. FFR fit was not affected by one cycle of decontamination with ethanol [34] or by multiple cycles of EtO or vaporized H 2 O 2 [34, 35] . The multiple decontamination cycle results from Kumar et al. [35] should be interpreted with some caution, however, as FFR fit was evaluated using only normal breathing and deep breathing exercises. The additional exercises from the OHSA protocol that involve movement of the mouth, head and body (to mimic movements performed by health care workers) were not included in the evaluation. Therefore, it is not clear if a fit factor >100 would have still been achieved had the full OHSA protocol been used. Additionally, the FFRss that were evaluated were new and had not been worn by health care workers prior to decontamination, or between decontamination cycles. There is evidence that FFR fit deteriorates through repeated donning and doffing [46] , therefore the number of decontamination and re-use cycles that can be applied to a FFR will be limited by physical stress imposed by both decontamination and donning and doffing. The FDA-commissioned Bioquell report [38] also evaluated fit following multiple Fischer et al. [34] are more representative of the real world, as it involved three cycles of decontamination and two-hour wearing. The FFR model they evaluated (3M Aura 9211+/37193) maintained a fit factor >100 following three cycles of vaporized H 2 O 2 ; however, they used a small sample size (six replicates of one FFR model). Further testing should be conducted using FFRs worn by health care workers prior to and between decontamination cycles in order to confirm whether FFRs maintain fit in the real-world setting following multiple decontamination cycles with vaporized H 2 O 2 .

Sodium hypochlorite consistently resulted in significant changes to FFR appearance.

Some of the changes reported were substantial enough that they could result in changes in FFR fit or comfort, such as stiffening of filter media and elastic straps and dissolving of half of the FFR nosepiece. Sodium hypochlorite also resulted in a bleach odor on the FFRs that would be unpleasant to the wearer. Some chlorine off-gassing was observed on FFRs that had been submerged in sodium hypochlorite and rehydrated, which the authors felt could be significant given that rehydration of the FFR could be compared with moisture in the exhaled breath of an individual wearing the FFR [9] . Therefore, as low-level exposure to chlorine may occur when wearing an FFR that has been decontaminated by submersion in sodium hypochlorite, sodium hypochlorite is not recommended for FFR decontamination and reuse. There are also potential safety concerns with the use of EtO. EtO in itself is hazardous and a known human carcinogen, and decontamination with EtO requires a lengthy aeration process to remove residual EtO [47] .

Following a 12-hr aeration cycle, Salter et al. did not find any EtO on six FFRs models, but did find traces of a hazardous contaminant, ethylene glycol monacetate [19] . Whether residual EtO remains on N95 FFRs when using a shorter aeration cycle, such as the four-hour cycle used by Viscusi et al. [9, 11] , is unknown. Given the potential safety concerns and the impracticality of a lengthy aeration cycle, the results of this review do not endorse EtO sterilization for N95 decontamination, which is consistent with the CDCs recommendations [48] . Residual levels of liquid and vaporized H 2 O 2 were reported in one and two studies respectively, though they remained within established safety limits, and in the case of vaporized H 2 O 2 , decreased over time. A "holding period" following decontamination, as described above, would provide extra assurance that hydrogen peroxide levels were under the permissible exposure limit set by OHSA (1 ppm) [49] .

The moderate overall risks of bias noted for germicidal outcomes are largely due to the use of unblinded assays to quantify bacterial and viral loads; although these are generally accepted as appropriate means of pathogen quantification, counting of plaques and colonies may be subjective. Assessments of physical appearance and odor were unblinded in all cases and results were often reported as additional comments about observations of damage, rather than through a systematic procedure for all FFR models, which makes it unclear to what degree sampling and assessment biases may have influenced these outcomes. The potential for bias in evaluations of post-decontamination safety was increased by the lack of control group in any study which reduces confidence in accuracy of the machine measurements.

Although this systematic review provides valuable information regarding the use of chemical disinfectants for the decontamination of N95 FFRs, a number of limitations must be acknowledged. Each study used a different combination of FFR types. In order to address this, we aggregated across FFR types within each study, treating the pooled replicates across FFR types as our statistically independent sampling unit. This is appropriate for our research question aimed at performance of FFRs in general (where we assume little difference between FFR types). If there are large differences between FFR types, our approach might artificially inflate our sample size; if this were the case it would unlikely change our findings, due to the consistency of these studies conclusions and low heterogeneity. For ease of completion, our review was limited to studies available in English or French, therefore evidence available in other languages is not included in our analysis. With the exception of the fit data from Fischer et al. [33] , all of the reported outcomes were evaluated on new, unworn FFRs. It is not clear whether extended FFR use prior to decontamination would alter our findings. Future work evaluating decontamination of N95 FFRs should perform testing in real-world conditions, using N95 FFRs that have been worn by health care workers in the clinical setting prior to decontamination. Finally, we did not have access to unpublished data, such as quality assurance work performed at hospitals or work performed by industry and submitted to regulatory agencies. For example, the FDA recently reissued Emergency Use Authorizations (EUA) and no longer authorize decontamination or reuse of respirators that have exhalation valves. In the current published literature, the reasons for this EUA are not evident [50] .

We identified thirteen studies that evaluated decontamination of NIOSH-approved N95 FFRs using a chemical disinfectant. Of these, the most promising chemical disinfectant evaluated J o u r n a l P r e -p r o o f was vaporized H 2 O 2 . A single cycle of vaporized H 2 O 2 successfully removes viral pathogens without affecting airflow resistance or fit, and maintains an initial filter penetration of <5%, with little change in FFR appearance. Residual hydrogen peroxide levels following decontamination were within safe limits. More than one decontamination cycle of vaporized H 2 O 2 may be possible but further information is required on how multiple cycles would affect FFR fit in a real world setting before the upper limit can be established. Although immersion in liquid H 2 O 2 does not appear to adversely affect FFR function, there is no available data on its ability to remove infectious pathogens from FFRs or its impact on FFR fit. Sodium hypochlorite, ethanol, isopropyl alcohol and EtO are not recommended due to safety concerns or negative effects on FFR function.

This systematic review provides valuable data on the efficacy and safety of decontaminating N95 FFRs using chemical disinfectants. At the time of this review, published data only supports one disinfectant approach. Literature in this area is rapidly evolving as researchers work to recognize solutions to widespread shortages in N95 FFRs. These circumstances are amenable to a living review that allows for the rapid identification and incorporation of important new data. Recognizing this need, our group has initiated a living review with an open-access database [51] ; the most recent scoping review update on 25-Jun-2020 identified 11 new publications evaluating decontamination of NIOSH-approved N95 FFRs using chemical disinfectants [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] , none of which would significantly alter the findings of this systematic review.

It is important that future studies on decontamination employ approved N95 FFRs and techniques for testing (e.g. NIOSH testing standards for aerosol penetration and airflow resistance). This is particularly relevant when evaluating germicidal effects, where significant heterogeneity in pathogen selection, application procedures, and assays were observed. It would be prudent for researchers working in this area to both consider the available literature on disinfectant efficacy and consult with their regulatory agencies about the most up to date testing requirements.

J o u r n a l P r e -p r o o f No change in aerosol penetration, aerosol penetration <5% maintained following one (n = 2) and three decontamination cycles (n = 1)

Aerosol penetration exceeded 5% for 4 of 6 FFR following 3 decontamination cycles (n = 1)

No change in airflow resistance, NIOSH standards maintained (n = 2)

No viable virus, or virus or bacteriophage below detectable assay limit (n = 4)

Fit factor >100 achieved following 1, 3, 5, 10 cycles with the VHP ARD system, but only following one cycle using the STERRAD 100NX (n = 1) b Fit factor >100 achieved following 3 X 2h wear + decontamination cycles (n = 1)

Slight tarnishing of metallic nosebands (n = 2) No changes in physical appearance (n = 3)

No changes in odor (n = 3)

Did not deposit significant quantities of toxic residues on the FFRs (n = 2)

J o u r n a l P r e -p r o o f Fit factor >100 achieved for 6 replicate FFRs following 1 decontamination cycle, but not all 6 replicates achieved a fit factor >100 following 2 and 3 decontamination cycles (n = 1)

No changes in physical appearance (n =1) Not assessed

Aerosol penetration exceeded 5% for particle penetration (n = 1)

Aerosol penetration post-sterilization exceeded 5% for particles larger than 76 nm (n = 1) 

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

Interim infection prevention and control recommendations for patients with suspected or confirmed Coronavirus Disease 2019 (COVID-19) in healthcare settings

Personal Protective Equipment Supply Chain: Lessons Learned from Recent Public Health Emergency Responses

Hospital preparedness for severe acute respiratory syndrome in the United States: views from a national survey of infectious diseases consultants

The national COVID-19 medical equipment clearinghouse

Conserving supply of personal protective equipment-A call for ideas

Evaluation of multiple (3-Cycle) decontamination processing for filtering facepiece respirators. DigitalCommons@University of Nebraska -Lincoln

Impact of three cycles of decontamination treatments on filtering facepiece respirator fit

Evaluation of five decontamination methods for filtering facepiece respirators

Impact of three biological decontamination methods on filtering facepiece respirator fit, odor, comfort, and donning ease

Effect of decontamination on the filtration efficiency of two filtering facepiece respirator models

Development of a test system to apply virus-containing particles to filtering facepiece respirators for the evaluation of decontamination procedures

Cleaning of filtering facepiece respirators contaminated with mucin and Staphylococcus aureus

Effectiveness of three decontamination treatments against influenza virus applied to filtering facepiece respirators

Effects of ultraviolet germicidal irradiation (UVGI) on N95 respirator filtration performance and structural integrity

Ultraviolet germicidal irradiation of influenzacontaminated N95 filtering facepiece respirators

Effects of relative humidity and spraying medium on UV decontamination of filters loaded with viral aerosols

Development of a test system to evaluate procedures for decontamination of respirators containing viral droplets

Analysis of residual chemicals on filtering facepiece respirators after decontamination

Hydrogen peroxide vapor sterilization of N95 respirators for reuse

Filter quality of electret masks in filtering 14.6-594 nm aerosol particles: Effects of five decontamination methods

Relative survival of Bacillus subtilis spores loaded on filtering facepiece respirators after five decontamination methods

CDC. Decontamination and reuse of filtering facepiece respirators

Decontaminating N95 and SN95 masks with Ultraviolet Germicidal Irradiation (UVGI) does not impair mask efficacy and safety: A Systematic Review

Microwave-and Heat-Based Decontamination of N95 Filtering FacepieceRespirators (FFR): A Systematic Review

Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement

The untold origin story of the N95 mask

Quality control for crowdsourcing citation screening: the importance of assessment number and qualification set size

The REDCap consortium: Building an international community of software platform partners

Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support

R: A language and environment for statistical computing. R Foundation for Statistical Computing

Conducting Meta-Analyses in R with the metafor Package

Disinfection of N95 respirators by ionized hydrogen peroxide in pandemic coronavirus disease 2019 (COVID-19) due to SARS-CoV-2

Assessment of N95 respirator decontamination and re-use for SARS-CoV-2

N95 Mask Decontamination using Standard Hospital Sterilization Technologies

Electronic Code of Federal Regulations (42 CFR 84.174). Filter efficiency level determination test-non-powered series N, R, and P filtration

National Institute for Occupational Safety and Health. NIOSH guide to the selection and use of particulate respirators

Final report for Bioquell HPV decontamination for the reuse of N95 respirators

The effects of particle charge on the performance of a filtering facepiece

Effect of Solvent Exposure on the Filtration Performance of Electrostatically Charged Polypropylene Filter Media

Information and FAQs on the Performance, Protection, and Sterilization of Face Mask Materials

Survival of influenza viruses on environmental surfaces

The survival and inactivation of enteric viruses on soft surfaces: A systematic review of the literature

Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1

Effect of fit testing on the protection offered by n95 filtering facepiece respirators against fine particles in a laboratory setting

Impact of multiple consecutive donnings on filtering facepiece respirator fit

Strategies for optimizing the supply of N95 respirators: Crisis/alternate strategies

COVID-19) Update: FDA Reissues Emergency Use Authorizations Revising Which Types of Respirators Can Be Decontaminated for Reuse

Live Scoping Review of N95 and Surgical Facemask Decontamination and Reuse: A Scoping Review Protocol

Aerosolized Hydrogen Peroxide Decontamination of N95 Respirators, with Fit-Testing and Virologic Confirmation of Suitability for Re-Use During the COVID-19 Pandemic

An Efficient Ethanol-Vacuum Method for the Decontamination and Restoration of Polypropylene Microfiber Medical Masks & Respirators

Analysis of SteraMist ionized hydrogen peroxide technology in the sterilization of N95 respirators and other PPE: a quality improvement study

Characterization of a novel, low-cost, scalable vaporized hydrogen peroxide system for sterilization of N95 respirators and other COVID-19 related personal protective equipment

Decontamination of face masks and filtering facepiece respirators via ultraviolet germicidal irradiation, hydrogen peroxide vaporisation, and use of dry heat inactivates an infectious SARS-CoV-2 surrogate virus

Development of a highly effective low-cost vaporized hydrogen peroxide-based method for disinfection of personal protective equipment for their selective reuse during pandemics

Effectiveness of Ultraviolet-C Light and a High-Level Disinfection Cabinet for Decontamination of N95 Respirators

Effects of Sterilization With Hydrogen Peroxide and Chlorine Dioxide on the Filtration Efficiency of N95, KN95, and Surgical Face Masks

Scalable In-hospital Decontamination of N95 Filtering Facepiece Respirator with a Peracetic Acid Room Disinfection System

Validation of N95 filtering facepiece respirator decontamination methods available at a large university hospital

Effect of various decontamination procedures on disposable N95 mask integrity and SARS-CoV-2 infectivity