Rapid Review of SARS-CoV-1 and SARS-CoV-2 Viability, Susceptibility to Treatment, and the Disinfection and Reuse of PPE, Particularly Filtering Facepiece Respirators

In the COVID-19 pandemic caused by SARS-CoV-2, hospitals are often stretched beyond capacity. There are widespread reports of dwindling supplies of personal protective equipment (PPE), particularly N95-type filtering facepiece respirators (FFRs), which are paramount to protect frontline medical/nursing staff, and to minimize further spread of the virus. We carried out a rapid review to summarize the existing literature on the viability of SARS-CoV-2, the efficacy of key potential disinfection procedures against the virus (specifically ultraviolet light and heat), and the impact of these procedures on FFR performance, material integrity, and/or fit. In light of the recent discovery of SARS-CoV-2 and limited associated research, our review also focused on the closely related SARS-CoV-1. We propose a possible whole-of-PPE disinfection solution for potential reuse that could be rapidly instituted in many health care settings, without significant investments in equipment.


Introduction
In pandemic situations, such as the ongoing COVID-19 pandemic, hospital resources are frequently stretched beyond capacity, as has already occurred in many countries across the globe [1]. Preventing the spread of COVID-19 to and from health care workers and patients relies on the availability and effective use of personal protective equipment (PPE) [1]. PPE includes masks, eye protection, gloves, gowns, and, for aerosol-generating procedures in particular, N95 filtering facepiece respirators (FFRs) or equivalent [2].
There has been a global shortage of PPE during the current pandemic [3], and the World Health Organization acknowledges the current global stockpile has been insufficient, especially for surgical masks and FFRs [2]. The supply of gowns and eye protection is also expected to be insufficient. Coordinating the supply chain of PPE in the midst of a pandemic with many closed borders and reduced freight is challenging. Individual behavior becomes a factor when people are scared or ill-informed [4]. Ideally, people need to have trust in the systems set up to support them in the workplace [5]. When this trust is compromised and local supply chains are affected, inappropriate use of PPE can occur, sometimes with theft of PPE further affecting supply, despite best-practice guidance on its use [2]. This shortage of PPE in the face of an exponential increase in demand also seems to have encouraged the production of counterfeit FFRs, potentially creating additional risks for health care workers [6]. Unsurprisingly, a call for ideas on conserving PPE was made through the Journal of the American Medical Association (JAMA) in March 2020 [7].
One key recommendation to deal with the unprecedented shortage of PPE has been disinfecting and reusing PPE, particularly FFRs [8]. It is acknowledged that PPE items are designed for single use. However, the reality during the course of the pandemic is that reuse has been undertaken by many health care workers across the world out of necessity [9]. Therefore, understanding how to effectively disinfect PPE items for potential reuse was the focus of this study. Given the ability of PPE decontamination and reuse to rapidly address supply issues close to the "frontline" (thus avoiding many of the upstream disruptions to the supply chain), we carried out a rapid review to summarize the relevant literature with three specific aims-first, to examine the current knowledge about the viability of SARS-CoV-2 on a variety of surfaces; second, to determine the efficacy of key disinfection procedures against SARS-CoV-2, specifically ultraviolet light and heat; and third, to determine the impact of these procedures on FFR performance. In light of the very recent discovery of SARS-CoV-2, our review also focused on SARS-CoV-1, a closely related sister clade virus from the same species [10]. Further, based on current knowledge, a possible whole-of-PPE solution for potential reuse is suggested that could be rapidly instituted in many health care settings without significant investments in equipment.

Methods
We carried out a rapid review, as it can provide valuable information for decision-making in a timely manner, particularly important in a pandemic scenario. There is no consensus in the literature for either the definition of a rapid review or the most appropriate methodology [11]. Nonetheless, our search strategy involved PubMed, Web of Science, and Google Scholar (in this order).
Searches were restricted to publications from 1 January 2003 (as the first recorded human infection of SARS-CoV-1 occurred in November 2002 [12]) and 18 July 2020. A number of keywords were used alongside the term "SARS" in combination with Boolean operators ( Table 1).
The results from the literature search (Table 1) had their title and/or abstract screened by the first author, and those deemed to be of relevance were immediately exported to a bibliographical software, and the respective full text subsequently obtained. Where appropriate, other articles were included, if discovered while examining the full text of individual studies. Any original study reporting quantitative data addressing any one of the three aims of this rapid review was included for data extraction. However, in the context of this rapid review, we acknowledge as a limitation the lack of a formal assessment of the evidence quality of the included studies. ((TI = SARS) OR (AB = SARS)) AND (TI = (ultraviolet OR UV OR heat OR N95 OR PPE OR "personal protect*" OR surviv* OR viability OR disinfect* OR decontam* OR inactivat*)) OR (AB = (ultraviolet OR UV OR heat OR N95 OR PPE OR "personal protect*" OR surviv* OR disinfect* OR decontam* OR inactivat* OR viability)) Google Scholar †~1 82,000 SARS AND (ultraviolet OR UV OR heat OR inactivation OR inactivate OR decontaminate OR decontamination OR disinfect OR disinfection OR N95 OR PPE OR "personal protective" OR "personal protection" OR survival OR survivorship OR viability) ‡ Search included three databases: Web of Science Core Collection, Current Contents Connect, and SciELO Citation Index. † Search excluded patents; results were sorted automatically by relevance based on the search engine's own ranking algorithms, and the top 2500 results were screened.

Filtering Facepiece Respirators
This rapid review has focused in particular on FFRs, which remain at the center of the PPE shortage worldwide. There is conflicting evidence on the superiority of FFRs over standard surgical masks to protect frontline staff against viral respiratory infections during standard care [13][14][15][16][17]. However, approximately 3.2% of patients with SARS-CoV-2 in China required intubation during the first wave [18], and evidence from the SARS epidemic showed that doctors and nurses involved in the early critical care period and endotracheal intubation of patients were over 13 times more likely to acquire SARS-CoV-1 infection themselves [19]. Thus, FFRs are particularly important for health care workers during aerosol-generating procedures in patients with SARS-CoV-2. Lack of adequate protection could result in a significant loss of highly specialized health care workers in an already strained workforce, exacerbating community transmission. Therefore, avoidance of cross contamination is critical in all health care settings, and FFRs play a key role.
We note that while the term "N95" has achieved global reach, it actually refers to the US National Institute for Occupational Safety and Health (NOISH) certification [20,21]. N95 FFRs are defined as respirators not resistant to oils, but with a particle filtration efficiency ≥95% when challenged with sodium chloride particles of a median diameter of 0.075 µm at a flow rate of 85 L/min [20,22]. The equivalent Conformité Européen (CE) certifications are FFP2 and FFP3 respirators, which have minimum required particle filtration efficiencies of 94% and 99%, respectively [21]. Thus, we have referred to FFRs instead of N95 throughout this manuscript, whenever referring to this generic group of respirators.

Virus Viability
During the SARS epidemic, SARS-CoV-1 was recovered from a variety of inanimate objects and surfaces, e.g., buttons of drinking water fountains, chairs, bookshelves, tables, and edges of a bed [23]. Two small studies from the same group in Singapore showed no SARS-CoV-2 contamination on PPE after some contact with infected patients [24,25], although none of those patients required ventilation support or aerosol-generating procedures. Nonetheless, SARS-CoV-1 has been recovered from door handles in a patient's room [26] and SARS-CoV-2 from uncovered shoes [24]. Such observations led the author of the former study to speculate that virus contamination of these surfaces may have led to infection among health care workers without documented contact with known hospitalized SARS patients [23]. As a result, it is important to understand the viability of SARS-CoV-2 on a variety of surfaces, particularly due to its relevance for PPE and health care/frontline worker protection.
Of note, SARS-CoV-1 remained infectious at room temperature for as long as 9-10 days on plastic petri dishes [31] and on respiratory specimens [30], and 21 days on plastic well plates [32], with limited loss of infectivity after 2 weeks at 4 • C shown by at least two other studies [33,35] (Table 2). However, Pagat et al. [34] showed a 2.5 log 10 TCID 50 /mL viral titre reduction in the first day when the inoculum passed from liquid to dried form, but afterwards it took as many as 42 days for complete inactivation of SARS-CoV-1 (~6.2 log 10 reduction) at room temperature on a glass surface (Table 2).
In addition, two studies [33,37] showed that lower temperatures are more favorable for SARS-CoV-2 viability, with little virus titre decay shown after 2 weeks at 4 • C (Table 3). However, even at room temperature (20-25 • C), the two studies showed that it can take 14 days to achieve a 4.5-5.0 log 10 reduction of SARS-CoV-2 in applied virus droplets (Table 3). Table 3. Studies reporting on the viability of SARS-CoV-2.
Importantly, some studies have determined the viability of SARS-CoV-2 directly on PPE or PPE-derived materials, including FFRs [37][38][39][40] (Table 3). Fischer et al. [39] reported a~4.0 log 10 reduction of SARS-CoV-2 on N95 FFR disks after 24 h but, according to Chin et al. [37], it took 7 days to achieve~4.8 log 10 and~3.0 log 10 reductions in SARS-CoV-2 infectivity on the inner and outer layers of surgical masks, respectively (Table 3). Notably, while yet to be peer reviewed, the study by Kasloff et al. [40] assessed SARS-CoV-2 viability using a soil load of mucin, bovine serum albumin, and tryptone as the inoculum, said to represent typical infectious body fluids of infected patients. That study showed that to achieve a 5.0 log 10 reduction in SARS-CoV-2 at room temperature (~20 • C), it took as long as 14 days on nitrile rubber gloves and 21 days on plastic face shields, N100 respirators, and polyethylene coveralls, with some residual infectivity still remaining on N95 respirators after 3 weeks [40] (Table 3).

Disinfection
A wide variety of potential disinfection methods for PPE have been examined and reported in the literature. These can be characterized as: (1) energetic methods (e.g., ultraviolet, dry and moist heat, and microwave generated steam), or (2) chemical methods (e.g., alcohol, ethylene oxide, bleach, and vaporized hydrogen peroxide). Some of these rapidly and markedly affect N95 particle filtration performance (alcohol [22,[41][42][43][44]), while others require chemical supplies and/or specialized facilities (e.g., ethylene oxide and vaporized hydrogen peroxide), or are not readily scalable to large numbers of PPE (e.g., microwave generated steam). Therefore, the focus of this review is on methods that may be easy to implement rapidly at large scale.

Ultraviolet Germicidal Irradiation (UVGI)
Across the ultraviolet (UV) light spectrum that is invisible to the human eye, for the purposes of this review, there are three classifications according to wavelength: UVA (320-400 nm), UVB (280-320 nm), and UVC (200-280 nm) [45]. UVC light has much stronger germicidal properties than both UVA and UVB [46,47]. UVC is strongly absorbed by RNA and DNA bases, leading to molecular structural damage via a photodimerization process; this results in virus inactivation, as the virus is no longer able to replicate [47,48]. Thus, for PPE decontamination, the focus is on UVC rather than UVA or UVB. There have been seven studies assessing the efficacy of UVGI against SARS-CoV-1 and six against SARS-CoV-2, although one of the latter was reported in such a way that was not deemed to be of relevance (Table 4).  .5 log 10 TCID 50 /mL 50 µL inoculum on stainless steel and N95 disks LOD = 0.5 log 10 TCID 50 /mL UVC 260-285 nm Irradiance 0.55 mW/cm 2 at point of exposure Stainless steel-below LOD (≥4 log 10 reduction) with 330 mJ/cm 2 N95-LOD not reached with 1980 mJ/cm 2 (visually estimated from figure as~3 log 10 reduction), but no data thereafter (i.e., beyond 60 min) Heilingloh 2020 [54] 600 µL 6.7 log 10 TCID 50  'Simulated' saliva:~2.5 log 10 reduction (from~3 to~0.5 log 10 TCID 50 /mL) FBS:~1.1 log 10 reduction (from~2.6 to~1.5 log 10 TCID 50 /mL) Study aimed to demonstrate SARS-CoV-2 inactivation by sunlight (which does not include the UVC spectrum) BSA, bovine serum albumin; CPE, cytopathic effect; FBS, fetal bovine serum; FFR, filtering facepiece respirator; LOD, limit of detection; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; RH, relative humidity; TCID 50 , median tissue culture infectious dose, corresponding to the concentration at which 50% of the experimental cells are infected after inoculation; UVA, ultraviolet light A; and UVC, ultraviolet light C. (!) Study only available as preprint at the time of manuscript preparation and therefore yet to be peer reviewed. Where necessary, the applied dose of ultraviolet light (in mJ/cm 2 ) was calculated by the authors using the standard formula, as the product of irradiance (mW/cm 2 ) and time (seconds).
For SARS-CoV-1, the applied dose of ultraviolet light C (UVC) used varied markedly from 300 to 14,500 mJ/cm 2 , with rather mixed outcomes (Table 4). At the lower end of the spectrum, a~6 log 10 reduction in virus titre was achieved in culture medium with 300 mJ/cm 2 [28] (Table 4). Conversely, in a high-protein solution, an applied dose of 14,500 mJ/cm 2 did not completely inactivate SARS-CoV-1 [45] ( Table 4), most likely due to competitive absorption of UV photons by the protein medium. Of note, a number of these studies were performed on culture media or under conditions that would not reflect the micro-environments likely to be found in FFRs or other PPE destined for disinfection and reuse in a real-world setting. As a result, some of these findings need to be interpreted in the appropriate context and be seen as guidance. Nonetheless, Heimbuch and Harnish (2019) showed complete inactivation (≥4.0 log 10 reduction) of SARS-CoV-1 from FFR coupons in the presence of artificial saliva (mucin) and artificial skin oil (sebum) with an applied UVC dose of 1000 mJ/cm 2 [51].
While there is no doubt that UVC is effective against both SARS-CoV-1 and SARS-CoV-2, efficacy of the applied dose (a function of irradiance and time) appears to be highly dependent on many factors, such as virus titre, inoculum size, the virus medium, and both shape, contours and type of material [30,47,51,58], likely explaining the highly inconsistent findings in the published literature. However, based on the available evidence, it seems that the effect of relative humidity on UVGI efficacy can be considered negligible [59].
Importantly, the applied dose is not necessarily the same as the actual dose the target virus receives at any specific point. While the applied dose is easy to measure experimentally using a radiometer, the received dose at each microscopic location is not. If there are shadowing or absorption effects from the surrounding medium, structures, or surface irregularities, the actual dose reaching the virus will be lower [60,61]. In addition, the penetration of UV across the multiple layers of an N95 FFR may vary from one model and manufacturer to another [62]. There is some limited evidence that the majority (approximately 90%) of captured aerosols occurs on the outer filter layer on an N95 FFR [63], but good disinfection efficacy is still desirable across all the layers.

Heat Treatment
Heat treatment is one of the most common methods utilized for disinfection, including for virus deactivation. Heat induces structural changes in virus proteins, disrupting the specific structures necessary to recognize and bind to host cells [64]. The challenge for heat treatment is to eliminate the virus without damaging PPE, in particular FFRs. Eight studies were found to have examined the efficacy of heat treatment against SARS-CoV-1 and eight against SARS-CoV-2 (Table 5).

Wang 2020 [70](!)
Unreported volume of virus stocks 7.2 log 10 TCID 50 /mL Heating conditions or RH undisclosed 37 • C-after 48 h 6.0 log 10 reduction but some infectivity remained (no data thereafter) 42 • C-after 24 h 6.0 log 10 reduction but some infectivity remained, which disappeared after 48 h 56 • C-30 min (7.0 log 10 reduction) 60 • C-15 min (7.0 log 10 reduction) BSA, bovine serum albumin; FBS, fetal bovine serum; LOD, limit of detection; RH, relative humidity; TCID 50 , median tissue culture infectious dose, corresponding to the concentration at which 50% of the experimental cells are infected after inoculation. (!) Studies only available as preprint at the time of manuscript preparation and therefore yet to be peer reviewed.
Environments with lower temperatures seem to be more favorable for virus viability and increased transmission rates [32,71,72], which also applies to SARS-CoV-1 and SARS-CoV-2 (Table 2). While the efficacy of heat treatment appears to be affected by relative humidity [32], this relationship for both viruses of interest here is unclear, as almost all experimental studies failed to report on this parameter (Table 5). However, the association between temperature and relative humidity was not monotonic for other coronaviruses, with virus survival lowest at moderate relative humidity (50%) [72].
Overall, heat treatment at 60 • C for 60 min would lead to SARS-CoV-1 inactivation according to six studies [31,34,35,45,50,65] (~3.5-8.0 log 10 reductions in a variety of media), and to SARS-CoV-2 inactivation as per six studies [33,37,66,67,69,70] (~4.6-7.0 log 10 reduction) (Table 5). However, one study showed residual SARS-CoV-1 infectivity until 90 min at 65 • C [46]. In addition, Fischer et al.'s findings on dry heat treatment at 70 • C against SARS-CoV-2 are surprising and difficult to interpret, as while their figure indicated a limited~2.2 log 10 reduction in virus titre on stainless steel after 60 min, the decay in the control samples on the same medium at room temperature was~1.5 log 10 over the same period [39].

UVGI
Thirteen published studies and two non-peer-reviewed reports have examined the effects of UVGI on the performance and structure of FFRs (Table 6).  Slight decrease in particle penetration, estimated as up to~1 percentage point. Small increase in flow resistance (<6% of the original value), independent of applied UV dose. At ≥710 J/cm 2 there was major loss of bursting strength for most respirator layers tested, some as much as 90%. For some layers of certain models (3M 9210 and K-C 46727) loss >80% occurred at 470 J/cm 2 . At 590 J/cm 2 the mean strap breaking strengths decreased by 10-21%. The lowest applied dose tested of 120 J/cm 2 reduced the bursting strength of the four models tested by 11% to 42% (depending on layer and model). Up to 20 cycles of UVGI treatment (20,000 mJ/cm 2 ) did not have a meaningful effect on fit, airflow resistance, or particle penetration for any model tested. Strap strength was unaffected by 10 UVGI cycles, but 20 cycles had some effect on certain models. Study difficult to interpret as aspects of UV disinfection were insufficiently reported. Negligible effect on filtration performance after 2 cycles (~3960 mJ/cm 2 ), but more marked after 3 cycles, although still within acceptable range.
Liao 2020 [44] Sterilizer cabinet 8 W bulb UVC (254 nm) Irradiance not described 10 cycles of 30 min 15 × 15 cm pieces of meltblown fabric, described as most important N95 FFR layer The ten 30 min cycles did not affect the fabric's filtration efficiency. In the absence of information on irradiance, it is not possible to ascertain the actual applied UVC dose. Studies have assessed a range of parameters, including the FFR's particle filtration efficiency, material strength, and "fit". The latter quantifies how tight the seal between the respirator and the wearer's face is, being generally derived from the ratio of non-toxic sodium chloride particles generated by the testing equipment present in the ambient air to that within the respirator on the wearer [86] (compared with a self-administered "fit check" prior to use, to determine if the FFR seals properly to the wearer's face prior to use [87]).
Exposure methodology for UVGI varied somewhat, ranging from a single cycle to as many as 20 cycles, or UVGI exposure to the outer-facing surface of the respirators or to both surfaces (Table 6). Seven studies (with applied UVC doses ranging from 180 to 6900 mJ/cm 2 ) [39,41,[77][78][79][80][81] reported negligible effects on FFR filter aerosol penetration, filter airflow resistance, fit, odor detection, comfort, donning difficulty, or physical appearance ( Table 6). Heimbuch and Harnish [51] evaluated the effects of multiple UVGI cycles on 15 different N95 FFR models; up to 20 UVGI cycles (total applied UVC dose 20,000 mJ/cm 2 ) did not have a meaningful effect on fit, airflow resistance, or particle penetration for any model. Strap strength was unaffected by 10 UVGI cycles (total applied dose 10,000 mJ/cm 2 ), but 20 cycles (20,000 mJ/cm 2 ) affected the material integrity of straps in certain models [51]. In other studies, applied UVC doses of 10,000 mJ/cm 2 had negligible effects on two N95 FFR models tested (e.g., particle filtration, polymer structure, and tensile strength) [85], and~18,000 mJ/cm 2 reduced fit scores of three N95 FFR models but which still remained within the required performance range [42] (Table 6).
Lindsley et al. [82] estimated the cumulative effect of extremely high exposures to UVC on N95 FFRs, in order to mimic repeated cycles of UVGI treatment. Their lowest applied dose of 120,000 mJ/cm 2 reduced the bursting strength of the four N95 models tested by 11% to 42% (depending on the model and the individual layer), with minor effects on filter aerosol penetration and filter airflow resistance [82] (Table 6). An applied dose of 590,000 mJ/cm 2 reduced the breaking strength of straps from the four N95 FFR models tested by 10% to 21% [82].
Of interest is the research letter by Ozog and colleagues [83], which reported marked differences in the effects of UVGI on fit testing among N95 FFR models. While one model was unaffected by a total applied UVC dose of 60,000 mJ/cm 2 , others failed fit testing after a single treatment cycle (i.e., 3000 mJ/cm 2 ) ( Table 6). It was concerning that some models failed fit testing even before treatment [83]. Therefore, while it seems that in general FFRs will withstand a total applied UVC dose ≥20,000 mJ/cm 2 , the evidence shows that findings from one model cannot be extrapolated to others. Further, it appears that Fischer et al. were the only researchers to investigate the combined effects of cycles of wear and UVC disinfection [39], but they have not examined more than three cycles of 2 h of wear and 1980 mJ/cm 2 disinfection, so further studies are required looking at more repeated FFR disinfection and reuse, particularly involving extended use. Table 7 summarizes 20 studies that examined the effects of heat treatment on the performance and structure of FFRs, most of which (14) were carried out in the current COVID-19 pandemic, including four that are yet to be peer reviewed. These studies consisted of dry heat treatment or moist heat treatment, with a limited number examining the effects of steam treatment (Table 7). While some studies have worked with temperatures above 100 • C (e.g., [41,77,88]), we have focused on lower temperatures, as most FFRs appear to be made of polypropylene [89], whose maximum operating temperature would be below 100 • C [90].

Heat Treatment
While moist heat has been reported to be better than dry heat at disinfection [91], the majority of studies have focused on dry heat (Table 7). This is likely because, in theory, such treatment could be relatively easily replicated, using for example, any oven with a thermostat. For two models (3M 8210 and Moldex 2200), there was a reduction in fit; for one model (3M 1860), there was a small increase in odor response; but both effects were deemed to be negligible. 3M 1870 samples experienced a slight separation of the inner foam nose cushion (some to a lesser or greater degree) from the FFR body. Authors concluded that moist heat incubation unlikely to lead to significant changes in fit, odor detection, comfort, or donning difficulty.

Lore 2012 [81]
Moist heat incubation Uncertain temperature, but likely 65 • C for 20 min, unknown RH 3M 1860s, 3M 1870 There was no significant decrease in filter performance. Rudimentary testing showing no effect on particle filtration efficiency after 5 cycles. Primary aim of the study seems to have been to demonstrate that is feasible to heat treat N95 FFRs at home using kitchen utensils on a gas stove.
Fischer 2020 [39] Dry heat (oven) at 70 • C, unknown RH Up to 3 cycles of 2 h of wear and likely 60 min of treatment 3M 9211+ There was a progressive reduction in filtration performance of respirators, which was below acceptable range after 3rd cycle.
Harskamp 2020 [88] Autoclave 34 min cycle: 12 min pre-heating, 17 min steam treatment at 121 • C, and 5 min drying Up to 3 cycles FFP2: 3M 1862+, 3M 9322+, Maco Pharma ZZM002, and San Huei 2920V FFP3: Safe Worker 1016 50% of FFP3 respirators were deformed and failed seal checks; all other respirators were intact upon inspection. The 3M 1862+ was the only respirator that continued to perform within the required range after 3 treatment cycles. All other respirators had particle filtering efficiency affected after one treatment cycle, performing below the required range (particularly for smaller particles-0.3 µm), with the magnitude of the reduction in performance varying between models. Methods lacking details, and amongst other things, unclear whether there was a cool down period between cycles (due to short duration). Authors concluded that 20 cycles "did not affect fit testing performance", but few details provided. There was a reduction in fit observed for all masks after one cycle, but the rate of reduction was highly variable, and most passed fit testing. Study was not standardized and it is difficult to interpret, but key message was variability between models. Note that one respirator failed the fit testing before any treatment.
Ou 2020 [43] Dry heat (oven): 30 min at 77 • C, unknown RH Steam treatment: 30 min with water vapor (i.e.,~100 • C) 3M 8210 10 cycles of dry heat or steam treatment had negligible effects on particle filtration efficiency. Dry heat treatment had no effect on the N95 fit. 5 cycles of steam treatment led to failure in fit testing, with evidence of some effect appearing after just one cycle.

Study Treatment Details FFRs Key Findings
Tsai 2020 [22] Dry heat at 92 • C, unknown RH Moist heat at 92 • C, 85% RH Treatment duration not provided, but context suggests 15 min One unidentified N95 respirator Minimal information provided, other than basic data showing no effect on particle filtration efficiency after 4 cycles (each 24 h apart) of either dry or moist heat treatment.

3M 1860
No reported change in "shape"; no details provided of fit testing results but authors imply that respirators were largely unaffected after 3 h treatment.
Minor reduction in filtration efficiency for bacterial aerosols (from 99% to 97% after 3 h) but still within acceptable range (i.e., ≥95%). The heat treatment studies were very inconsistent in regard to the adopted temperatures and length of exposure (Table 7). Thus, the findings reported from the 20 studies described in Table 7 have been further condensed in Table 8 to facilitate their interpretation. The studies that have utilised an upper temperature range of 90-100 • C (potentially damaging to plastic polymers) and found no effect on FFR fit testing or filtering performance (including two using steam treatment [44,94]) had a relatively short total cumulative treatment time ≤60 min (Table 8). In contrast, a number of studies have shown that both dry and moist heat treatment of FFRs in the range of 70-85 • C were possible for extended periods of time (Table 8), without marked effects on respirator performance and/or fit (Table 7). These have ranged from 90 to 600 min (cumulative) at 70 • C using dry or moist heat [39,68,93,96,97], to 150-400 min (cumulative) at 85 • C with low-moisture (30% RH) to 100% RH [44,92] (Table 8). Liao et al. [44] also reported that the particle filtration efficiency of the meltblown fabric of N95 FFRs was not markedly affected after 1500 min of dry heat treatment at 75 • C or after 1000 min at 85 • C and 30% RH (Tables 7 and 8). Cells contain the citations for a given study, with the corresponding temperature tested and the reported cumulative treatment time after which FFR performance/fit remained within the acceptable range. Studies where temperatures tested were above 100 • C have been excluded. 1 There was, however, separation of inner foam in one FFR model. 2 Temperature was 86 • C but rounded for simplicity. 3 Total time is an estimate. 4 Average temperature and length of exposure are rough estimates. 5 Low-moisture heat (30% RH). (!) Studies only available as preprint at the time of manuscript preparation and therefore yet to be peer reviewed.
Of note, even at the lower end of the temperature range (60 • C), three studies from the same group using moist heat incubation (80% RH) [78][79][80] reported that while cumulative treatment times of 30-90 min did not significantly affect FFR performance and fit, there was separation of the inner foam nose cushion for a given respirator model (Table 7). In addition, as reported for UGVI treatment, some FFRs failed fit or particle filtration testing even before treatment [92,95,97], and the ability of different models to withstand high-temperature insults varied [88]. Further, while a number of studies have shown that many FFR models can withstand multiple disinfection cycles with heat at 70 to 85 • C (Table 8), the findings reported by Fisher et al. suggest that this may not be necessarily applicable in practice, when treatment cycles are interpolated with periods of actual FFR wear [39] (Table 7).

Viability
The viability of both SARS-CoV-1 and SARS-CoV-2 will vary markedly depending on the material in question, the ambient temperature, the medium in which the virus is deposited, and possibly the initial viral load.
SARS-CoV-2 could potentially remain infectious for many days on inanimate objects (including PPE) under the right conditions in infectious bodily fluids, potentially as long as 3 weeks at room temperature.

Disinfection
We advise against attempts to disinfect and reuse soiled PPE; disinfection and viability studies show a protective effect of protein and aqueous substrata on SARS-CoV-1 and SARS-CoV-2 infectivity. Therefore, it would be ill-advised to attempt to disinfect any PPE that is clearly contaminated upon visual inspection.
The data on UVGI remain scarce, heterogeneous, conflicting, and consequently difficult to interpret. Nonetheless, the existing data indicate that an applied UVC dose of approximately 1000 mJ/cm 2 would likely be effective against SARS-CoV-2 on a relatively flat surface and in the absence of soiling agents (e.g., bodily fluids), leading to a~4 to 5 log 10 TCID 50 /mL reduction in virus titre. However, based on some of the SARS-CoV-2 inactivation data from N95 FFRs [39,56], a conservative dose of 1500-2000 mJ/cm 2 should be considered, given: (i) possible errors in applied dose estimation; (ii) uncertainties regarding the actual susceptibility of SARS-CoV-2 to UVC; (iii) the effects of different materials on SARS-CoV-2 susceptibility to UVC; and (iv) the challenge to reach the inner filtering layers of FFRs [62,98] and overcome potential shadowing effects, so that sufficient UVC is applied to their various segments (e.g., straps).
Unpublished data from our group show that there is minimal UVC radiation on the wearer-facing side of FFRs when the outer side is irradiated (outer 7.34 mW/cm 2 vs. inner 0.10 mW/cm 2 ), with 99-100% blockage of UV light in N95 FFRs also shown by Ontiveros et al. [98]. There are also reports of widespread SARS-CoV-2 infection among frontline medical staff [99], thus, it has to be assumed that SARS-CoV-2 contamination of FFRs would likely occur on both sides, particularly when there is strong evidence that asymptomatic cases may be responsible for the transmission of a large proportion of SARS-CoV-2 infections [100,101]. Therefore, we recommend that both wearer-facing and outer-facing surfaces of FFRs be equally treated at the recommended UVC dose (i.e., at a total dose 3000-4000 mJ/cm 2 ).
For heat treatment, the higher the temperature, the faster the virus inactivation occurs. Conservatively, dry heat treatment at 60-65 • C for 90 min or 70-75 • C for 60 min would most likely lead to inactivation of SARS-CoV-2 on PPE, with the suggested treatment period advisable to ensure adequate heat transfer to the inner layers of FFRs, particularly if a number of respirators are being treated simultaneously (in which case we would caution against stacking them). While we cannot recommend a target RH due to the paucity of data for SARS-CoV-1 and SARS-CoV-2, there is some evidence that higher relative humidity (i.e., moist heat) would likely increase treatment efficacy [91].

Impact of Disinfection on FFRs
If the conservative applied UVC dose of 1500-2000 mJ/cm 2 per FFR surface is adopted (i.e., total dose of 3000-4000 mJ/cm 2 ), it may be possible to subject FFRs to approximately five UVGI disinfection and reuse cycles without compromising respirator function and material integrity. Similarly, using heat treatment at 60-65 • C for 90 min or 70-75 • C for 60 min, five disinfection cycles would likely be possible.
However, the feasibility of multiple disinfection cycles needs to be ascertained for a given FFR model, as there is extensive evidence of variability. Extended use also needs to be considered in the achievable cycle number.
Due to the widespread use of alcohol-based disinfectants, it is important to emphasise that FFRs should not be sprayed with alcohol, as it can remove the electrostatic charge from the respirator filter material, severely reducing the filter's effectiveness at collecting particles, as shown by a number of studies [22,[41][42][43][44].

Disinfection of Other PPE
While this study focused primarily on FFRs, the supply of other PPE will be seriously affected in a pandemic situation, in particular surgical masks (potentially as important as FFRs [16,17]) and isolation gowns [102], but also face shields and eye protection.
Surgical masks-UVGI would not be appropriate for disinfection of surgical masks due to their folded construction. Thus, heat treatment would be the most readily available option, likely at similar levels as FFRs, although 60 min at 70-75 • C would seem more appropriate to maximize thermal viricidal activity deep within mask folds.
Isolation gowns-Heat treatment would be recommended due to their size and folds, as wiping with chemical disinfectants would be laborious and prone to failure. We are not aware of disinfection studies undertaken on isolation gowns, but heat treatment at 60-65 • C for 90 min would be advisable as the plastic polymers that often make up isolation gowns tend to have relatively low maximum operating temperatures, and higher temperatures would likely lead to permanent structural damage.
Goggles and other eye protection-These should be immersed for at least 10 min in a chlorine solution at a conservative dose of 5000 mg/l, which would account for the gradual reduction in chlorine concentration throughout the day. Alternatively, in the absence of any signs of soiling, these could be thoroughly cleaned with an 80% ethanol solution for at least 30 s [75,103]. Afterwards, the goggles/eyewear should be rinsed well with warm water to remove the disinfectant solution, which could otherwise damage the equipment or cause skin irritation for the wearer. In addition, as goggles and other eyewear can be made of different materials, we recommend testing to make sure the disinfectant will not damage the equipment (e.g., 'fogging' the lenses) before implementing a chemical disinfection procedure.
Face shields-These are made of thin plastic and would likely be damaged if treated at temperatures ≥60 • C. The best approach may be to clean face shields using the same procedures as for eyewear; however, face shields usually have a foam-like material or thicker plastic band on the area that is in direct contact with the face, which may be difficult to thoroughly clean with chemical disinfectants. There is a lack of information on the use of UVGI for face shields, and they may be constructed from a wide variety of transparent plastics with different sensitivities to UVC effects. It is therefore unclear whether UVGI can be used once or repeatedly without discoloration or 'fogging' due to UV damage, and testing would be recommended.

Proposed Disinfection and Reuse Protocol
Based on the available evidence, a possible disinfection and reuse protocol is proposed as outlined in Figure 1. Following the use of new PPE, at point of doffing PPE, the wearer is to remove and inspect items, looking for any damage or soiling (e.g., bloodstains or presence of organic material). If the PPE is damaged or visibly contaminated, this is to be placed in a bin for biohazard waste. If not damaged or contaminated, PPE is to go into a separate clearly marked bin for reuse. This PPE is to be bagged and transported in a bin to the storage area using locally approved standard operating procedures.
Key steps in the proposed disinfection and reuse cycle include (Figure 1): (a) Inspection and sorting-careful inspection of PPE (including straps); any soiled and damaged PPE to be discarded, intact PPE to be stored.
(c) Re-inspection and sorting-after disinfection, careful re-inspection of PPE (including straps of FFRs) must take place; any PPE with any sign of damage must be discarded; intact PPE to be Following the use of new PPE, at point of doffing PPE, the wearer is to remove and inspect items, looking for any damage or soiling (e.g., bloodstains or presence of organic material). If the PPE is damaged or visibly contaminated, this is to be placed in a bin for biohazard waste. If not damaged or contaminated, PPE is to go into a separate clearly marked bin for reuse. This PPE is to be bagged and transported in a bin to the storage area using locally approved standard operating procedures.
Key steps in the proposed disinfection and reuse cycle include (Figure 1): (a) Inspection and sorting-careful inspection of PPE (including straps); any soiled and damaged PPE to be discarded, intact PPE to be stored. (b) Treatment-UVGI, heat, or chemical disinfection, as appropriate. (c) Re-inspection and sorting-after disinfection, careful re-inspection of PPE (including straps of FFRs) must take place; any PPE with any sign of damage must be discarded; intact PPE to be packaged for reuse, after being appropriately marked as PPE derived from disinfection, including the number of the disinfection cycle. (d) Fit checking-frontline staff to ensure that FFR passes fit check prior to use, and any disinfected PPE fit properly. At any sign of suboptimal fit, disinfected PPE to be immediately discarded.
It is likely that FFRs should be discarded after the fifth reuse, although further research is required to determine the number of cycles possible. An exception to this rule would be under extreme circumstances, where the alternative to further reuse of suboptimal PPE would be not wearing any protection at all.
This protocol provides recommendations for a possible pragmatic disinfection process for most PPE, that could be rapidly implemented, based on best available evidence.

Reuse of FFRs
In an ideal world, the reuse of FFRs is not encouraged if at all possible, as complete disinfection cannot be guaranteed for all FFRs under all circumstances. As highlighted by the US Centers for Disease Control and Protection (CDC), it is not possible to determine a maximum possible generic number of safe reuses for FFRs [104]. Nonetheless, the CDC recommend that in the absence of the manufacturer's guidance, FFRs should not be reused more than five times [104], as suggested by two previous studies [89,105] based on the observed reduction in FFR fit.
Importantly, at least one small study showed that, unsurprisingly, the reuse of N95 FFRs was associated with higher fit failure rate [106]. Thus, it is fundamental that the combination of disinfection and reuse are properly investigated, as the number of paired cycles that FFRs can be subjected to would most likely be lower than the number of disinfection cycles alone.

Extended Use of FFRs
According to Fisher and Shaffer 2014 [89], extended use would be preferable over limited reuse due to a lower risk of contamination with lesser contact with FFR surface. However, extended used leads to an increase in non-adherent behaviors (e.g., adjusting or touching the FFR) over time [107], increasing the risk of self-contamination. In a recent study from China during the COVID-19 response, 97% of 542 frontline health care workers had some form of skin damage, which was greater with longer wear of FFRs [108]. An accompanying editorial highlighted that this increases the likelihood of non-adherent FFR-wearing behavior, and consequently an increased risk of viral transmission [109]. Notably, cases of contact dermatitis in health care workers are relatively common [108,110], including those resulting from the use of FFRs [108,[111][112][113][114][115][116]. Thus, it is almost inevitable that the prevalence of dermatological conditions would increase with extended use of PPE. As prolonged skin breakdown increases health care workers susceptibility to infection and improper PPE use, access to virtual dermatology clinics for health care workers is strongly recommended to manage and treat skin breakdown in health professionals wearing PPE for extended periods. Further, it is also important to attempt to mitigate other adverse effects associated with extended use of FFRs, such as headaches [117].

Ultraviolet Light Toxicity
While UVC is perfectly safe when the equipment in question is appropriately designed and handled, it is worth stressing that accidental exposure to UVC is harmful to humans [118]. There has been at least one case in the current pandemic scenario of photokeratitis and epidermal phototoxicity caused by the improper use of a UVC disinfection apparatus in the home environment [119]. Therefore, we advise against the use of homemade devices for PPE disinfection using UVGI.

Other Pathogens
While different regimens of UVGI and heat treatment are effective against a large number of human pathogens, there is a high degree of variability among the susceptibility to temperature and UVC among microorganisms [59,120]. Thus, while the disinfection procedures reviewed here have focused on SARS-CoV-1 and SARS-CoV-2, they would not necessarily be efficacious against all other pathogens, in particular spore-forming bacteria. However, we contend they would likely be effective against many important human pathogens, and Xiang et al. for example [96] showed that dry heat treatment of FFRs at 60 • C for at least 60 min inactivated H1N1 virus, one fungus (Candida albicans), and six bacterial species (including Escherichia coli, Staphylococcus aureus, and Pseudomonas aeruginosa). The literature on UVGI is extensive, and the applied doses proposed here would be effective against a large number of human pathogens [59,98], but covering these would be outside the scope of this review. Nonetheless, the potential risk posed by other pathogens is another reason to make sure that visibly soiled PPE is not reused, as disinfection would be more difficult to achieve.

Conclusions
Given the shortage of equipment for frontline staff worldwide, the authors believe that there is sufficient evidence to support the disinfection and potential reuse of FFRs and other PPE in the current pandemic scenario when necessary. The authors have applied a would-I-wear-it (WIWI) test to the process for developing protocol recommendations. Further, based on the literature that was examined, the proposed methodology would likely achieve disinfection against most other important pathogenic organisms.
Proper disinfection and reuse of PPE would not only address the problem of short-term supply in the frontline during the pandemic, but also likely lead to considerable cost savings in the long term. Further, it would also improve the environmental footprint of a given health care facility, potentially allowing for consideration of long-term reuse of PPE. According to estimates from US hospitals for example, 5.17 tons of waste are generated per staffed bed every year [121], and in the current COVID-19 pandemic, increases of as much as 280 tons/day of extra medical waste have been reported in Southeast Asia [122]. Ultimately, it is the right of every health care worker responding to the current pandemic to have PPE available not only for their protection, but also to reduce the spread of COVID-19 [123].