A method to determine the available UV‐C dose for the decontamination of filtering facepiece respirators

Introduction

The supply of National Institute for Occupational Safety and Health (NIOSH)‐certified N95 filtering facepiece respirators (FFRs) may become limited during an influenza pandemic [Institute of Medicine (U.S.) Committee on the Development of Reusable Facemasks for Use During an Influenza Pandemic et al. (2006)]. Current guidance states that these disposable devices should be discarded after each use (donning and doffing) to prevent self‐inoculation with infectious material on the surface of the respirator. Extending the lifetime of FFRs for multiple uses (e.g. multiple donnings) may help to alleviate the supply demand (Viscusi et al. 2007, 2009a,b; Roberge 2008). One option that may permit FFR reuse is the decontamination or removal of the infectious material from the FFR through one or more physical or chemical treatments. For this option to be practical, the decontamination treatment must maintain FFR fit and filtration performance and not leave hazardous residues. Other desired attributes for a decontamination method for FFR reuse would be low cost, high throughput and ease of use (Viscusi et al. 2009b). Today, no validated decontamination methods for FFRs exist (OSHA 2008; HPA 2009), but research is underway to determine whether this is feasible (Viscusi et al. 2007, 2009b; Fisher et al. 2009; Vo et al. 2009).

Ultraviolet germicidal irradiation (UVGI) using ultraviolet‐C (UV‐C, 254 nm) has been suggested as a viable option for this application (Viscusi et al. 2007). Most recently, nine FFR models were evaluated for changes in physical appearance, odour and laboratory performance (filter aerosol penetration and filter airflow resistance) following simulated decontamination using five different methods, including UV‐C (Viscusi et al. 2009b). In that study, UV‐C treatment did not affect the filter aerosol penetration, filter airflow resistance or physical appearance of the FFRs. In another study, (Vo et al. 2009) demonstrated that a 4·32 J cm⁻² (43 200 J m⁻²) dose of UV‐C achieved approximately 3 log reduction (LR) of MS2 virus applied as droplets to an FFR, while higher UV‐C doses (>7·20 or 72 000 J m⁻²) yielded no detectable MS2 virus. However, concerns remain about the ability of UV‐C to inactivate infectious particles within the fibrous web of layers of an FFR.

The effectiveness of UV‐C as a decontamination method is affected by several parameters, including the topography of the contaminated surface and the location of the micro‐organisms within the substrate. The use of UV‐C for surfaces is mainly for hard, nonporous substrates (Andersen et al. 2006; Rastogi et al. 2007). For these substrates, the microbial contamination is deposited on the surface exterior, thus UV‐C decontamination of viruses (Tseng and Li 2007), spores (Gardner and Shama 1998) and other microbes (Gorsuch et al. 1998) has been shown to be effective. Irregular and porous surfaces are considered problematic because of the lack of UV‐C penetration. UV‐C does not readily penetrate into solid surfaces as the light is absorbed or reflected by the substrate material. Decontamination efficacy decreases if UV‐C cannot effectively penetrate shielded areas (Gardner and Shama 2000) (Tseng and Li 2007).

UV‐C transmittance is possible through the small gaps and fibres of porous surfaces such as textiles. In fact, UV‐C transmission through clothing and textiles (both woven and nonwoven) has been previously described (Kerr et al. 2000; Hoffmann et al. 2001). Furthermore, a NIOSH health hazard evaluation report of surgical suite personnel found that UV‐C penetrated some surgical masks and gowns, suggesting that UV‐C should also penetrate FFRs (Sylvain and Tapp 2009).

Research to evaluate the penetration and decontamination efficacy of UV‐C applied to FFRs is lacking. NIOSH‐certified N95 FFRs are manufactured using a variety of materials in assorted shapes and colours, although they all share some common attributes. For example, the nine FFR models studied previously (Viscusi et al. 2009b) were comprised of multiple layers of woven and nonwoven materials. Generally, the inner and outer layers provide form, stability, comfort, water resistance and filtration of large particles. The internal filtering medium (IFM) is usually comprised of spun‐bonded polypropylene and provides the bulk of the filtration efficiency (Barrett and Rousseau 1998). The aim of this study was to develop a method to assess model‐specific parameters for UV‐C decontamination of FFRs using measured UV‐C transmittance values. UV‐C transmittance through the layers of several FFR models was quantified and used to calculate FFR model‐specific UV‐C doses. The methodology used to generate this data will be useful to respirator manufacturers, infection control experts and other researchers to optimize FFR design and UV‐C parameters (irradiance, time, etc.) for possible application to FFR reuse if authorized during an FFR supply shortages for emergency situations.

Materials and methods

Calculation of α, the available UV‐C for decontamination

The measured transmittance data were used to determine the effective UV‐C irradiance for each layer of each model. Simply, the measured transmitted irradiance (i.e. the UV‐C that gets through a given layer (i), or , becomes the incoming irradiance for the next layer (). The surface irradiance (I_S) is the () for the most exterior layers. The factor α, which is the intercepted or nontransmitted UV‐C fraction, can be calculated for each layer (relative to its orientation in the FFR) in either direction using the following equation:

1

Summing the α‐values of the filtering layers (α_i) of the IFM made from both directions (n = number of filtering layers × 2) results in α_IFM, which is the available UV‐C reaching the IFM for each model as a percentage of the I_S.

2

This factor represents the percentage of UV‐C available for decontamination and is similar to the exposure factor calculated by Gardner and Shama 2000. The α_IFM values are related to the UV‐C transmittance of the exterior layers. Higher transmittance values of exterior layers can lead to higher α_IFM values, while exterior layers that shield the UV‐C from reaching the internal layers (i.e. lower I_T) will lead to lower α_IFM values.

UV‐C decontamination experiment

The α_IFM was used to categorize FFRs as most shielded (≤5%), moderately shielded (>5% but ≤30%) and least shielded (>30%). UV‐C decontamination of MS2 was determined for FFR coupons excised from two FFR models from each shielded category. Experimental coupons (n = 4) and control coupons (n = 2) were loaded with approximately 10⁷ plaque‐forming units of MS2 as previously described (Fisher et al. 2009). For models C and F, the MS2‐contaminated coupons were exposed to UV‐C (25 ± 1·0 W m⁻²) for combined bidirectional treatment times of 1, 2, 4 and 10 min (0·5, 1, 2, 5 min per side). Only the 10‐min exposure was performed on model A because of a low α_IFM. A supplemental experiment for Model A was performed using a bidirectional treatment time of 5 h.

Specific UV‐C doses (J m⁻²) to the IFM (D_IFM) for each model can be calculated as follows:

3

where α_IFM is given by eqn (2), t is the exposure time (s), and I_s is the surface irradiance (W m⁻²) from eqn (1). FFR models B, D and E were exposed to UV‐C (25 ± 1·0 W m⁻²) in a bidirectional manner for the model‐dependent treatment times corresponding to D_IFMs of 300, 1000 and 3000 J m⁻². To account for any temperature effects from long exposure times, contaminated control coupons were covered with a plastic cap and exposed alongside experimental coupons under the UV‐C source.

A dose–response curve was constructed for the MS2 decontamination by plotting the average LR per coupon as a function of the D_IFM for each model for each time‐point.

Results

Table 1 identifies the filtration layers (bold values) of each FFR model and lists the bidirectional available UV‐C (α‐values) for all layers. The exterior layers received the majority of the available UV‐C with α‐values measuring from 23 to 50%. The α‐values for the individual filtering layers ranged from 0·05 to 22%. Four models, A, B, C and D, demonstrated α_IFM values of <10%, while the models E and F had values of 30 and 31%, respectively. The α_IFM values permitted the division of FFR models into three UV‐C shielding categories as shown in the bottom row of Table 1.

The FEs of the individual layers of each FFR model using the TSI 8130 AFT confirmed the visual determination of the layers which comprise the IFM. The filtration layers, marked as bold values in Table 2, had efficiencies >87% with the majority performing at 94% or better. The filtration performance for layers not determined to be filtering in function were <31%. Likewise, Table 2 shows the differential airflow resistances of the layers of the IFM (ranging from 2·5 to 7·6 mm H₂O) exceeded the values for all other layers for all FFRs (ranging from 0·1 to 0·7 mm H₂O).

Table 3 lists the D_IFMs and measured LRs of Models A, C and F for the bidirectional treatment times of 1, 2, 4 and 10 min. The supplied dose at each treatment time was the same for each FFR model, but the D_IFM varied because of the model‐specific α_IFM values. LRs within a given treatment time differed among the FFR models. A 10‐min treatment (the only treatment condition with data for all three models) produced LRs of 0·1 (model A), 2·9 (model C), and >4·8 (model F). Across models A, C and F and within models C and F, LR increased with increasing D_IFMs.

The calculated treatment times and LRs for models B, D and E are shown in Table 4. The treatment times required to reach the targeted D_IFMs of 300, 1000 and 3000 J m⁻² differed for each model. LRs were statistically different among the models for D_IFMs of 300 J m⁻² (P = 0·006) and 3000 J m⁻² (P < 0·001). The results for the 1000 J m⁻² treatment among the models were statistically similar (P = 0·79). LRs increased with increased D_IFM for models B (P < 0·001) and D (P < 0·001). The LRs reported for model E did not differ significantly among all three doses (P = 0·18).

Table 5 contains the calculated exposure times required to achieve a D_IFM of 1000 J m⁻² for each model‐specific α_IFM with a UV‐C irradiance of 25 W m⁻², which range from 266 to 2 min. Table 5 also depicts the LRs associated with tested D_IFMs of 1000 ± 125 J m⁻². LRs for the approximate 1000 J m⁻² D_IFM ranged from 2·86 to 3·59, with the lowest LR value (2·86) corresponding to the lowest D_IFM (917 J m⁻²).

Discussion

The information presented in this study provides an effective method to calculate UV‐C doses for the decontamination of FFRs. UV‐C decontamination is based on supplying an adequate dose to the contaminated area. UV‐C dose required for decontamination, which is microbe specific, is a function of irradiance and time. The UV‐C irradiance decreases with distance, especially when travelling through a substrate (e.g. FFR), where UV absorption and reflection are factors. Irradiance is also affected by microbial concentrations and the level of protective residues; however, consideration of these factors is the subject of another manuscript. Determining the UV‐C transmittance through the FFR material provides for a more accurate estimate of the irradiance supplied to the microbial contaminant. With a more accurate assessment of the irradiance, treatment times can be adjusted to achieve the targeted dose for the specific micro‐organism.

UV‐C radiation applied to FFRs will be transmitted, reflected or absorbed by the fibres of the multilayered substrate similar to other textiles (Hoffmann et al. 2001; Duleba‐Majek 2009). The transmittance of UV‐C through the layers of the FFR will largely occur through the gaps and pores between the fibres of the material. The materials in the inner and outer layers are more porous than the IFM as demonstrated by a comparative decrease in airflow resistance during the aerosol penetration tests (Table 2). The porosity of the inner and outer layers allows UV‐C to reach the IFM. The reflection and absorption of UV‐C will also be dependent upon other textile properties (e.g. weight and thickness) and the chemical composition of the fibres and/or other materials present such as dyes and delustrants. For a given layer, the reflected and absorbed UV‐C, which is the collective radiation that has impacted the surface, is assumed to be available for decontamination. The UV‐C that is transmitted through the pores or gaps of a layer becomes the potential available UV‐C for the next stratum.

The irradiance values used in this study for calculating UV‐C doses for FFR decontamination are based on the effective irradiation for the IFM. Being in the interior of the FFR, the IFM is the most challenging layer(s) to expose to UV‐C irradiation. The IFM, by design, provides the majority of the filtration performance to FFRs and should be the target of the UV‐C dose. In some instances (e.g. cough or sneeze that contain large wet droplets), the layers exterior to the IFM may capture the majority of the particles. However, even a fraction of the microbial contamination deposited on the IFM would be challenging to decontaminate. Thus, a specific dose calculated to decontaminate the IFM will provide an excess amount of UV‐C to the inner or outer layers. The excess UV‐C to the inner and outer layers provides a ‘safety factor’ to ensure a higher level of decontamination for the FFR exterior, which may reduce the hazard caused by contact transmission of infectious organisms.

The effectiveness of calculating model‐specific D_IFMs is supported by the goodness of fit (R² = 0·87, RMSE = 0·39) of the biphasic curve for all the decontamination experiments performed (Fig. 2). The biphasic analysis of the LRs (not shown) vs the supplied dose is more scattered because of the exclusion of model‐specific α_IFM values (R² = 0·36, RMSE = 0·88). Biphasic curves have been used to model thermal, nonthermal, pulsed field and UV‐C inactivation and occur as a result of a resistant subpopulation of microbes or other factors, such as, microbial clumping, microbial protection via particulate matter or localization on the substrate (Cerf 1977; Kowalski 2000; Geeraerd et al. 2005).

An examination of the data for the 10‐min UV‐C exposure of FFR representatives of the most shielded, moderately shielded and least shielded categories reveals the importance of calculating the D_IFM (Table 3). A 10‐min exposure to a UV‐C surface irradiance of 25 W m⁻² equates to a supplied dose of 15 000 J m⁻² for each tested model. The LRs of 0·1, 2·9 and >4·8 for models A, C and F, respectively, would have been inconsistent if an equal dose was assumed. However, D_IFMs of 38, 917 and 4707 J m⁻² for models A, C and F, respectively, are consistent with the LR values. Variation in the decontamination efficiency among the FFR models is expected considering the differences in D_IFMs.

The results of the targeted dose treatments (Table 4) further support the relationship of D_IFM and MS2 decontamination, although unanticipated variation in decontamination efficacy across FFR models was evident for the 300 and 3000 J m⁻² D_IFMs. This variation may provide insight into factors important to UV‐C decontamination of FFRs. For the D_IFM of 300 J m⁻², some variability would be expected because of the steep slope of the initial phase of the biphasic curve (Fig. 2). In this initial phase, the shielding effect, or protection of the virus from UV‐C, seems to be limited, which is in contrast to the results of the 3000 J m⁻² treatment. The statistical variation for this treatment may hint at an FFR model‐specific decontamination limit, which results from shielding or blocking of UV‐C by the composite layers of the FFR. This is similar to the shielding effect described in Gardner and Shama (2000). Across the models, this effect can be observed as the tailing of the biphasic curve (Fig. 2), although other factors such as virus clumping and UV‐C resistance of a subpopulation of virus may produce similar results (Kowalski 2000). The results for the UV‐C decontamination of model E provide further support for a model‐specific decontamination limit. For this FFR model, UV‐C efficacy approaching the decontamination limit was achieved with the lowest tested D_IFM of 300 J m⁻² and remained statistically the same for all three D_IFMs tested (Table 4).

UV‐C treatments of an approximate D_IFM of 1000 J m⁻² produced similar LRs for all models tested (Table 5). The consistent MS2 LR values demonstrate the importance of α_IFM values in determining the duration of UV‐C exposure times. Calculated exposure times to achieve a model‐specific D_IFM of 1000 J m⁻² ranged from 266 min for Model A to 2 min for Model F, representing the most shielded to least shielded categories. Exposure times of <8 min for the targeted D_IFM, given an irradiance of 25 W m⁻², are possible with α_IFM values of 8·5 and greater (Models D–F). Model B, of the most shielded category (α = 2·18%), demonstrated a 3·2 LR with a treatment time of 32 min. Even with low α_IFM values, significant LRs are possible with longer exposure times as demonstrated in this study or increased UV‐C irradiance, which would also provide an increase in dose. The demonstrated decontamination of FFRs with low α_IFM values is promising for real world application.

The determination of the UV‐C transmittance and α‐values for the layers of an FFR is simple when using a radiometer. This process can be further simplified because of the ease by which filtering media is visually distinguishable from the other layers of an FFR (although filtration data would be preferable). The material used in the IFM appears to be less porous than the surrounding layers and of similar composition across model types. For all models examined, the filtering medium was located in the interior of the mask. By combining all the media that comprise the filtering layers into one stratum (the IFM) and compiling separate strata for the media to the exterior of the IFM in both directions, the number of layers can be minimized along with the number of measurements required to ascertain the transmittance data. Simple calculation [eqns (1) and (2)] using the transmittance data produces α_IFM values, which is necessary to generate FFR model‐specific D_IFMs. Reducing the number of layers by separating the IFM from the other composite strata of the FFR also minimizes the complexity of deciphering what is or is not a separate and distinctive layer.

Model‐specific α_IFM values will determine the necessary UV‐C exposure time, given a targeted dose and the UV‐C source. The potential to decontaminate specific FFR models to a targeted LR value can be determined by evaluating the exposure times required to achieve desired doses. A minimum α_IFM could be adopted during a reuse emergency, thereby defining FFR models that are more likely to lead to successful UV‐C decontamination. The α_IFM values may also help in the selection of an adequate UV‐C source as an increase in irradiance would decrease treatment times given the targeted D_IFM.

Producing an IFM model‐specific bidirectional (from both directions) D_IFM provides the highest level of decontamination potential. Microbial contamination may emanate from the environment and/or from the user and deposit on all the layers of an FFR. A bidirectional UV‐C exposure targeted for the IFM will provide a dose adequate for microbial decontamination on the IFM, while providing an excessive dose for the exterior layers. However, a bidirectional target of the IFM may not be essential. Limiting the exposure to one direction, from outer to inner, may provide an adequate dose for decontamination while limiting the complexities of a two‐sided bidirectional treatment. The method to determine α_IFM values and D_IFMs described previously for bidirectional treatments is applicable to single direction exposures as well. Depending on the model‐specific composition of the exterior layers, the α_IFM may be greater for a single direction exposure compared to a bidirectional treatment. The outer layer and IFM are more likely to contain the microbial threat, assuming that the wearer is healthy.

Acknowledgements

The authors express their sincere gratitude to Dr Debra Novak, Dr Samy Rengasamy, Dr Evanly Vo, Dr Benjamin C. Eimer, Mr Dennis Viscusi, Mr Mike Bergman and the manuscript reviewers for their suggestions and contributions.

Disclaimer

The findings and conclusions in this manuscript are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health (NIOSH). Mention of company names or products does not constitute endorsement by NIOSH.