Statistical issues with respect to workplace protection factors for respirators

A critique of Occupational Safety and Health Administration's halfmask respirator assigned protection factor

Halfmask air-purifying respirators are used by millions of workers to reduce inhaling air contaminants, both chemical (e.g., asbestos, styrene) and biological (e.g., SARS-CoV-2, Mycobacterium tuberculosis). In 2006, the federal Occupational Safety and Health Administration (OSHA) promulgated a standard that gave halfmask respirators an assigned protection factor (APF) of 10. This signified that OSHA assumes a fit-tested and trained wearer will experience a 10% maximum total inward leakage of contaminated air into the facepiece. To derive APF = 10, OSHA analyzed data from 16 workplace studies of the efficacy of halfmask respirators worn against particulate contaminants. In this commentary, I contend that, in considering the data, OSHA made several errors that overstated halfmask respirator efficacy. The errors were (i) failing to properly account for within-wearer and between-wearer variability in respirator efficacy; (ii) ignoring the effect of particle deposition in the respiratory tract; (iii) aggregating unbalanced data within and between studies, and effectively double-counting the data in some studies; and (iv) ignoring the effect that particle size exerts in penetrating respirator facepiece leak paths. The net result is that OSHA's APF = 10 can lead to excessive toxicant exposure for many workers.

Correcting Coal Miner Respirator Total Inward Leakage Values for Respiratory Tract Deposition

The great majority of workplace respirator efficacy studies have measured total inward leakage (TIL) for particulate contaminants. One of the first such studies, designated the Harris study, was conducted in the early 1970s in US underground coal mines. As in other particle-based studies, inside-the-facepiece dust sampling was continuously conducted across the inhalation and exhalation phases of the breathing cycle, although unlike in other studies, respirable dust cyclones were used in air sampling. Because exhaled air was partially depleted of dust particles due to deposition in the respiratory tract, the measured time-averaged dust concentration inside the facepiece underestimated the time-averaged dust concentration inspired into the facepiece. In turn, the reported TIL values underestimated the true TIL values experienced, which is to say, overestimated respirator efficacy. This paper describes a method to correct the Harris study's reported TIL values for respiratory tract deposition while accounting for particle size-selective sampling by the cyclone devices. Given the estimated coal mine particle size distribution outside the respirator, it is shown that the reported TIL values should be increased by the multiplicative factor 1.69. This paper also discusses the assigned protection factor (APF) of five for the quartermask respirator class and shows that 4/5 quartermasks in the Harris study did not meet the criterion for complying with this APF value when using the corrected TIL values.

Predicting future protection of respirator users: Statistical approaches and practical implications

The purpose of this article is to describe a statistical approach for predicting a respirator user's fit factor in the future based upon results from initial tests. A statistical prediction model was developed based upon joint distribution of multiple fit factor measurements over time obtained from linear mixed effect models. The model accounts for within-subject correlation as well as short-term (within one day) and longer-term variability. As an example of applying this approach, model parameters were estimated from a research study in which volunteers were trained by three different modalities to use one of two types of respirators. They underwent two quantitative fit tests at the initial session and two on the same day approximately six months later. The fitted models demonstrated correlation and gave the estimated distribution of future fit test results conditional on past results for an individual worker. This approach can be applied to establishing a criterion value for passing an initial fit test to provide reasonable likelihood that a worker will be adequately protected in the future; and to optimizing the repeat fit factor test intervals individually for each user for cost-effective testing.

Criteria for the collection of useful respirator performance data in the workplace

Workplace protection factors (WPFs) are intended to measure the ability of a respiratory protective device (RPD) to reduce contaminant exposure when used in the context of an effective respiratory protection program. In 1992, members of the American Industrial Hygiene Association Respiratory Protection Committee (RPC) published a review of important issues and considerations for measuring respirator performance in the workplace. The RPC recognized that respirator testing in workplaces can have a variety of objectives and endpoints, and that not all workplace measurements are WPFs. That paper addressed concerns in the general categories of 1) study objectives; 2) site selection; 3) subject selection and preparation; 4) sampling and analytical methods; and 5) data analysis. No specific protocol for measuring WPFs was recommended by the RPC, and attempts to reach a U.S. consensus on a WPF protocol since 1992 have not succeeded. Numerous studies have implemented the principles for WPF measurement described in the RPC paper. Modifications to the original recommendations have been made to reflect the current state of the art. This article describes what has been learned in recent years in each of the five categories identified in the 1992 discussion. Because of the wide variety of workplaces and work activities, contaminants and respiratory protective devices, a strict protocol is not appropriate for collecting WPF data. Rather, the minimum requirements for the collection and presentation of meaningful respirator performance data in the workplace are described. Understanding of these principles will permit useful RPD performance data to be generated.

Determining sample size and a passing criterion for respirator fit-test panels

Few studies have proposed methods for sample size determination and specification of passing criterion (e.g., number needed to pass from a given size panel) for respirator fit-tests. One approach is to account for between- and within- subject variability, and thus take full advantage of the multiple donning measurements within subject, using a random effects model. The corresponding sample size calculation, however, may be difficult to implement in practice, as it depends on the model-specific and test panel-specific variance estimates, and thus does not yield a single sample size or specific cutoff for number needed to pass. A simple binomial approach is therefore proposed to simultaneously determine both the required sample size and the optimal cutoff for the number of subjects needed to achieve a passing result. The method essentially conducts a global search of the type I and type II errors under different null and alternative hypotheses, across the range of possible sample sizes, to find the lowest sample size which yields at least one cutoff satisfying, or approximately satisfying all pre-determined limits for the different error rates. Benchmark testing of 98 respirators (conducted by the National Institute for Occupational Safety and Health) is used to illustrate the binomial approach and show how sample size estimates from the random effects model can vary substantially depending on estimated variance components. For the binomial approach, probability calculations show that a sample size of 35 to 40 yields acceptable error rates under different null and alternative hypotheses. For the random effects model, the required sample sizes are generally smaller, but can vary substantially based on the estimate variance components. Overall, despite some limitations, the binomial approach represents a highly practical approach with reasonable statistical properties.

Elastomeric, half-facepiece, air-purifying respirator performance in a lead battery plant

This workplace protection factor (WPF) study of a half facepiece air-purifying respirator with P100 filters was done in a lead battery manufacturing plant. Paired air samples for lead were collected inside and outside respirators worn by workers who were properly trained and quantitatively fit tested. Of the 45 valid sample sets, only four had detectable lead on the inside sample. WPFs were calculated for these sample pairs by dividing the outside sample lead concentration (C(o)) by the inside concentration (C(i)). For the remaining 41 sample pairs, the detection limit for lead was used to calculate a maximum estimated C(i) concentration. The C(o) for each of these sample pairs was divided by the C(i) estimate to obtain a minimum WPF value. All the WPFs were rounded down to two significant figures, resulting in values ranging from 12 to > 2500. A rank and percentile procedure resulted in a 50th percentile WPF of 270 and a lower 5th percentile estimate > 50. These WPFs exceed the assigned protection factor of 10 for half facepieces published by the Occupational Safety and Health Administration. This study's results support the APF of 10 and indicate the respirator provided appropriate protection as it was used in this study. The comparability of the two analytical methods commonly used together in WPF studies was also evaluated. The samples collected outside the respirators were analyzed for lead by proton-induced X-ray emission analysis (PIXE) followed by inductively coupled plasma spectrometry (ICP). While the two methods were highly correlated (r(2) = 0.965), the mean PIXE lead mass was approximately 45% higher than the mean ICP value. This systematic bias was explained by the assumptions used to interpret the PIXE analytical results. When WPF studies use ICP and PIXE for C(o) and C(i) samples, respectively, the calculated WPF values are conservative estimates of respirator performance.