Preamble - Respiratory Protection
V. Significance of Risk
Respirators are used by American workers as a means of protection against a multitude of respiratory hazards that include chemical, biological, and radiological agents. Situations in which respirators are relied upon to provide protection from these hazards include those that involve immediately life-threatening situations as well as routine operations where engineering controls and work practices are not able to provide sufficient protection from these hazards. In these situations, respirators must "seal off" and isolate the worker's respiratory system from the contaminated environment. The risk that a worker will experience an adverse health outcome when relying on respiratory protection is a function of the toxicity or hazardous nature of the air contaminants present, the concentrations of the contaminants in the air, the duration of exposure, and the degree of isolation provided by the respirator. When respirators fail or do not provide the degree of protection expected by the user, the user is placed at an increased risk of any adverse health effects that are associated with exposure to the respiratory hazards present. Therefore, it is critical that respirators perform as they are designed to do to ensure that users are not at an increased risk of experiencing adverse effects caused by exposure to respiratory hazards.
OSHA has discussed the nature of adverse health effects caused by exposure to airborne chemical hazards many times in previous rulemaking efforts (see, for example, Appendix A of the Hazard Communication standard, 29 CFR 1910.1200 and the preambles to any of OSHA's single substance standards codified in 29 CFR 1910.1001 to 1910.1052). In all instances where OSHA has promulgated new or revised PELs for chemical air contaminants, OSHA has determined that the health effects associated with exposure to the contaminants represent material impairment of health because the effects are life-threatening, cause permanent damage, or significantly impair the worker's ability to perform his or her job in a safe manner. As discussed in Section VI of this preamble, OSHA expects that thousands of illnesses and hundreds of fatalities that are presently being caused by exposure to hazardous substances will be avoided annually among respirator wearers as a result of improvements and clarifications made to the earlier standard by this final rule.
Evidence on current workplace exposure levels confirms that respirators are needed in many work situations to protect workers against serious work-related illness. To illustrate, OSHA identified several substances that represent a range of adverse effects and for which OSHA's Integrated Management Information System (IMIS) database has documented workplace exposures that exceed the current PELs for these substances. The effects represented by this subset of the IMIS and the associated substances for which there are documented overexposures include:
-- Sudden death/asphyxiation -- carbon monoxide, carbon dioxide;
-- Loss of lung function -- wood dust, welding fume, manganese fume, copper fume, cobalt metal fume, silica;
-- Central nervous system disturbances -- carbon monoxide, trichloroethylene;
-- Cancer -- chromic acid, wood dust, silica; and
-- Cardiovascular effects -- carbon monoxide.
When respirators are used during operations where exposures exceed OSHA's PEL, OSHA believes that there is little or no margin that would protect the worker in the event that the respirator does not perform as well as designed or expected. For all of the substances for which OSHA has promulgated a comprehensive health standard (i.e., Arsenic, 29 CFR 1910.1018; Asbestos, 29 CFR 1910.1001; Benzene, 29 CFR 1910.1028; Lead, 29 CFR 1910.1025; Ethylene Oxide, 29 CFR 1910.1047), OSHA has determined that exposure above the PEL is associated with a significant risk of material impairment of health, and believes as a matter of policy that exposures below the PEL may be associated with risk levels that are significant. That is, there is no exposure level near or somewhat above the PEL that can be considered to be at a low or insignificant risk level. Therefore, where workers perform jobs that result in exposures above the PEL for any of these substances, use of properly functioning respirators is essential to ensure that workers are not placed at significant risk of material impairment of health.
Throughout this preamble, OSHA has demonstrated that adequate fit testing, proper respirator selection, worker training, and thorough inspection and maintenance are essential elements of a respirator program. Without these requirements, OSHA believes that there is a greater chance that a respirator user will inhale potentially dangerous air contaminants, either by improper selection of equipment, excessive respirator leakage, improper use of the respirator, or any combination of these. This section presents an analysis conducted by OSHA to evaluate the improved protection to workers who use respiratory protection equipment by the type of effective respirator program required by the final rule.
In the context of a respiratory protection program, the health risk presented to workers can be represented as the risk that a respirator will fail to provide some minimum expected level of protection, which increases the possibility that the user of the respirator will be overexposed to a harmful air contaminant. This presumes that respirators will be selected and used in work settings where exposure to ambient concentrations of air contaminants poses an unacceptable health risk, and, if the respirator performs as expected, the wearer will be protected from that risk. For example, an employer who provides a half-mask, chemical cartridge respirator for employee use might typically assume that the respirator will filter out 90 percent of the contaminant and base his or her choice of respirator on that assumption. If the respirator performs less effectively than expected, the employer's expectation that the respirator will provide effective protection will not be fulfilled.
This concept of risk differs from that used by OSHA in its substance-specific health standards, in which the Agency typically defines risk as the probability that a worker will acquire a specific work-related illness. Quantifying that kind of risk requires the analysis of data that relates the magnitude or intensity of exposure to the incidence or prevalence of adverse effects seen among exposed populations or experimental animals. In contrast, the kinds of hazardous situations covered by the final respiratory protection standard are varied in terms of the nature of the hazard present (i.e., acute, chronic, or both), the frequency and magnitude of exposure, and the types of illnesses associated with exposure to those hazards. As a consequence, the health risks addressed by the final rule cannot be described in terms of an illness-specific risk, but instead relate to the more general probability that a respirator will provide insufficient protection causing the wearer to be exposed to a dangerous level of one or more air contaminants.
Certain studies, referred to as "workplace protection factor" (WPF) studies, have attempted to measure the effectiveness of respirators under actual conditions of use in the workplace. The WPF is a measure of the reduction in exposure achieved by using respiratory protection and is represented by an estimate of the ratio of the concentration of a contaminant found in the workplace air to the concentration found inside the respirator facepiece while the respirator is being worn. As the degree of protection afforded by the respirator increases, the WPF increases. Alternatively, the degree of protection provided by a respirator can be expressed as a penetration value, which is the reciprocal of the WPF and reflects the ratio of the concentration of contaminant inside the facepiece to the concentration outside. For example, a WPF of 50 equates to a penetration value of 0.02 and means that the concentration inside the respirator facepiece is one-fiftieth of the ambient level.
Because WPF studies are designed to evaluate the field effectiveness of respiratory protection equipment, study protocols usually have been designed to minimize factors that can reduce respirator performance. Such factors include selecting the wrong type of respirator for the working conditions under which the study is being conducted, use of poorly fitting respirator facepieces (i.e., testing of respirator fit is routinely done in well-conducted WPF studies), inadequate training of wearers in proper respirator adjustment and use, or excessive leakage caused by malfunctioning or dirty respirator parts. Typically, WPF study protocols include procedures for properly selecting respirators and ensuring that they are in good working order, assigning respirators to workers on the basis of valid qualitative or quantitative fit tests, training wearers on how to adjust strap tension properly and use the respirator, and ensuring that neither facial hair nor other personal protective equipment is likely to interfere with respirator fit. In addition, workers included in WPF studies are usually monitored throughout the period that respirators are worn to verify that the equipment is being properly used. All of these conditions reflect the principal elements of a strong respirator program in which respirator performance is optimized; therefore, the results from a good WPF study can mirror the results obtained by an employer who implements a well-run respiratory protection program.
To quantitatively evaluate the impact of implementing a good respirator program on respirator performance, OSHA identified several WPF studies that were conducted using methods that reflect a comprehensive program, and compared these results to other workplace studies that did not employ all of the elements of a good program. Quantitative approaches are used to develop (1) aggregate estimates of respirator effectiveness in both the presence and absence of a good respiratory protection program, and (2) estimates of the frequency with which workers are likely to achieve inadequate protection while using a respirator, given the presence or absence of a good underlying program. All of the studies used in this analysis pertain to the effectiveness of half-mask, negative-pressure respirators, and all are contained in OSHA's rulemaking docket (H-049).
Many of the well monitored WPF studies conducted were reviewed by Nelson et al. in 1995 (Ex. 64-514); these authors selected data from seven such studies to evaluate the overall field effectiveness of half-mask, negative-pressure respirators. Each of the studies described by Nelson et al. ensured selection of properly fitted respirators either by an accepted qualitative fit test (QLFT) (i.e., isoamyl acetate or saccharin) or by a quantitative fit test (QNFT) where only respirators that provided a minimum protection factor to the wearer of at least 100 were selected. Each of these studies provided for worker instruction in proper respirator use, and workers were monitored during each study to ensure proper use. An additional six studies were reviewed by Nelson et al. but were rejected either because they allegedly used biased sampling methods to determine ambient and in-facepiece contaminant concentrations or because the authors believed that improper or invalidated fit test procedures were employed.
In the studies selected by Nelson et al. for analysis, workers used elastomeric or disposable respirators equipped with dust-mist, dust-mist-fume, or high-efficiency particulate (HEPA) filters, and the collection of studies represented a range of workplace exposure situations, including pigment production, metals refining, asbestos exposure during brake-repair work, welding, and spray painting. Geometric Mean (GM) WPF values from these studies ranged from 47 to 3,360, with an overall GM WPF of 290. The 5th percentile WPF from the data set was estimated to be 13, with a 95% confidence interval of 10-18. Nelson et al. concluded from the analysis of the overall data set that the assigned protection factor of 10 for half-mask, negative- pressure respirators was reasonable given that a WPF of less than 10 would not likely occur more than 5 percent of the time. In addition, Nelson et al. found no significant difference in the field performance of disposable respirators compared to elastomeric models. OSHA has not conducted a detailed comparative evaluation of WPF values obtained from disposable vs. elastomeric respirators; if, in fact, disposable respirators provide less protection than elastomeric respirators, the WPFs that can be achieved under a good respirator program will be overstated in this analysis since Nelson et al.'s compiled data reflect the use of both types of respirators.
Each of the studies reviewed by Nelson involved worker exposures to dusts. OSHA could identify only one WPF study, by Galvin et al. in 1990 (Ex. 64-22), that examined respirator effectiveness against exposure to a vapor-phase contaminant rather than a particulate. In this study, WPF measurements were taken on a group of 13 styrene workers who used halfmask, air-purifying respirators equipped with chemical cartridge filters. All employees were assigned respirators based on passing an irritant smoke fit test, and all were trained on how to properly don the respirator and conduct fit checks. In-mask and ambient styrene concentrations were measured over one-hour periods, during which employees were instructed not to readjust the facepiece. Chemical cartridges were changed with each new sampling period to ensure that there was no breakthrough. In-mask styrene concentrations were adjusted upwards by 40 percent to account for pulmonary retention, which avoided potentially overestimating the WPF. The GM WPF for the overall cohort was reported to be 79, with a geometric standard deviation (GSD) of 3.51. There was no significant difference in WPF values between those workers engaged in relatively physical operations, such as spraying, compared to those performing less physical work tasks. The GM WPF found by Galvin et al. for styrene-exposed workers lies within the range of GM WPF values reported in the studies reviewed by Nelson for worker cohorts exposed to particulate-contaminated environments.
Nelson in his 1995 report (Ex. 64-514) excluded the Galvin et al. study from his analysis because fit tests were performed using the irritant smoke protocol. As discussed in the Summary and Explanation section of this preamble, OSHA has determined that the irritant smoke qualitative fit test provides a valid, effective test of respirator facepiece fit. The procedures used by Galvin et al. to ensure adequate worker training and respirator use are consistent with the elements of a permissible respirator program, and OSHA, therefore, finds it appropriate to include this study in the set of WPF studies that are representative of effective respiratory program practices.
In contrast, OSHA has identified three studies where investigators also determined WPF values for half-mask, negative-pressure respirators, but where few steps were taken to ensure maximum respirator performance. OSHA believes that these studies illustrate the relative lack of protection afforded by respirators when certain critical elements of the respiratory protection program are missing or inadequate. The studies identified by OSHA are those by Toney and Barnhart in 1972 (Ex. 64-68), Moore and Smith in 1976 (Ex. 64-49), and Harris et al. in 1974 (Ex. 27-11).
Toney and Barnhart (Ex. 64-68) conducted a WPF study to evaluate the effectiveness of half-mask, chemical-cartridge respirators on reducing exposures of spray painters to solvent vapors and aerosols. Data were obtained from painters working at 39 different sites and included both in-mask and ambient concentrations. WPFs were found to be low; from the raw data presented in the study, OSHA calculated a GM WPF of 3.8 for solvent exposure (GSD=2.28, N=39) and a GM WPF of 11.4 for aerosol exposure (GSD=4.12, N=40). Penetration tests performed on unused respirator cartridges of the same types used in the field indicated that the poor WPFs achieved in the field tests were caused by poor respirator fit and a lack of respirator maintenance, and were not due to any inherent defect in the cartridges. The authors concluded that respirators being used by painters were not effective and cited several reasons, all pointing to the lack of a respiratory protection program at the facilities tested. For example, 28 percent of respirators used by the painters were poorly maintained. Some of the conditions found by the investigators included deteriorating rubber on the facepieces, the presence of stuck or warped valves, missing head straps, and evidence of leakage around the cartridge seal. In addition, it was apparent that some of the cartridges had not been changed for extended periods of time. Many of the facilities studied supplied non- approved respiratory protective devices (respirators were approved by the Bureau of Mines at the time of the study), and most had no formal training or maintenance program in place. The authors found that "* * * management and workers are extremely uninformed on the subject of selection, use, and care of respiratory protective devices." (Ex. 64-68, p. 93).
The second study, conducted by Moore and Smith in 1976 (Ex. 64-49), measured WPF values obtained by workers exposed to sulfur dioxide (SO2) during a furnace charging operation at a copper smelter. Three models of half-mask, chemical cartridge respirators were tested on each of nine workers; in-mask and ambient SO2 concentrations were measured during the furnace charging operation while the respirators were worn. There is no indication in the study that qualitative or quantitative fit testing was performed to verify adequate facepiece fit. A total of 81 samples were collected, 5 of which were excluded from the analysis because the subjects removed or lifted the respirator facepiece during the sampling period. Average ambient SO2 concentrations varied in the range of 53 to 61 mg/m3 (20.4 to 23.5 ppm) during the sampling period. Geometric mean WPF values reported for each of the three models of respirator were 22.1 (SD=22.6), 18.4 (SD=14.2), and 12.9 (SD = 11.0). Moore and Smith concluded that the overall protection afforded by the respirators was poor, and that between one-third and one-half of the protection factors achieved would be below 10, the accepted minimum protection factor for that type of respirator. Reasons given by the authors for the poor fits observed among the subject workers included the possibility that strap tension was not properly adjusted (the authors did not control or monitor strap tension), variation in facial hair (despite the lack of beards or wide sideburns), and normal work activities that caused head motion and deep breathing associated with heavy work.
The third study is that of Harris et al. in 1974 (Ex. 27-11), who evaluated the performance of five half-mask dust respirators among 37 miners working in 4 coal mines. In-mask and ambient dust measurements were made throughout the workshifts, during which miners intermittently used respiratory protection. Thus, this study differs from the others described above in that the ratio of in-mask to outside concentrations included periods of time where the respirator was not worn, in contrast to the typical WPF study. The ratio of in-mask to outside concentration determined during periods of intermittent respirator use, termed the "effective protection factor" (EPF), is not directly comparable to WPF values because, to the extent that workers spend time in contaminated atmospheres without respiratory protection, the WPF will tend to understate the actual protection obtained while the respirator is being worn. However, according to Poppendorf in 1995 (Ex. 54-512), it is possible to use EPF data to estimate the WPF that was likely to have been achieved during periods of respirator use if both of the following are known or can be estimated: (1) The fraction of time during which the respirator was not worn by the subject, and (2) the ratio of contaminant concentration in areas where the respirator was worn to that in areas where the respirator was not worn. Poppendorf (Ex. 54-512) described the mathematical relationship between the EPF and WPF and suggested that the likely range of average WPF values achieved by the miners during periods of respirator use was 3.6 to 5.7. This estimate of WPF is based on an observation by Harris et al. that miners wore their respirators about half of the time during the sampling periods, and an assumption by Poppendorf (Ex. 54-512) that the dust levels in the air while respirators were worn were at least 5 times higher than airborne dust levels during periods of respirator non-use. OSHA believes that the latter assumption is reasonable given that Harris et al. reported that, for the most part, miners wore their respirators only when visible airborne dust was present. Harris et al. noted that the hard hats worn by the miners interfered with proper respirator strap positioning and adjustment; OSHA believes that this factor, as well as the apparent lack of fit testing, is likely to have contributed to the low protection factors experienced by the miners.
OSHA believes that the studies described above demonstrate that improved respirator performance can be achieved under actual workplace conditions if fit testing is used to select respirators, if respirators are clean and in good working order, and if employees are properly trained and supervised in their use. This is evident when the summary statistics from aggregate protection factor data obtained from field studies on groups of employees using respirators in the absence of a strong respirator program (i.e., Moore and Smith, Toney and Barnhart, Harris et al.) are compared with those obtained from cohorts using respirators under the condition of a strong program (i.e., the studies reviewed by Nelson and the study by Galvin et al.). Summary protection factor data from these studies are presented in Table V-1 as geometric mean and mean WPF values, and the geometric standard deviation (GSD) of the distribution of WPF values. From these summary statistics, OSHA computed a weighted geometric mean WPF across cohorts exposed to particulate contaminants to compare the central tendency in protection factors achieved both with and without an adequate underlying respirator program (see footnote on Table V-1).
In general, groups of employees using respirators against particulate exposures under a strong program achieved an overall GM protection factor about 25-fold higher than groups using respirators without the elements of a strong respiratory protection program. In studies that did not implement all of these elements, mean WPF values among the particulate-exposed worker cohorts tested ranged from about 6 to 22. Mean WPF values for particulate-exposed worker cohorts included in the WPF studies where elements of a good program were implemented ranged from 72 to 2,400, with the mean WPF from one study estimated to be 11,500. The results from studies that examined respirator effectiveness against gas or vapor, also included in Table V-1, show an 8-fold difference in overall GM WPF values. With only one exception, the 95 percent confidence intervals around the GM WPF values computed from the studies reflecting inadequate program practices do not overlap with those computed from the studies reflecting strong program elements (see Table V-1); thus, the hypothesis that there are no differences in the GM WPF values between the two groups of studies is rejected. This analysis suggests that implementation of a good respiratory protection program containing the elements described by the final rule can contribute to a substantial increase in the overall performance of respirators used in actual workplace settings, as measured by the mean WPF across groups of workers.
The three WPF studies representing deficient program practices were all conducted 10 to 20 years earlier than the WPF studies reflecting good program elements. Thus, differences between the two groups of studies in working conditions, processes and exposures, or respirator equipment and technology could confound the comparison of respirator effectiveness measures. OSHA is not aware of any recent studies that have been conducted that were designed to evaluate the impact of respirator program elements on respirator effectiveness, nor are recent studies available that have attempted to measure respirator effectiveness under conditions of a poor respiratory protection program. OSHA believes that this analysis of program impacts on respirator performance is based on the best available data. However, OSHA has considered whether confounding factors related to the elements of a good respirator program may also have contributed to the differences in respirator performance reported by the two groups of WPF studies. For example, respirator fit can be adversely affected by vigorous work activity requiring head motion and deep breathing. Heavy work loads also contribute to respirator discomfort, which may cause a worker to wear a respirator too loosely. The nature of the air contaminant affects respirator performance in that different types of respirator filters have different capabilities in purifying contaminated air and gas-phase contaminants and small-particulate aerosols pass more readily through leak points than do aerosols comprised mostly of larger particles.
OSHA does not believe that any systematic differences in working conditions or respirator technology contribute substantially to the differences in respirator effectiveness found between the two groups of studies included in the analysis. For example, both groups of studies represent a range of workplace situations that involve strenuous and non-strenuous work. In the studies that do not reflect good program practices, workers were engaged in active, strenuous work (smelter operations and coal mining) as well as less active work (spray painting). Similarly, studies that reflect good program practices have also been conducted on worker cohorts engaged in both active work (metals refining) and less active work (spray painting, brake repair). Both groups of studies also involve a range of contaminants, including both gas-phase and various kinds of particulate. Some of the studies reviewed by Nelson included information on the size distribution of particulates to which workers were exposed, with the range across these studies including both respirable and non-respirable particles. Other studies included in the Nelson analysis reported that workers were exposed to both dust and fume. Therefore, the differences in WPFs found between the two groups of studies cannot be explained by differences in particulate sizes or characteristics. Both groups of studies also represent a variety of half-mask respirator designs and filters, including single-use respirators and respirators equipped with dust/ mist (i.e., non-HEPA) filters. OSHA believes it unlikely that the 14-fold difference in overall WPFs between the two groups of studies can be primarily attributed to any fundamental differences in respirator equipment or technology. Therefore, OSHA finds that the differences in WPF values obtained from the two groups of studies are more likely to reflect differences in how well the respirators fit the subject workers, the condition of the respiratory equipment used, and the extent to which the equipment was used properly, rather than any confounding caused by systematic differences in work settings, the nature of the exposures, or the age of the WPF studies.
The kinds of summary statistics presented in Table V-1 have been used by several investigators to demonstrate how poorly or how well respirators can protect workers under actual conditions of use (see, for example, Moore and Smith (Ex. 64-69), Nelson et al. (Ex. 64-514)). However, such descriptive measures can only provide information on the aggregate frequency distribution of protection factor values in a group of workers. Although it is useful to rely on summary statistics from aggregate protection factor data to make general statements about the effectiveness of respirators, such measures do not adequately convey information on the number or proportion of workers who remain at risk of overexposure to air contaminants despite the use of respiratory protection, or how frequently an individual worker might experience poor fits.
Nicas (Ex. 156) and Nicas and Spear in 1992 (Ex. 64-425) have suggested that using statistics from aggregate protection factor data does not adequately describe the true risk of overexposure to workers using respirators because the approach fails to recognize that there are two different sources of variability that account for the overall variation in protection factor values measured from a given cohort of workers. One source of variability in protection factors is the variation typically experienced by a single worker from one day to the next; this is termed within-worker variability. The second source of variability reflects the observation that different workers within a group will achieve different average protection factors over a given period of time; this is termed between-worker variability. In a peer-reviewed article, Nicas and Spear (Ex. 64-425) have described a statistical model that accounts for both sources of variability. This model has been used by OSHA to estimate the following from the protection factor studies described above to better characterize risks to workers who use respirators both in the absence of and under a strong respiratory protection program:
(1) The proportion of workers who fail to achieve a long-term average protection factor at or above some specified target level, exposing the worker to an increased risk of a chronic health hazard (i.e., a health hazard that is typically associated with long-term cumulative exposure); and
(2) The proportion of workers who achieve a protection factor below some specified target level at least 5 percent of the time that the respirator is worn, thus increasing the frequency with which a worker may be exposed above an effect concentration associated with an acute health hazard.
The Nicas and Spear model (Exs. 64-425, 156) used by OSHA in this analysis is a one-factor analysis of variance and is described briefly as follows. Let P denote a penetration value experienced by the wearer of a respirator during a randomly selected wearing time (P is defined as the reciprocal of the protection factor PF measured in the workplace, or 1/PF). For example, a P value of 0.1 for a respirator wearer reflects that a protection factor of 10 was achieved in the workplace for that individual. If one were to measure the penetration values among members of a group of workers over time and aggregate the results, the total distribution of P values can be described by the following parameters:
Where:
P = the penetration value for a worker for a particular wearing period,
µp = the arithmetic mean penetration value for the population,
B = a lognormally distributed factor that transforms µp to the arithmetic mean
penetration value for the individual worker, and
W = a lognormally distributed factor that transforms µp x B to the P value
experienced by the individual worker for a particular wearing time.
The factors W and B describe within-worker variability and between-worker variability, respectively.
Since workplace protection factor studies typically report the geometric mean and geometric standard deviation of protection factor values obtained from a cohort of respirator wearers (i.e., GM[P] and GSD[P]), the parameters described above for within-worker and between worker variability can be estimated as follows if the relationship between GSD[B] and GSD[W] are known or assumed. Let R represent the ratio of GSD[W]/GSD[B]; then GSD[B] can be estimated from GSD[P] and R by the relationship
GSD[W], GM[B], and GM[W] are estimated by:
The arithmetic mean of the total distribution of penetration values across the whole cohort, µp, is estimated by:
Nicas (Ex. 156) defines two additional values, d and k, that are based on the parameters described above. The value d represents the 95th percentile of the between-wearer distribution of average penetration values among a cohort of respirator wearers; thus, there is a 5 percent chance that a respirator wearer in the cohort could have an average penetration value of d or higher. If d is set to some penetration value reflecting some minimum acceptable value of protection, the probability that a respirator wearer would fail, on average, to achieve the minimum acceptable penetration value is Pr(Z>z), where
and Z is the standard normal deviate. By estimating the parameters µp, GM[B], and GSD[B] from WPF data, one can estimate the probability that a respirator wearer could have an average penetration value greater than some specified value d.
The value k is defined by Nicas (Ex. 154) based on the distribution of each worker's 95th percentile P value and represented the P value experienced at least 5 percent of the time by 95 percent of workers in the cohort. If k is set to some minimum acceptable P value, the estimated probability that a respirator wearer could fail to achieve the minimum P value at least 5% of the time is Pr(Z>z), where
and Z is the standard normal deviate. Thus, the proportion of workers who fail to achieve a P value of k at least 5 percent of the time can be determined by estimating the parameters µp, GM[B], and GSD[W] from WPF data.
The following hypothetical example illustrates OSHA's use of the model to estimate the risk to workers of experiencing an overexposure while using respiratory protection. Suppose that the WPF values obtained from a group of workers using half-mask, negative-pressure respirators are found to have a geometric mean of 50 (i.e., GM[P] = 1/ 50 = 0.02) and a geometric standard deviation of 3.0 (GSD[P] = 3.0). Furthermore, from one of the WPF studies reviewed by OSHA (Galvin et al.) (Ex. 64-22), it was reported that within-worker variability exceeded between-worker variability in workplace protection factors, with the ratio GSD[W]/GSD[B] = 1.5. From equations 4 through 7 above, and assuming that R = 1.5, then GSD[B] = 1.73, GSD[W] = 2.60, GM[W] = 0.63, and GM[B] = 0.86. The arithmetic average of the cohort's P values, µp, is estimated from equation 8 to be 0.037. If a protection factor of less than 10 (the NIOSH minimum assigned PF for half-mask respirators) is considered to place the worker at risk of an overexposure, then equation 9 predicts a probability of 1.8 percent that a worker in the group would be expected to have an average WPF value of 10 or less (i.e., d is set to 0.1 in equation 9); that is, 1.8 percent of the group of respirator wearers would frequently encounter situations where they are working in a hazardous environment without the minimum protection expected from the respirators being used. By equation 10, there is a substantial probability (47 percent) that a worker in the cohort would not achieve a minimum protection factor of 10 at least 5 percent of the time that respirators are used (i.e., k is set to 0.1 in equation 10).
OSHA used the Nicas and Spear model, the summary data from the WPF studies reviewed above, and the method outlined in the example described above to estimate the probability that a respirator wearer would fail to receive adequate protection from their respirator; the detailed results of this analysis appear in Table V-1, and summary findings are listed in Table V-2. From the studies that reflect the lack of an adequate respiratory protection program, the Nicas and Spear model predicts a high probability (between 36 and 100 percent) that a wearer would not achieve an average protection factor of 10. Data from two of these studies by Toney and Barnhart (Ex. 64-68), and Harris et al. (Ex. 27- 11), when used in the model, suggest a probability of between 13 and 39 percent that the average WPF for a respirator wearer could be 2 or less, which may be considered equivalent to receiving no long-term protection at all. In contrast, workers included in the studies reflecting good respirator program elements would be expected to experience low WPFs much less frequently. The probability that a wearer would attain an average WPF of 10 or less is estimated to be between <0.01 and 3 percent. Results from the studies that reflect good respiratory program practices also indicate that long-term average WPF values at or below 2 would rarely occur. The results from this analysis demonstrate that deficiencies in implementing a good respirator program can greatly increase the chance that the wearer of a negative-pressure respirator will receive less than the minimum expected average protection from the respirator over the long-term, thus increasing the chance that the worker will be exposed to a higher chronic health risk.
OSHA's analysis (Tables V-1 and V-2) also demonstrates that workers using respiratory protection under a deficient program will be exposed more frequently to higher concentrations of airborne contaminants, which may increase the risk that the worker will experience acute health effects. The Nicas and Spear model applied to the studies that reflect inadequate respirator programs predicts nearly a 100 percent chance that a protection factor of less than or equal to 10 would be experienced at least 5 percent of the time. Under conditions of a good respirator program, use of the model suggests no more than a 32 percent chance that WPFs of less than or equal to 10 will occur more than 5 percent of the time.
OSHA finds that, without an adequate respiratory protection program in place, a substantial fraction of respirator users are at risk of being overexposed to hazardous air contaminants due to poor respirator performance. The studies conducted under conditions of a poor respirator program, when analyzed using the Nicas and Spear model, suggest a greater than 50 percent probability that the wearer of a half-mask, negative-pressure respirator will regularly fail to attain the expected minimum level of protection, and that the chance of receiving essentially no protection is substantial. OSHA considers these risks of overexposure to be significant. The studies reviewed by Nelson and the Galvin study indicate that these risks are considerably lower in situations where respirators are used in conjunction with the implementation of strong respiratory protection program elements such as appropriate fit testing, adequate employee training, use of clean respirators in good working order, and regular monitoring of employees to ensure proper respirator use. Thus, OSHA finds that implementation of a comprehensive respiratory protection program, such as the one prescribed by the final rule, will substantially reduce the risk of overexposure that is due to respirator failure. Because such overexposures can place workers at a significant risk of health impairment, as described earlier in this section, OSHA also finds that promulgation of the final rule will substantially reduce the significant health risks associated with those overexposures.
[63 FR 1152, January 8, 1998]
