One type of statistical analysis for WPF data was proposed in 1978. Leidel and Busch suggested the use of 2- and 3-parameter lognormal distributions and tolerance limits for the reporting and interpretation of respirator leakage data.[1] However, these authors specifically omitted any recommendation for the use of lognormal tolerance limits for APF computations. They recognized that APFs based on tolerance limits would tacitly permit some wearers to receive less than APF-level protection.
In 1983, Myers et al. proposed the use of 1-sided lower tolerance limits for APF estimates.[2] However, except for one WPF study reported in 1984,” subsequent respirator researchers have not reported tolerance limits for their WPF results and APF estimates. In 1987, the ISEA stated with regard to the tolerance limit approach for determining APFs:
The proposed rule requires that during analysis of the workplace protection data, 95% of the test subjects must achieve a workplace protection factor with 95% confidence. There is too much variability in the test methods to require the use of confidence intervals. When the confidence interval is added to the prediction, no field test performed to date indicates any tested respirator can meet its assigned protection factor. For example, a half mask respirator with a minimum woricplace protection factor (WPF) of 22 in the DuPont [sic] asbestos study would have a WPF of 6 using the NIOSH [confidence interval] methods.”
If one were to able to conduct multiple WPF respirator-performance studies under conditions similar to the initial study reported by a research team, the resulting study-to-study 5th-percentile WPF point estimates™ would vary considerably due to sampling errors. In general, the smaller the number of test subjects, the larger the potential sampling error and uncertainty are associated with a computed point esti- mate.
- Lenhart, S.W. and D. L. Campbell: Assigned Protection Factors for Two Respirator Types Based
Upon Workplace Performance Testing, Ann. Occup. Hyg. 28(2):173~182 (1984), pp. 180-181.
“Industrial Safety Equipment Association: Key Issues on NIOSH Notice of Proposed Rulemaking for Testing and Certification of Respirators for Use in Mines and Mining, enclosure transmitted in a letter “To Our Customers and Distributors” from C. D. Cowan of the 3M Company, St. Paul, Minnesota (October 9, 1987), Item 1.4.2, p. 2.
“A 5th-percentile WPF point estimate from a sample of WPF s estimates the WPF value for which 5% of all similariy-obtained WPF's will be less than or equal to. Correspondingly, it estimates the WPF value
for which 95% of all similarly obtained WPF will exceed.
- ↑ Leidel, N. A. and K. A. Busch: Statistical Methods for Analysis of Respirator Data, paper presented at the 1978 American Industrial Hygiene Conference, Los Angeles, CA (May 10, 1978), pp. 12-13.
- ↑ Myers, W. R, Lenhart, S. W., Campbell, D. and G. Provost: The Forum—Letter to the Editor, Am. Ind. Hyg. Assce. J. 44(3):B25-26 (1983), p. B-26.
