A previous study has determined the ability of five fit-test methods (BitrexTM, saccharin, generated aerosol, PORTACOUNT Plus corrected for filter penetration, and PORTACOUNT Plus/N95 COMPANIONTM) to screen out poorly fitting N95 filtering-facepiece respirator models. In that study, the results were compared to the 5th percentile of the simulated workplace protection factor (SWPF). No fit-test method met the new American National Standards Institute (ANSI) 288.10 standard of no more than 50 percent for the alpha error (probability of rejecting an adequately fitting respirator; erroneous failure) and no more than 5 percent for the beta error (probability of accepting an inadequately fitting respirator; erroneous pass). The correlation between the PORTACOUNT Plus SWPF values and actual exposures is not 1.0, and their relationship can be better described by a quadratic equation based on available data. In this study, the original analysis was repeated using five alternate approaches to the 5th percentile as the reference test: (1) the unadjusted mean SWPF; (2) the bias-adjusted mean SWPF (i.e., the mean SWPF adjusted to account for possible bias in the fit-test instrument); (3) the bias-adjusted 5th percentile SWPF; (4) the unadjusted individual minimum SWPF; and (5) the bias-adjusted individual minimum SWPF. With these alternate references, the range of alpha errors for the Bitrex, saccharin, generated aerosol, PORTACOUNT Plus, and PORTACOUNT Plus/N95-COMPANION fit tests were 41 to 59 percent, 38 to 67 percent, 67 to 86 percent, 51 to 79 percent, and 38 to 87 percent, respectively. The corresponding beta errors were 6 to 25 percent, 9 to 16 percent, 0 to 3 percent, 2 to 6 percent, and 0 to 19 percent, respectively. Even with the alternate reference criteria for respirator performance, none of the fit-test methods met both error goals of the ANSI 288.10 standard.
NIOSH, Division of Respiratory Disease Studies, 1095 Willowdale Road, Morgantown, WV 26505-2888