VARIATION IN QUANTITATIVE RESPIRATOR FIT FACTORS DUE TO FLUCTUATIONS IN LEAK SIZE DURING FIT TESTING

Variation in fit factors during quantitative respirator fit testing wa s studied for a high degree of fit (aerosol fit factors > 1000) and a low degree of fit (aerosol fit factors < 1000). In a controlled human study, fit factors were determined sequentially for three different ex ercises by (1) an aerosol fit rest (using a portable condensation nucl ei counter and room aerosol as the test agent) and (2) the newly devel oped dichotomous-flow fit test (Dichot). For the higher level of respi rator fit, the aerosol fit factors were 30 to 60 times the correspondi ng flow fit factors, and for the lower level of respirator fit they we re 2 to 4 times the flow fit factors. A coefficient of variation (CV) of 84% (GSD 1.6) for the higher respirator fit and 178% (GSD 2.2) for the lower respirator fit data was observed in the human study when aer osol fit factor's for the three exercises were pooled. In a similar ma nnequin study, the center sampling probe gave aerosol fit factors with a CV of 5.4% (GSD 1.05). The flow fit factors for all three exercises pooled had a CV of 36% (GSD 1.3) for the higher respirator fit and 40 % (GSD 1.5) for the lower respirator fit data, while the mannequin stu dy gave flow fit factors with a CV of 2.2% (GSD 1.02). This, the varia tion in fit factors obtained in the human study was much higher than t hat obtained in a mannequin study. However, the variation in the aeros ol method relative to the flow method, in the human study, is of the s ame order of magnitude as in the mannequin study. This suggests that t he higher variation in the human tests is mainly due to variations in faceseal leak size and not to increased systematic errors. It was esti mated that the fluctuations in faceseal leak size for the subject with the high fit factor varied between 0.5 mm and 0.7 mm, and between 1.0 mm to 1.3 mm for the subject with the low fit factor. Thus, the fit f actor determined for a human cannot be expected to be constant even fo r the most perfect test system.