Relationship Between Fit Factors, Penetration, and Mask Leakage

(Reprinted with permission from: Respiratory Protection Update 7(4):1-9. ISSN 1048-6658, Respirator Support Services, 2028 Virts Lane, Jefferson, MD 21755, 1995.)

Quantitative respirator fit testing (QNFT) was developed and commercialized more than 25 years ago. QNFT results have historically been expressed as fit factors. The fit factor is defined as the ratio of outside challenge agent concentration to in-mask agent concentration measured during a fit test. It presumably represents the number of reductions in outside agent concentration achieved by the respirator system during the fit test. A more direct expression of mask leakage is produced by the ratio of measured in-mask concentration over outside concentration. This variable is defined as penetration, and is related to the fit factor as a reciprocal function (penetration = 1 / fit factor). Mask penetration presumably represents challenge agent penetration, or leakage, into the respirator system during the fit test.

Because the measured in-mask concentration variable appears in its denominator, the fit factor becomes extremely volatile when low levels of in-mask challenge agent concentration are measured. Table I illustrates this phenomenon. An extremely small change in measured aerosol penetration from 0.02% to 0.01% produces an apparent large change in fit factor from 5,000 to 10,000. The common misperception that large changes in fit factors represent large changes in mask leakage has been so widespread that it has highly influenced the structure and application of QNFT. Elaborate fit test exercise protocols have been developed and written into national standards such as OSHA=s asbestos and lead standards. The focus of fit testing has been almost exclusively on exercise protocols, which require a substantial amount of time to complete. As a consequence, the more important aspects of fit testing, which involve 1) trying to determine the best fitting mask for each individual required to wear a mask, and 2) ensuring that each individual can consistently achieve an acceptable fit when their mask is donned, have historically been ignored.

The influence of fit factors on our expectations of respirator performance is perhaps best illustrated by what has happened with gas masks. Historically, a fit factor of 10,000 was used as a minimum passing level during laboratory studies of gas mask fit. Generated aerosol fit test systems were normally used to conduct those studies, and fit factors exceeding 10,000 were commonly measured. The expectation that military gas masks could routinely provide such levels of fit was wide-spread, based on comments made by researchers from both the United States and Europe during a 1995 Vancouver, B.C. conference sponsored by the International Society for Respiratory Protection. A closer examination of Table I provides some interesting insights regarding such expectations.

^a In-mask particle count (particles/ml) given typical outside count of 5,000 particles/ml

^b (% penetration) x (default inspiratory flow rate of 53,800 ml/min)

^c Pass/fail level recommended for TSI Portacount by several members of ANSI

Committee on Respirator Fit Test Methods

The % penetration (% PEN) column shown in Table I was produced by multiplying the reciprocal of the corresponding fit factor by 100. The result provides an indication of mask leakage associated with the fit factor. For example, a fit factor of 100 indicates that 1% of the aerosol concentration outside the mask penetrated (leaked) into the mask through a leakage path. When HEPA filters are used, it is generally assumed that essentially no aerosol penetrates into the mask through its air-purifying (filtration) path. Since air is the medium that carries aerosol or any other contaminant into the mask, a fit factor of 100 therefore means that 1% of the air entering the mask flowed through a leakage (non-filtering) path while 99% of the air entered through the air-purifying (filtering) path.

A fit factor can be converted into an equivalent leakage flow rate in ml/min by multiplying % penetration by the total flow rate into the mask during inspiration, when leakage into the mask actually occurs. Respirator inspiratory flow rates are primarily a function of the work rate of the wearer and the resistance of the air purifying cartridges used with the respirator.⁽¹⁾ The total flow rate used in Table I is the mean inspiratory flow rate used by the FitTester 3000 Controlled Negative Pressure (CNP) fit test system to convert directly measured mask leakage rates into equivalent fit factors. It should be noted that the CNP programmed mean inspiratory flow rate of 53.8 l/min is representative of mask inspiratory flow rate at a moderate work rate. If an aerosol fit test is conducted at a lower work rate (lower inspiratory flow rate), then the equivalent leak rates shown in Table I would be lower.

With a better understanding of the relationship between fit factors and respirator leakage, a closer look at the aerosol-based fit factors that have played such a major role in defining the structure and application of quantitative fit testing is most informative. Generated or ambient aerosol fit test systems have produced most of the available data on respirator fit. Based on the magnitude of the fit factors reported by those systems during fit tests, a number of experts currently advocate using a fit factor of 5,000 as a pass-fail level for the TSI Portacount ambient aerosol fit test system.

If we establish 5,000 as the passing fit factor level for air-purifying respirators, we obviously expect that most respirator wearers will be able to achieve fit factors above that level. The US Marines currently use 6,667 as the passing level for the M-40 gas mask, which is also a negative pressure air-purifying respirator. Based on the leakage levels associated with these fit factors, is it reasonable to assume that air-purifying respirators can routinely deliver such performance? The results of a whole lot of CNP fit tests conducted on both respirators and gas masks indicate that it is not.

A substantial number of problems and measurement biases related to aerosol fit test systems have been identified and investigated by a number of research scientists working in different laboratories. Leak geometry has been shown in laboratory experiments to be an important determinant of aerosol penetration into a respirator.^(2-7) The location and depth of sampling probes used to collect in-mask samples during a fit test have also been shown to significantly affect fit test results.^(8-10) Phenomena related to aerosol streamlining past the sampling probe can add positive or negative bias to in-mask sample results. Aerosol deposition in the lungs decreases the likelihood that penetrating aerosol will be collected by the sample probe and consequently detected as leakage by an aerosol system's detector. Although such phenomena can reduce the amount of aerosol leakage detected, they have little or no effect on the actual penetration of contaminant into the respiratory system. The presence of such biasing factors during aerosol fit testing can produce inflated fit factors, and decreases measures of leakage, relative to actual lung exposures.

The failure to detect mask leakage efficiently results in low in-mask aerosol counts, which produce artificially high, and extremely volatile, fit factors. Small changes in detected leakage are often mis-interpreted as large changes in mask fit. Inflated expectations relative to actual mask performance are also produced. As pointed out in Table I, the maximum allowable penetration or leakage through a HEPA filter is 0.03%. That level of penetration is equivalent to a fit factor of 3,333. It seems ironic to expect less leakage through the facepiece-to-face seal of air purifying respirators than would be allowed through a HEPA filter.

Analyzing QNFT results by comparing individual exercise fit factors can be extremely misleading. The volatility of fit factors based on low measurements of in-mask aerosol concentration can mask small changes in mask leakage appear to be large changes in mask fit related to the exercises. As a consequence, fit test protocols today are defined in terms of a series of test exercises, even though no scientific evidence has been produced to show that such exercises significantly alter mask fit. On the other hand, mask donning has been shown to have a substantial effect on mask fit.

Conducting more than one repetition of a QNFT today seems to be the exception rather than the rule. Most fit tests are conducted with a single mask donning, even when multiple repetitions of the test protocol are accomplished. As a consequence, most decisions regarding the acceptability of mask fit are based on a single mask donning. Based on observing what happens to measured respirator fit when a mask is removed and redonned, we conducted a study to determine the relative importance of mask donning and test exercises on measured respirator fit when the same mask is donned by the same person a number of times.⁽¹⁴⁾ The objectives of the study were to: 1) examine the effects of mask donning and fit test exercise protocols on measured fit factors, and 2) determine how effectively two fit test systems can differentiate between respirators with different levels of fit. Fourteen active-duty Air Force personnel previously assigned to a respiratory protection program participated in the study. A broad range of respirator fit was generated among the test subjects by assigning two different sizes of the same respirator type (half-mask or full-face) to each subject.

A controlled negative pressure (CNP) fit test system (FitTester 3000, DNI, Inc., Carson City, NV) and an ambient aerosol (PC+) fit test system (Portacount Plus, TSI, Inc., St. Paul, MN) were used to conduct three fit tests per day with both respirators assigned to each subject for a period of five consecutive days (ie. 2 fit test systems x 14 subjects x 2 masks/subject x 3 fit tests/mask/day x 5 days = 840 total fit tests.) Respirators were removed and redonned by the subjects between each fit test. With the exception of positive/negative pressure checks conducted by the subjects after each donning, no guidance or feedback was provided by test administrators when each mask redon was accomplished.

Fit test failure rates determined by each system for each mask are presented in Table II. In general, Mask 1 for each subject fit better than Mask 2 (same brand and type, different size). The mask fit failure rate detected by the CNP system was more than three times higher than the fit failure rate detected by the ambient aerosol system.

Failure Rate, %

CNP PC+

Mask 1 21.4 4.8

Mask 2 41.4 14.8

Overall 31.4 9.8

All CNP and PC+ fit factors were log-transformed prior to statistical analysis. One-way analyses of variance were performed to test for significant differences between mask leakage during the individual exercises of the fit test protocols. Comparisons between the first normal breathing fit factors and the overall fit factors were made to examine exercise effects by determining how often a pass-fail decision based on the overall fit factor differed from the decision that would have been made had only the first normal breathing fit factor been used. No significant differences were detected between the overall respirator failure rates predicted by the first normal breathing versus the overall fit factors. In other words, the information we had about respirator fit after conducting the first normal breathing fit test did not differ substantially from the information we had after conducting all of the exercises in most of the 840 tests conducted.

The effects of donning on mask fit paint another picture. In order to graphically show that picture, a different way of looking at fit test results was devised. Instead of directly comparing fit factors, which can be very misleading, we wanted a way of showing two pieces of information so that direct comparisons of different fit tests could be shown graphically. The first piece of information was did the person pass or fail the fit test? The second piece of information was did the person pass or fail the test by a little or a lot?

Overall fit factors were transformed into multiples of the criterion value (MCV) where the criterion value = the pass/fail level established for each respirator type. For half-mask respirators, the pass/fail level was a fit factor of 100, while a value of 500 was used for full-face respirators. For example, an MCV of 2.5 would mean that the overall fit factor for the fit test was 2.5 times higher than the minimum passing fit factor. Therefore a respirator that just passes a fit test would have an MCV value of 1.0 (ie. a fit factor of 500 for a full-face respirator has MCV = 1.0). Since the log of 1.0 is zero, graphing the log of the MCV for each fit test provides an immediate visual determination of whether the respirator passed (positive value) or failed (negative value) each fit test by a little (shorter bar) or a lot (taller bar). An example of transforming fit test results into MCV values is shown in Figure 1, which gives fit test results for two of the subjects in the study. Each bar represents the results of a fit test conducted after a mask donning. The three fit tests conducted each day with each mask and each system are grouped together. The fifteen fit tests conducted with each mask are presented for each fit test system.

Transforming fit test results into log MCV values provided an opportunity to observe the variability associated with 14 different subjects donning two different sizes of the same type respirator on 15 separate occasions over the course of one week. Unlike the relatively limited variability exhibited in the fit test exercise data, a much greater degree of variability in mask leakage and fit was shown by the mask donning data.

Quantitative respirator fit testing has been conducted in the United States for more than three decades. During that time, elaborate and time-consuming fit test protocols have evolved that stress fit test exercises. Missing from current fit test protocols is a focus on selecting the best fitting mask for each individual, and ensuring that each individual can consistently achieve an acceptable fit when their mask is donned. Study results indicate that mask donning has a much more dramatic effect on mask leakage and fit than do exercises accomplished during the fit test. It is apparent that more time and effort should be directed towards assessing donning efficiency at the expense of trying to determine the effects of exercises.

1. Crutchfield, C.D., T.K. Pham, and M.D. Van Ert: "Determination of Inspiratory Pressures and Flow Rates for Work Rate Dependent Quantitative Respirator Fit Testing". Appl. Occup. Environ. Hyg. 8(2):103-107, 1993.

2. Holton, P.M., D.L. Tackett and K. Willeke: Particle Size-Dependent Leakage and Losses of Aerosols in Respirators. Am. Ind. Hyg. Assoc. J. 48(10):848-854, 1987.

3. Holton, P.M., and K. Willeke: The Effect of Aerosol Size Distribution and Measurement Method on Respirator Fit. Am. Ind. Hyg. Assoc. J. 48(10):855-860, 1987.

4. Chen, C.C., and K. Willeke: Characteristics of Face Seal Leakage in Filtering Facepieces. Am. Ind. Hyg. Assoc. J. 53(9):533-539, 1992.

5. Hinds, W.C. and G. Kraske: Performance of Dust Respirators with Facial Seal Leaks. I. Experimental. Am. Ind. Hyg. Assoc. J. 48(10):836-841, 1987.

6. Hinds, W.C., and P. Bellin: Effect of Facial Seal Leaks on Protection Provided by Half-Mask Respirators. Appl. Ind. Hyg. 3(3):158-164, 1988.

7. Loebach, D.V.: TSI Portacount Side-by-Side Comparisons. OSHA Cincinnati Laboratory, Cincinnati, May 2, 1989.

8. Myers, W.R., J. Allender, R. Plummer and T. Stobbe: Parameters that Bias the Measurement of Airborne Concentration Within a Respirator. Am. Ind. Hyg. Assoc. J. 47(2):106-114, 1986.

9. Myers, W.R., J. Allender, W. Iskander and C. Stanley: Causes of In-Facepiece Sampling Bias - I. Half-Facepiece Respirators. Ann. Occup. Hyg. 32(3):345-359, 1988.

10. Myers, W.R. and J. Allender: Causes of In-Facepiece Sampling Bias - II. Full-Facepiece Respirators. Ann. Occup. Hyg. 32(3):361-372. 1988.

11. Oestenstad, R.K., J.L. Perkins, and V.E. Rose: Identification of Faceseal Leak Sites on a Half-Mask Respirator. Am. Ind. Hyg. Assoc. J. 51(5):280-284, 1990.

12. Oestenstad, R.K., H.K. Dillon, and J.L. Perkins: Distribution of Faceseal Leaks in a Half-Mask Respirator and their Association with Facial Dimensions. Am. Ind. Hyg. Assoc. J. 51(5):285-290, 1990.

13. Willeke, K., and U. Krishnan: Present Procedures in Quantitative Respirator Fit Testing: Problems and Potential Solutions. Appl. Occup. Environ. Hyg. 5(11): 762-768, 1990.

14. Crutchfield, C.D., E.O. Fairbank, and S. Greenstein: Effects of Exercises and Mask Donning on Respirator Fit. Presented at the American Industrial Hygiene Conference, Washington, D.C., May, 1996.