Document XO4k6v6rOe1oGvpga0KMeb32R

TO: Distribution TGG: JCL: FGJ: AC: AJO: RF XF: * * Interoffice Communication FROM: DATE: SUBJ: T. G. Grumbles September 24, 1991 VISTA COMPARISON OF FIT BETWEEN NATURAL AND SILICONE RUBBER FACEPIECES The attached article describes a study comparing the fit factors obtained by natural and silicone rubber facepieces. In summary , the fit factors for the silicone rubber facepiece were found to be significantly greater than those for natural rubber face pieces. However, the variability of fit factors was higher for the silicone facepieces. This data should be considered when evaluating new respirators. T. G. Grumbles dlj .328 Attachment SAFETY DIRECTORS Bruce Trego-Aber, Brent White-Bait, George Williams-Blane, Matt Tonkovich-Hmd, K. L. Fogg-LCCP, R. V. Gantz-LCLAB, Mike LunsfordLCVCM, Mark Marker son-Okc, Greg Lipps-Premiere, R. B. Martin-Austin, J. R. Drurawright, J. G. Farrier, L. L. Zimmerman T* " yvv 000007921 Comparison of Fit Provided by Natural d Silicone Rubber Facepieces of the me Brand of Half-Mask Respirator K. Oestenstad* and Angela M. Zwissler* ersity of Alabama at Birmingham, School of Public Health, UAB Station, Birmingham, Alabama 35294; ielin Tire Corporation, P.O. Box 2846, Greenville, South Carolina 29602-2846 The fit provided by natural and silicone rubber facepieces of If properly formulated, silicone rubber is more pliable, be same brand and size of half-mask respirator were compared by performing multiple fit tests in random order with both facepieces on 45 subjects. The tests consisted of a six-exercise pro tocol while measuring the penetration of ambient room aerosols with a condensation nuclei counter fit test device. Fit factors obtained by the two facepieces were compared by nonparametric Statistical methods because fit factors for the silicone rubber facepiece were not lognormally distributed. The lit factors for the silicone rubber facepiece were found to be significantly greater tfrn those for the natural rubber facepiece. Although the dif ference was statistically significant, it was not of the same mag superior in chemical resistance, and just as durable as natural black rubber. However, narural rubber may have better tear resistance. Because of their increased pliability, it would be reasonable to expect that silicone rubber facepieces would conform better to wearers' faces and provide more effective faceseals than their natural rubber ana logues. However, there is only limited information in the literature to support this assumption. As part of quantitative fit tests (QNFT) to develop per formance data for self-contained breathing apparatus (SCBA), nitude as reported in previous studies. It was noted that the Hyatt0* reported results of tests on silicone and neoprene maximum and minimum fit factors obtained with the natural rubber full facepieces made from the same mold. Fit fac rubber facepiece were higher than those for the silicone rubber tors (FFs) for the two facepieces equipped with demand facepiece. The variability of fit factors for the silicone rubber facepiece was also found to be higher than that of the natural rubber facepiece. This would affect the calculation of lit factor lower tolerance limits and should be considered in estimating the lowest likely fit provided by a respirator for an individual or a group of wearers. Oestenstad, R.K.; Zwissler, A.M.: A Comparison of Fit Provided by Natural and Silicone Rubber Facepieces of the Same regulators were measured with DOP aerosol while subjects performed a five-exercise test protocol. The test panel con sisted of 31 firemen selected according to face length and face width criteria of the Los Alamos full-face respirator test panel.0* Hyatt found that 95 percent of the subjects achieved a 8rand of Hatf-Mask Respirator. Appl. Occup. Environ. Hyg. 6:785-789; FF^ 100 while wearing the silicone rubber facepiece, 1991. T1* whereas only 87 percent achieved a FF & 100 with the neoprene rubber facepiece.11 * Log-probability plots of the Introduction data from this study indicate that the FFs for both facepieces were approximately lognormally distributed. The geo During the past decade there have been significant metric mean for the silicone facepiece was about 23,000, changes in respirator technology that have resulted in in and the geometric mean for the neoprene facepiece was creased comfort and efficiency of these devices. These only 570. The researcher described the silicone facepiece changes include: improved filter materials to reduce as being soft and pliable and the neoprene facepiece as breathing resistance; better sealing surfaces from changes being hard and rigid. in facepiece design; the availability of sized facepieces; As part of a report of the effectiveness of an anthropo- improved weight distribution by using crown straps and metricallv designed half-mask respirator, Jones etal.(i) facepiece yokes; and the use of silicone rubber to provide compared the fit obtained with medium-sized facepieces more comfortable facepieces. Silicone rubber facepieces made of silicone and natural rubber. They determined FFs have been promoted extensively by manufacturers as being for the two facepieces using a com oil aerosol while sub significant improvements to their product lines. They cite jects performed a six-exercise test protocol. The test panel a more comfortable fit, lighter weight, ease of cleaning, consisted of 25 subjects (12 female and 13 male) who were and better durability as advantages. selected according to face length.and lip width criteria of APPL OCCUP. ENVIRON. HYG. N9) SEPTEMBER 1991 1047-322X/91/0609-785S2JXV5 1991 AIH VVV 000007922 785 the Los Alamos half-mask test panels2* The range of FFs for the silicone rubber facepiece was 174-18,750 with an average of 2?35/3) The range of FFs for the rubber facepiece was 20-838 with an average of 214. The researchers attributed the better fit to silicone's lower coefficient of friction that allowed it to seal/conform better to facial surfaces. The researchers did not report tests for the distribution of FFs or for the difference be tween the mean FF for the two facepieces. These studies113) would lend strong support to the as sertion that respirators made of silicone rubber provide a much more efficient faceseal than those made of natural rubber. However, there are limitations in the study designs that might restrict interpretations of those results. Neither study accounts for intrasubject variability in fit because only one QNFT was performed by each subject witiveach res pirator. daRoza etal.^ found that the fit obtained by one subject with a given facepiece was defined by a lognormal distribution and that the geometric standard deviation for that distribution varied widely among individuals. There fore. multiple QNFTs must be performed on the same subject/respirator combination to make an estimate of that distribution. The FF for that subject can then be defined as the geometric mean, or the probability of experiencing a minimum FF can be calculated from the geometric mean and geometric standard deviation. The variation of FFs for a group of subjects has also been shown to be lognormallv distributed/"*55 Failure to test for this distribution and perform the appropriate transfor mations when comparing FFs obtained by subjects wearing different respirators can be misleading and can result in the use of inappropriate statistical tests. It is not clear from the data reported by Jones etalyi) if these factors were considered. The means they reported for the two facepieces were not identified as arithmetic or geometric. The purpose of the present study was to compare the fit obtained by silicone and natural rubber facepieces of the same model and size of a half-mask respirator while accounting for intrasubject fit variability of the test subjects. This was accomplished by conducting replicate tests with each facepiece on 45 subjects and by testing for the sig nificance of any observed difference between FFs. Materials and Methods Respirator The respirators used in this study were size medium, U.S. Safety (United States Safety' Service Co., Kansas City', Missouri) Series 100 silicone rubber and Series 200 natural rubber half-mask facepieces equipped with HEPA filters. Each facepiece was fitted with a sampling port on the midline at a location between the subjects nose and mouth. The facepieces were disassembled, cleaned, reassembled, and inspected after the tests on each subject. Subjects The 45 subjects used in this study were university stu dents, staff, and faculty. Of these, 23 were male and 22 were female, 20 had previous experience wearing a res- S pirator for more than one hour at least once in a workplace X setting, and 25 did not have that experience. Persons with facial features that would have resulted in obvious faceseal ^ leaks were not used in the study. Subjects were not selected ' on the basis of anthropometric dimensions of the Los Ala mos respirator test panel/25 The rationale for this selection criteria was to compare the fit of the two facepieces over a wide range of FFs, recognizing that one size facepiece would not be expected to provide a good fit for even* member of a group of wearers/65 Quantitative Fit Test Methods Respirator fit tests were conducted with the Portacount* Respirator Fit Tester- (TSI, Inc., St. Paul, Minnesota). This device is a continuous flow condensation nuclei counter (CNC) that utilizes ambient room particles as a test aerosol. The CNC increases the size of sampled aerosols as small as 0.02 p.m to about 0.1 |xm by condensing isopropanol on the particles. The resulting droplets are then counted bv a light-scattering laser photometer. It can measure particle counts in the range of 0.1 panicles/cm3 to 5 x 105 particles/cm3. Willeke et aiS7) first proposed the use of the CNC for fit testing with nonhazardous submicrometer aerosols. The CNC has since been found to provide good agreement in measuring FFs when compared with the traditional DOP aerosol photometer method/8-95 Test Protocol A series of six fit tests were performed on each subject: three tests while wearing the silicone rubber facepiece and three tests while wearing the natural rubber facepiece. The type of facepiece worn in each test was randomly selected by the flip of a coin until one type had been tested three times. A positive pressure check was performed be fore each test, and an initial measurement was made with the CNC to confirm a FF > 10. FFs were measured while the subjects performed each of the following exercises for 1.5 minutes: normal breathing, deep breathing, moving the head side-to-side, moving the head up-and-down, talk ing, and normal breathing. The respirator was removed at the end of each test, and the sealing surface was wiped with an alcohol swab. The subjects also wiped their faces with paper towels to remove any accumulated perspiration. Results and Discussion Data from each test were entered into a SOLO database (BMDP Statistical Software, Inc., Los Angeles, California) for sorting and for performing necessary transformations and appropriate statistical analyses/105 The overall FF for each test was calculated by the following equation/115 g Overall FF = ------------------- y1 ^Exercise FF, VVV 000007923 (1) The geometric mean of the three FFs for each facepiece on each subject was used as the best estimate of the FF 786 APPL OCCUR. ENVIRON. HYG. 6(3} > SEPTEMBER 1991 Subject Rt Factors Natural Rubber Facepiece It F2 f3 r4 l5 b6 .. 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 Test 1 713 32233 3790 3626 17031 15912 457 57829 12740 8211 8270 188 14333 4404 2287 12944 27116 2891 6505 389 3324 9982 2097 1385 922 11221 46813 23148 6369 16050 6650 2329 12083 4083 1104 513 3097 10637 2903 10475 11442 6933 6649 12160 15349 Test 2 53 29321 1975 5548 10173 30059 2055 9952 12704 4135 8521 5128 8768 1208 2399 10697 48835 2058 5197 463 1834 7714 2357 4059 338 15036 42726 16120 7224 10141 7790 5494 9248 517 1975 554 481 7390 566 8830 6834 5539 6499 12985 5289 Test 3 311 29127 1875 5657 9760 4998 5349 32024 7870 2827 6509 2472 8213 1025 1944 8959 37763 1919 3072 1727 12626 7665 1362 3663 2440 28220 18582 22412 5125 10358 10666 4961 6346 9463 8445 2186 1687 18999 596 2337 7087 5894 5634 2425 3931 GM* 228 30194 2412 4845 11914 13371 1713 26415 10840 4578 7712 1335 10106 1760 2201 10745 36641 2252 4700 678 4254 8388 1888 2741 913 16824 33376 21105 6184 11902 8206 3995 8933 2714 2641 853 1360 11431 994 6001 8214 6094 6244 7262 6833 *GM - geometric mean. GS0 - geometric standard deration ol n-3 replicates. GSO* 3.76 1.06 1.48 1.29 1.36 2.48 3.45 2.45 1.32 1.72 1.16 5.68 1.36 2.22 - 1.12 1.20 1.34 1.25 1.47 2.26 2.69 1.16 1.33 1.81 2.68 1.60 1.66 1.14 1.19 1.30 1.27 1.60 1.38 4.46 285 2.26 2.58 1.61 2.53 2.27 1.33 1.12 1.09 2.59 205 Test 1 7038 33477 24135 63706 29970 6008 35350 31324 61 8 264 4277 3531 15755 24380 2332 2086 35840 29209 11362 26368 15794 2106 7741 1120 18319 21581 27425 6458 2451 543 42913 2466 218 27022 7391 10770 13143 9622 10193 18968 4265 6828 1212 10730 SIDcone Rubber Facepiece Test 2 Test3 GM* 8749 33575 19985 14703 19698 18340 17279 27143 36 2159 4019 7107 3861 32777 27654 2972 309 16564 5522 5164 29543 16571 18935 7957 120 22415 15610 40086 5844 13557 10832 35653 13622 210 10236 2202 7877 20439 9663 13631 12869 15808 6620 13172 2990 5365 30214 16061 10965 20527 14413 32827 24127 13 50 1997 6688 2681 - 12322 2697 3001 2921 17685 9360 7729 18277 14246 216 7981 19 18607 13552 17577 9400 646 10650 13470 12329 1022 16426 6292 1887 16194 563 8747 9810 16585 6877 15268 9160 6921 32384 19795 21746 22969 11667 27167 27374 31 96 1284 5680 3324 18531 12205 2750 1235 21897 11472 7683 24237 15507 2055 7892 136 19695 16589 26834 7079 2779 3971 27417 7454 361 16563 4678 5430 16324 3742 10572 133778 10380 6774 6247 6648 GSO* 1.28 1.06 1.23 2.57 1.26 1.80 1.48 1.14 2.18 17,09 4.11 1.32 1.21 1.66 3.70 1.15 3.36 1.53 2.34 1.48 1.29 1.08 9.33 1.02 7.78 1.12 1.27 1.51 1.28 4.60 5.60 1.86 2.61 2.47 1.62 1.93 2.53 125 5.15 1.25 1.39 2.16 1.02 4.14 201 for that subject/facepiece combination. This method of cal culation was intended to account for intrasubject variability of fit as identified by daRoza etalS4> The FFs for the three tests with both respirators and the geometric mean and the geometric standard deviation of those tests for all sub jects are shown in Table I. A review of the data revealed that the geometric standard deviations for the silicone rubber facepiece for Subjects 10 and 23 were 17.09 and 9 33, respectively. These values were much higher than expected/4-5) therefore, Natrella's test2) was applied to the variances of the three log FFs for the silicone facepiece to determine if the data for the two subjects could be classified as outliers. That test re jected the variance for Subject 10 as an outlier'(ct. = 0.05). As a consequence, all the data for that subject were ex cluded from subsequent statistical analyses. The cumulative distributions of the mean log FFs for both facepieces on the remaining 44 subjects are shown in Figure 1. It was noted that the values for the natural rubber facepiece followed an approximately straight line indicating a lognormal distribution. However, the values for the silicone rubber facepiece deviated substantially from a straight line. The Martinez and Inglewicz tes/13^ was applied to test the data for normality. That test found the log FFs for the natural rubber facepiece to be normally distributed (a = 0.05), but the distribution of the log FFs for the silicone rubber facepiece was not. Therefore, nonparametric statistical tests were used to compare FFs ob- APPL OCCUR. ENVIRON. HYG. Ml . SEPTEMBER 1991 VVV 000007924 787 11.00 10.00 9.00 TT + Natural Rubbar Sllcone Rubber ... y *** ** ***** 700 600 500 - t 400 - ________ i 1 e -- 10 i i i i__ i, .a i 20 90 40SOM 70 SO Percent Less Than i. 90 90 FIGURE 1. Cumulative distributions of loo FFs for the natural rubber and silicone rubber facepieces. tained with the two facepieces. geometric mean FF obtained by a subject with the natural The observation that log FFs for the silicone rubber rubber facepiece was higher than that obtained with the facepiece were not normal was unexpected and does not silicone rubber facepiece, and the minimum value for the follow previous studies that found FFs to be lognormaily silicone rubber facepiece was lower than that for the nat distributed.*1,4-5,14* It is not known why this distribution ural rubber facepiece. However, the median and mean FFs was observed with the silicone rubber facepiece only, be for the silicone rubber facepiece were higher than their cause fit tests on both facepieces were performed in a respective values for the natural rubber facepiece. It was random sequence at the same time. When the mean log also noted that the variation in fit was greater for the sil FFs for the silicone rubber facepiece were sorted by gen icone rubber facepiece as indicated by higher maximum, der. it was found that the distribution for males was log median, and mean values of the geometric standard de normal, whereas that for females was not*13* Therefore, viation for replicate tests on each subject. Variations in this phenomenon may be due to the gender differences standard deviations would affect the lower confidence lev in critical facial dimensions identified by Johnson*15* and els determined for the two facepieces. to the gender differences in faceseal leak sites observed The 95 percent lower tolerance limit is the value above by Oestenstad etalSl6) which it could be predicted with some level of confidence Correlation coefficients were determined as a test of the that 95 percent of the FFs in a population would lie, and possible effect of room particle count on the observq^j^s. -- this value has been used to determine assigned protection Spearman's correlation coefficient was 0.25 for room'par factors from workplace protection factor data.*17* If both ticle count and the natural rubber FFs, and it was 0.15 for distributions were lognormal, the mean of the logs and room particle count and the silicone rubber FFs.*10* There their standard deviations could be used to calculate the 95 fore. it was assumed that room particle count had no effect percent lower tolerance limits for FFs for the two respi- on measured FFs. In a sequence of fit tests, it is possible that repeated donning of a respirator might result in a learning effect producing higher FFs for the last tests. To test for this effect, the FFs for both facepieces were sorted into the sixtest sequence in which they were performed on each sub ject. and a Kruskal-Wallis ANOVA was performed.*10* That test indicated there was no significant difference within the six-test sequence (p -- 0.470). Therefore, it was as sumed that a learning effect did not influence the FFs obtained with the two facepieces. Examination of Table II indicates that the maximum TABLE II. Summary of Facepiece FB Factora Natural Rubber Facepiece Sldcone Rubber Facepiece GM* GS0* GM* GSO* Maximum 36845 5.68 32386 9.33 Median 6140 1.54 9136 1.60 Mean 6401 1.94 11800 2.34 Minimum 228 1.06 31 1.02 *GM - geometric mean of three replicates with the same respirator on to same subject *GSD " geometric standard deviation ot replicates with the same respirator on the same subject VVV 000007925 788 APPL QCCUF. ENVIRON. HY6. &9> SEPTEMBER 1991 se values would be 456 for the natural rubber and 352 for the silicone rubber facepiece at a ^jence level of 95 percent. This indicates that the rub)iece would be expected to provide a better fit percent of the subjects wearing that respirator. Althese values exceed the commonly accepted asprotection factor of 10 for half-mask respirators,<l8) point illustrates the importance of using quantitative or fit testing for selecting the best-fitting respirator subject and performing multiple tests of the respiptor/subject combination to determine the variation of fitt4) i: The statistical significance of differences between FFs |r the two facepieces was tested by the Wilcoxon Sign lank Test and the two-sample sign tesL{10) The Wilcoxon ;ign Rank Test showed that the FFs for the silicone rubber acepiece were significantly greater than those for the-natfural rubber facepiece (p = 0.0134). The two-sample sign test found that the probability of observing higher FFs for the silicone rubber facepiece in 30 of the 44 pairs was 0.0123- Conclusions The median FF for the natural rubber facepiece was 6140 and the mean was 8401; corresponding values for the sil icone rubber facepiece were 9136 and 11,800, respectively. Both independent and matched nonparametric tests found the FFs for the silicone rubber facepiece to be significantly greater than those for the natural rubber facepiece. Al though these differences are statistically significant, they do not confirm the order of magnitude differences re ported by Hyatt^ and Jones etaL<3) The results of this study indicate that this brand of halfmask silicone rubber facepiece provided a better average fit than its identical natural rubber twin in a laboratory study. The study was designed to account for intrasubject variability in making the comparison. Fit was compared over a wide range of FFs and an anthropometric test panel was not utilized. It was assumed that if any sampling bias due to conditions beneath the respirator facepiece was present/19) it was consistent between respirators. The observation that the log FFs for the silicone rubber, facepiece were not normally distributed contradicts the accepted assumption that FFs are lognormally distributed. If that assumption were accepted in this study, parametric t-tests would have found no significant difference between the mean log FFs for the two facepieces (p = 0.2150); an outcome that is opposite of that found with the nonpara metric tests used in this study. This would indicate that distributions of FFs should be tested before comparing the fit provided by different respirators for a group of subjects. Although the silicone rubber facepiece provided higher mean and median FFs, it was noted that the natural rubber facepiece provided higher maximum and minimum FF values. In addition, variation in fit for the silicone rubber facepiece was greater than that for the natural rubber facepiece. This would affea the calculation of FF lower or tolerance limits and should be considered in estimating the lowest likely fit provided by a respirator for an indi vidual or a group of wearers. In this case, if the distribution of mean log FFs for the silicone rubber facepiece had been normal, the 95 percent lower tolerance limit would have been lower than that for the natural rubber facepiece. Persons selecting respirators are cautioned to consider the variation of fit as well as the FF. References 1. Hyatt, EC,: Respirator Protection Factors. LASL Report la-6084-MS. Los Alamos Scientific Laboratory of the University of California, Los Alamos, NM 0976). 2. Hack, A; Hyatt, E.C.; Held. B.J.; etaL: Selection of Respirator Test Panels Representative of U.S. Adult Sizes. LASL Report LA-5488. Los Alamos Scientific Laboratory of the University of California, Los Ala mos, NM (1974). 3. Jones. JA; Brissette, J.E.; Holm. J.M.: Fit Factors of An Anthropomeirically Designed Three Size Half Mask. J. Int. Soc. Resp. Prot. 5:6-9 0987). 4. daRoza, RA; Cadena Fix, CA; Carlson, G.J.; etal.: Reproducibility of Respirator Fit Tests as Measured by Quantitative Fit Tests. Am. Ind. Hyg. Assoc. J. 44:788-794 (1983). 5. Leidel, NA: Performance of Face Seal Fit Tests for Respiratorv Pro tection Programs. Sc.D. Thesis. Harvard University, Boston, MA(1979). 6. Hyatt, E.C.; Pritchard, JA; Richards, C.P.: Respirator Efficiency Mea surements Using Quantitative DOP Man Tests. Am. Ind. Hyg. Assoc. J. 33:635-643 (1972). 7. Wtiieke, K.; Ayer, H E; Blanchard, J.D.: New Methods for Quantitative Respirator Fit Testing with Aerosols. Am. Ind. Hvg. Assoc. J. 42:121135 0981). 8. Emstberger, H.G.; Gall, R.B.; Turok, C.W.: Experiments Supporting the Use of Ambient Aerosols for Quantitative Respirator Fit Testing. Am. Ind. Hyg. Assoc. J. 49:613-619 0988). 9. daRoza, RA; Biermann, AH.; Sacken, C.R.; etal.: Fit Factor Measure ments with the Portacount and the Aerosol Photometer. Presented at the International Society for Respiratory Protection Conference, San Francisco, CA (November 6-9. 1989). 10. Hintze, J.L: SOLO [Computer Software J. BMDP Statistical Software. Inc. Los Angeles, CA (1989). 11. American National Standards Institute. American National Standard for Respiratory Protection, Respirator Fit Methods, Draft ANS1-Z88.10. ANSI. New York (1989). 12. Nairella, M.G.: Experimental Statistics, pp.17-1-17-6. NBS Handbook 91. United States Department of Commerce, National Bureau of Stan dards, Washington. DC (1966). 13. Martinez,J.; Inglewicz, B.: A Test for Departure from Normality Based --on a Biweight Estimator of Scale. Biometrica 68:331-333 (1981). 14 Dixon, S.w.; Nelson, TJ.: Workplace Protection Factors for Negative Pressure Half-Mask Facepiece Respirators. J. Int Soc Resp. Prot 2:347361 (1984). 15. Johnson, BA: Gender Differences Affecting Respirator Mask Sizing Systems. Presented at the American Society of Safety Engineers Profes sional Development Conference. Baltimore, MD (1987). 16. Oestenstad, R.K.; Dillon, H.K.; Perkins. LL: Distribution of Faceseal Leak Sites on a Half-Mask Respirator and Their Association with Facial Dimensions. Am. Ind. Hyg. Assoc. J. 51:285-290 (1990). 17. lenhart, S.W.; Campbell, D.L.: Assigned Protection Factors for Two Respirator Types Based Upon Workplace Performance Testing. Ann. Occup. Hyg. 28:173-182 (1984). 18. National Institute for Occupational Safety and Health: NIOSH Respi rator Decision Logic, p. 13. DHHS (NIOSH) Pub. No. 87-108. NIOSH. Cincinnati, OH (1987). 19. Myers. W.R.; Allender,J.; Plummer, R.; Stobbe, T.: Parameters that Bias the Measurement of Airborne Concentration Within a Respirator. Am. Ind. Hyg. Assoc. J. 47:106-114 (1986).' Received 9/1QA0; iwiew decision 10/30/90; revision 1/14/91; accepted 2/1/91 APPL OCCUP. ENVIRON. HYG. 6m SEPTEMBER 1991 VVV 000007926 789