Document ppNpbVjxpQNv9nmO9XGRmXndX

Interoffice Communication TO: Distribution Ifi&v JGt? ERT- M3J; AJO: RF XF:_____ FROM: DATE: SUBJ: T. G. Grumbles December 11, 1990 USE OF "PORTACOUNT" FOR QUANTITATIVE FIT TESTING VISTA The attached is a detailed study that further verifies the accuracy and utility of the Potacount device when compared to classical quantitative fit-testing methods. T. G. Grumbles dlj .225 Attachment SAFETY DIRECTORS Bruce Trego-Aber, J. D. Harris-Balt, Harry Peirce-Blane, Kathy Perez-Hmd, K. L. Fogg-LCCP, R. V. Gantz-LCLAB, G. M. Shirley-LCVCM, Brent White-Okc, R. B. Mar tin-Austin, J. R. Drumwright, Rick Quy VVV 000011530 A Comparison of Respirator Fit Factors Determined by Portable Condensation Nuclei Counting and Forward Light-Scattering Photometric Methods John C. Rose,*-0 Rledar K. Oestenstad, * and Vernon E. Rose6 ATexaco, Inc., P.O. Box 1404, 42nd Floor, Houston, Texas 77251, BUniversity of Alabama at Birmingham. School of Public Health, UAB Station, Birmingham, Alabama 35294 The purpose of this squiy was ro compare fit factors determined by a recognized photometer quantitative fit test system (Model 264, Dynatech-Frontier Corp.) and a portable condensation nuclei counter respirator fit test instrument(Portacount, TSI, Inc.). The study was performed by conducting sequential fit tests with the two methods on human subjects wearing the same respirator at the same fitting. The fit factors obtained by the two methods were compared by several statistical tests. The results of these analyses indicated that there was good agreement in determining pass/fail at a critical fit factor. It was further concluded that fit factors for groups of wearers measured by the condensation nuclei counter instrument are comparable to those obtained by the pho tometer method. Rose, J.C.; Oestenstad, R.K.; Rose, V.E.: A Com parison of Respirator Fit Factors Determined by Portable Condensation Nuclei Counting and Forward Light-Scattering Photometric Methods. Appl. Occup. Environ. Hyg. 5:792-797; 1990. Introduction The quantitative measure of a respirator fit is generally expressed as a tit factor. This factor is an indication of the effectiveness of the respirator faceseal. It is defined as the ratio of the concentration of die contaminant outside the respirator (C,,) to the concentration of the contaminant inside the respirator (Cj), orU) Fit Factor = Cq/C, (1) All measures of respirator performance are based on this ratio, with limitations and conditions that may be imposed on the measured concentrations, particularly Q. When these concentrations are determined in the laboratory-, the ratio is expressed as a Ht factor. If the measurements are made in the workplace, the resulting value is called a protection factor or workplace protection factor. In quantitative Ht testing (QNFT) of respirators w ith aero sols, Ht factor determination can he affected by a number of factors, including panicle size distribution and mea surement methods.121 It has been demonstrated that there is a large variation in fit for individuals wearing the same respirator as well as for a population of respirator wear ers.'In both cases, the variation was best described by a lognormal distribution. Sampling bias can also be present as a result of sample probe location, leak site, and the interaction between leak site and breathing patteras.U) Studies have also indicated that laboratory Ht factors do not correlate well with those determined in the work place.1 >Despite these problems, QNFT can be an effec tive t(x>I in selecting the specific respirator that provides the best Ht for a wearer.'-' The use of QNFT for testing respirators has been included in the Occupational Safetv and Health Administration (OSHA) comprehensive health standards for hazardous waste operations and emergency response, asbestos, lead, acrylonitrile, and formaldehyde.'8' A widely accepted QNFT method which uses an oil aero sol as a test agent and forward light-scattering photometry as a measurement method was hrM reported in 1972.<,;> Since its development, the method has been used exten sively in research and in industrial respirator programs. The aerosol size distribution, detection method, and test protocol have been incorporated into the American Na tional Standards Institute s 'Practices for Respirator Pro tection" (ANSI Z88.2-1980)/,0` This method requires the use of an exposure chamber and an accompanying aerosol generator. Therefore, the equipment is relatively expen sive and must be set up and operated at a fixed location. As an alternative to this method, a portable QNFT in strument has Iwen developed. This light-weight, batteryoperated instrument uses a constant flow condensation nuclei counter (CNC) as a sensor. It alternately measures aerosol concentrations from inside and outside the res pirator by means of a solenoid valve. The instrument cal culates the fit factor from average particle counts during VVV 000011531 792 1M?-322X/30>'05n-792S2.00/6 A!K APPL OCCUP. ENVIRON. HYG. S<W - NOVEMBER 1990 90-second sampling intervals. This method allows the use of particles in the size range of about 0 2-1 micrometer < ftm) in room air to he used as a test agent, thus eliminating the need for an aerosol generator and test chamber. There fore. tile CNC instrument can he used to perform QNFT m the workplace or at remote locations such as hazardous waste sites. One of the initial studies involving a CNC for hi testing was conducted in 1981.'Simultaneous measurement of respirator ht was conducted five times on two subjects using a continuous flow laboratory CNC instrument and a conventional light-scattering photometer. The subjects were exposed to an aerosol generated by the fit testing system. The researchers found that the fit factors determined by the photometer method were 1.64 times that for fit factors measured by the CNC. However, the correlation coefficient for the percent leakage measured by the two methods was 0.967. Respirator fit was also measured with the CNC alone using a variety of different aerosols, including ambient rcKim panicles. The researchers concluded that the CNC compared well with the photometer technique and that unv fine particle cloud could he used as a test agent. A study which used a portable CNC for fit testing was reported in I98"',n Comparative testing of the CNC and the standard photometer method was conducted in two stages. In the first comparison, the fit factors determined by the rwo instruments had correlation coefficients of 0.9"60.998 when sampling simultaneously in respirators with controlled leaks and mounted on mannequin head forms. Tile second comparison in that study*111 involved se quential fit tests with the two methtxls on 100 human subjects wearing full facepiece respirators. The authors did m>t report direct correlation of these tests because of the effects of nonoveriapping sampling periods and the nonreproducible nature of leaks which occur at the faceseal on the measurement of lit factors. However, the research ers did find that the rwo methods found 96 percent agree ment in determining pass/fail fit at a critical fit factor of 666?. Another si udv a impartng respirator lit with ambient aer< >sols and the conventional photometer method was re ported in 1988 The instrument used to measure fit factor with ambient aerosols was a laboratory panicle counter, both methods employed forward light-scattering photometers as sensors: however, the higher sensitivity of the particle counter allowed it to use room aerosols as a test agent. The methods were compared hv sampling inside facepieces with controlled leaks and by sequential sam pling of subjects wearing half-mask and full facepiece respirators. The controlled leak comparison found significant dif ferences between the two methods for 0.0075 and 0.55 percent leak levels but not at a 2.75 percent leak level on * half-face mask. The full-face mask analysis showed no gnilicant difference for leak levels between 0317 and 1.88 percent. While the difference of fit factors among subjects was significant, there was no significant difference hetween the means of five fit tests on 30 subjects deter mined by the two methods. Tlie purjiose of the present study was to make a direct comparison of Hi factors determined by the portable CNC instrument and the conventional photometer method rather than to determine the accuracy of lit factors determined by two methods of QNFT. This was accomplished by per forming sequential measurements of fit factor with the two methods at die same fitting of the respirator. Since the measured fit factors were paired data, any difference in the fit factor could be attributed to the method. This design w ould also include the effects of nonoveriapping sampling periods and nonreproducibie nature of leaks that occur at the faceseal during the measurement of fit factors. It was intended to determine if fit factors obtained in the field w ith the CNC instrument could be expected to be similar to those obtained in the laboratory with the photometer method. Materials and Methods VVV 000011532 The rwo instruments for determining fit factors com pared in this study were the Mode! 264 polvdispersed aerosol/photometer test system f Dynatech-Frontier Corp., Albuquerque, New Mexico) and the newly developed Portacount CNC Respirator Fit Tester (TSI, Inc. St. Paul, Min nesota). The photometer system consisted of an aerosol generator, exposure chamber, and forward light-scattering photometer sensor. The mechanical generator produced a polydisperse oil aerosol with an aerodynamic mass me dian diameter between 0.5 and 0.7 g.m with a geometric standard deviation of about 1.8. The aerosol was mixed with filtered air to produce an aerosol concentration in the range of 20-25 mg/m-* in the test chamber. Measure ment of the aerosol in tlie test chamber and inside the respirator was by forward light-scattering photometry. The CNC is a compact, portable condensation nuclei counter, in this instrument, sampled air was first saturated with alcohol vapor while passing through a heated satu ration mix*. The air was then cooled in a condenser tube and the alcohol vapor condensed on panicles causing them to increase in si/.e so they would scatter light as they passed through a focused light beam.'1,1 Scattered light was fo cused onto a photodiode which in turn generated an elec trical pulse for each aerosol particle that passed through the viewing volume. The CNC output only indicated the number of panicles detected and did not account for dif fering size or mass of those panicles. This technology allowed the device to count particles as small as 0 02 p.m over a concentration range of 0.1 to 5 x 10^ particles/ cm\l 1,11 This measurement range allowed the determina tion of fit factors greater than 100,000. The respirator used in this study was a 3M Easi-Air (3M Corp., St. Raul, Minnesota) silicone half-mask respirator equipped with high efficiency filters. Twc sizes were used: small/medium and mediunvlarge. The respirators were cleaned and sanitized after each use to ensure sanitary conditions. The study was conducted by fitting 24 subjects with a respirator and then performing sequential quantitative tit tests, first in nx>m air with the CNC instrument and then tn the aerosol test chamber with the photometer method During both tests, the Mihtects performed the following exercises: normal breathing, deep breathing. moving head side to side, niov mg head up and down, talking.and normal breathing, ('are was taken not to disturb the fit of the respirator from live time of the initial fitting until after the conclusion of both fit tests. Both instruments were operated in accordance with the manufacturers' operating instructions. The tests were con ducted in separate locations so as not to expose the CNC instrument to the corn oil aerosol from the photometer method. The CNC instrument measured three fit factors for each test exercise. The exercise tit factor was calculated as the average of the three values. The overall fit factor (KFI for the CNC test was then calculated by the following equation.11M Overall FF = 6/[(t/NBlFF) + (1/DBFf) + (PSSFF) + (1/UDff> + (1/TKff) + (1/NB2ff)J (2) The photometer tests consisted of three 30-second in tegrated photometer values while sampling inside the res pirator (C,) during each exercise and integrated photom eter values while sampling in the test chamlter (Cu) between exercises. Fit factors for each exercise were calculated bv dividing the average of the chamber values prior to and after that exercise by the average value of the three mea surements inside the respirator during the exercise. Hie overall fit factor for the test was then calculated by Equation 2 (i s i it should be noted that both methods determined in tegrated values of aerosol concentration rather than jx*ak values i'stng integrated output while sampling inside the l.acpicee mav overestimate fit factors hv about 10-13 ;x-r .cut vine to sampling exhaled air of the wearer.'-'''1 Mea sured tit factors were not corrected for this Ih.ls. Data were entered into a computer spreadsheet, and appropriate transformations and statistical tests were per formed using True Epistat Statistical Software.1 '*1 Two-Luled tests of hypothesis were performed at an alpha value of 0.03. With 2-4 pairs of measured ht factors, the study could detea a difference of 120 between the mean fit factors of the two methods at an alpha value of 0.05 and a power of 90 percent.* ^ This assumes a geometric standard devia tion of 2 26 for mean difference between fit factors pre viously reported.'l~' Results Exercise ht factors are show n m Tables i and II, and the paired overall fit factors are shown in Table III The dis tributions of the fit factors for the two methods were found to be significantly different from the normal distribu tion.1 1" The natural logarithm transformed data were found not to be significantly different from the normal distri bution11'*1 and were used for subsequent parametric sta tistical analysis. The range of measured ht factors by the CNC instrument method was 61-15.560 with a geometric mean of 188-4 and a geometric standard deviation of 3-9. The range of fit factors measured by the photometer method was 30-52,130 w ith a geometric mean of 2092 and geometric standard deviation of (v". However, an inspection of the data m TABLE I. CNC Instrument Exercise Fit Factors Subject Number i 2 3 4 5 6 3< 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Normal Breathing 245 34823 63167 3390 3210 2667 5i2 869 2993 3969 1823 3357 3? 4590 4450 1060 2740 332 3103 14577 3980 3284 33027 3627 Deep Breathing 163 37567 44100 2533 6170 3113 351 1847 4600 3041 969 2927 74 4200 3347 1797 2817 867 5333 12447 22200 6903 1337 2243 Side to Side 177 85067 25795 1797 4750 2917 281 1620 1663 1391 1897 2977 61 3780 4120 3673 3293 1203 7683 11903 5500 3933 3033 2440 Up and Down 177 77167 52256 2580 6867 3720 315 3080 1863 84 3563 3310 63 2717 4880 3553 2843 116 7573 23600 4350 12433 2833 2897 Talking 585 3427 2507 1667 4210 2847 359 1250 1409 124 1029 1960 111 2137 1873 1713 1204 258 1913 6653 4463 1371 1363 2103 Normal Breathing 610 72767 82533 2857 3423 3297 301 2930 2930 99 4053 3210 60 2293 2413 2053 1303 221 3879 4627 4510 13867 2250 2180 wv 000011533 794 APPL 0CCUP FWWffON HYG 5iW ,nnri- TABLE II. Photometer Instrument Exercise Fit Factors Subject Number i 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Normal Breathing 296 36831 98749 1427 4680 3369 243 1514 5494 516 3980 2739 23 2085 10826 59254 9155 42 1368 12423 3978 7885 900 2020 Deep Breathing 312 36738 83425 1514 4538 3414 330 1544 3286 52 4789 3491 27 3007 6120 47368 10571 81 2968 4309 1871 60485 1122 1815 Side to Side 221 92161 10)669 1311 1605 4236 217 1150 2824 109 4394 2674 24 1144 5497 85367 17443 32 2768 26668 927 26485 2123 2348 Up and Down 398 64676 29438 2018 1529 3084 201 1634 1602 146 4904 3806 44 6152 16662 8971 7761 31 1627 7496 1451 20449 2369 2013 Talking '61 44&80 '.2362 1223 4696 3932 371 1735 6414 125 1413 1765 31 1394 5256 97193 14480 37 1461 12797 855 31293 2147 2274 Normal Breathing 637 73938 63214 1398 2826 3587 230 1266 2422 136 4775 2506 51 2492 4499 86783 7715 36 3001 1831 2049 17998 2092 1883 Table 111 indicates that lit factors for photometers were lower than for CNO in W of the 2-i pairs. A linear regression mode! with the photometer log fit factors as the independent variable and the CNC log fit factors as the dependent variable yielded a slope of 0.600 which was significantly different from a slope of zero (p < 0.0001).'lHl A plot of the regression with 9s percent confidence intervals for the curve is shown in figure 1 The regression equation accounted for "3.9 percent of the variation in the dependent variable. Pearson's Correlation coefficient for the log fit (actors by the two methods was 0.839 and was significantly different from zero tp < 0.001 ).< Student s i-tesi for paired samples found that the mean difference between log lit factors determined by the two methods was 0(M3h. which was not significantly different from zero (p = 0 627V 1,1 The hi factor data were also compared by the 'X'ilcoxon Signed Rank Test.<Ul Thai test found the median difference between fit factors obtained hv the two methods not to lx* significantly different from zero (p = 0.992 J. The two methods were also compared as to their agree ment with regard to passing or failing a fit test rather than a direct comparison of fit factor values. Critical fit factor values of 100 and 1000 lor pass/fail criteria and fit factors determined by the photometer as the standard were used for this comparison. These results were entered in a 2 x 2 table, and sensitivity, specificity, and predictive values were calculated. These results are shown in Table IV It is noted that there was only one disconcordant pair in the 48 comparisons at the two critical fit factors. However, the statistical significance of these distributions cannot be tested because .some cells have less than live observations.1 H) Discussion Results of this study indicate that fit factors determined by the CNC instrument and the photometer method, on a TABLE III. Overall Fit Factors Subject CNC Instrument Number Measured Log 1 241 5486 2 15560 9 652 3 11820 9 378 4 2320 7750 5 4410 8392 6 3060 6.025 7 340 5.829 8 1600 7 378 9 2400 7 782 10 191 5.254 11 1660 7.417 12 2860 7.958 13 61 4 110 14 3010 fl.QtO 15 3120 8.047 16 1920 7.560 17 2020 7 610 18 272 5.608 19 3860 8.260 20 9350 9 143 21 5200 8 557 22 3780 8.238 23 2270 7 729 24 2490 7.820 GM* 1884 GSD 3.9 *GM = geomemc mean. "GSO = geomeinc standard deviation. Photometer Method Measured Log 364 52130 36890 1450 2630 3560 252 1440 2950 114 3320 2660 30 2020 6590 32790 10190 39 1960 5420 1420 18880 1560 2040 5896 10862 10516 7278 7.876 8179 5.531 7275 7.900 4.738 8.107 7.887 3.417 7.612 8.794 10.398 9.229 3610 7.581 8.597 7.262 9.846 7.354 7.621 2092 i^raooov^ Log Photometer FF FIGURE 1. Regression ol log photometer lit lector (FF) and log CMC fit factors. group basis, are highly comparable. The results are in agreement with those reported in other studies.'"'1 1 -1-'The range of Hi factors obtained by both methods were ex tremely large and followed a lognormal distribution. These results correspond with those of a previous study on var iability of respirator fit."' The geometric standard devia tion of 3.9 for the CNC instrument Ht factors is almost the same as that found for half-mask respirators in the previous study. However, the geometric standard deviation of 6.7 for the photometer method was higher than those previ ously reported. The difference in variance between the two methods in this study was found not to be significantly different when tested by Bartlett's Test for Homogeneity of Variance (p = 0.091)/ Ui The correlation coefficient for log fit factors by the two methods in testing 24 human subjects determined in this study was 0.859 One previous stud)' found a correlation coefficient of 0.96-7 from repeated, simultaneous tests on two subjects/"' and another found coefficients of 0.9760.998 from testing respirators with controlled leaks and mounted on mannequin headforms/n 1 The slightly lower value obtained in this study may be due to intersubject variation in leakage on a large sample of human subjects uud'or die effects of nonoverlapping sampling periods and nonreproducthle nature of faceseal leaks."' However, the coefficient obtained in this study was still not significantly different from 1.00. The results of the regression analysis of the log fit factors from this stud) were also very similar to results reported in previous work.1 11 ' Those studies found slopes of 0.609 and 0.8I6-0.9',2 and correlation coefficients of 0.967 and 0.953-0 996, respectively, from simultaneous sampling bv the two methods. The regression slope in this study was 0.600 with a correlation coefficient of 0 859. Again, the fact TABLE IV. Pass/fail Agreement Photometer FF' >100 FF < 100 --- -- Photometer ff > 1000i FF < 1000 FF > 100 CNC FF < 100 22 0 t FF > 1000 CNC 1 FF < 1000 19 0 0 5 Sensitivity: A/(A C) Specificity: D/(B + D) Positive Prediclive Value: MA + 81 Negative Predictive Value: D/(C + D) FF = 100 100 0.50 0.96 1.00 FF = 1000 100 1.00 100 100 Sensitivity: The probability that the CNC FF will be greater than the critical value given (hat the photometer FF is greater than that value. Specilicity The probability that the CNC FF will be less than the critical FF given that the photometer FF is less than that value. Positive Predictive Value: The probability that a CNC FF greater than the critical value will reflect a photometer FF greater than that value. Negative Predictive Value: The probability that a CNC FF less than the critical value will retied a photometer FF less that value. 'FF = tii lactw VVV 000011535 796 APPL 0CCUP. ENVIftON. HVG. smt NOVEMBER 1990 that the regression equation accounted tor '3 9 percent of the variation in (he dependem variable is probably due to interMibject variation in leakage antL-or the ettects < >f nonoverlapping sampling periods1" The comparahilitv ot the two methods is further supported bv results of the Student's t-test and VC'ilcoxon Sitn Rank Test. The west found that the mean difference be tween the lug fit factors for the two methods was not significantly different front zero, and the Sign Rank Test found the median difference between fit factors not to be significantly different from zero. These results are similar to those reported in a siudv of repeated measures of the same fit by a photometric method and a laboratory CNC.( 121 Finally, the two methods provided very good agreement in determining fit factor pas.vfai! criteria. Only one disconcordant pair was found in the set of 48 comparisons. This resulted in a specificity of 0.50 at a critical fit factor of 100 because of the small number of observations in the cells used to calculate that value. However, the two meth ods were found to be in complete agreement at a critical fit factor of 1000. These results are very similar to those found while testing full facepiece respirators with the two methods.'111 The study assumed any bias present as a result of facepiece sampling conditions would be the same during test ing bv the two techniques. Other possible sources of bias were the consistent sequence of test methods and facial exercises during the tests. These differences could be elim inated by randomizing the sequence of test methods and of exercises in the test. Also, simultaneous, as opposed to sequential, sampling of the two techniques could be considered Conclusions The CNC instrument gave very similar results when tested in sequence w ith the photometer method. Fit factors de termined by the two methods, compared on a group basis, were highly correlated. Both parametric and nonparametric statistical analysis of the data indicated no significant difference between the two methods, and the two methods were in gtxid agreement in determining pasx/fai! at two critical fit factors Differences in fit factors measured by the two methods could be attributed to intra- and intersubject variation in leakage on a large sample of human subjects and nonoverlapping sampling periods. It wus concluded that fit factors measured by the CNC instrument for the group of subjects are comparable to those measured bv the photometer method. This would mean fit factors obtained for groups of wearers by users of the CNC instrument in the workplace or at remote lo cations could be expected to be equivalent to those mea sured hv a photometer method at a laboratory location. However, caution should be exercised when comparing tit factors on an individual basis. In applying the results of QNFT, the practicing industrial hygienist should keep in mind the purpose ol these tests. QNFT is {XMformed to assist in selecting the respirator that provides the highest .imouni of protection for an individual wearer and to as sure that an appropriate lower confidence limit of the measured fit factors exceeds the assigned 111 factor for tile type of respirator being used. This studv indicates that tile CNC instrument will provide similar results to the pho tometer in the performance of these tasks. Acknowledgments Tiie authors appreciate the assistance with statistical analysis by Dr. Laura L. Perkins of the Epidemiology De partment, 1'niversity of Alabama at Birmingham, School of Public Health. This study was supported in part bv NIOSH Educational Resource Center grant 2T15 OHO?210-06. References ]. Nhers. Vi R.: Lenh.in. s W . Caniplx*ll. l). Provost, G.: Tlic Forum Am I mi H>n Asms- I -o B2x~B26 < t`)KS). 2. Holloa, I'M . VI illekc. k . The El tea of Aerosol Size Distribution and Measurement on Respirator Fit Am. Ind Hyg. Assoc. J. 4 8SS-K60 t19H"l 3 daKoza. K A.. Cudena-Fix. C.A.. Carlson. G.I., et al: Keproducihilib of Kcspiralor Fit Tests as Measured by Quantitative Fit Tests. Am Ind. Mvg Assoc J ~9-t < 19y I -i M\ers, WR. AHender, I. Plummer. K; Stobbe. T Parameters Thai Bias die Measurement of Airborne Concentrations Within a Respi rator .Am. Ind. H\ii Asmx. -TdtXi-t U U9H9V s Dixon. s.W , Nelson. T I Workplace Protection Factors tor Negative Pressure. Halt Mask Facepiece Respirators j. ini Nx; Rcspir Frol 2:.3-T-?bl < IVHh ). tv Miers, W.R., Peach. M l: Cutright, K., Iskander. VC Workplace Pro tection Factor Measurements on Powered Air-Purifi in> Respirators ai a Secondary Lead snu-lter. Results and Discussion .Am Ind Hvj> Assoc. I. -is (sH I -<>SK I 1984) Willeke. K.: Aver. H F^: Blanchard, I D New Methods lor Quantitaiive Respirator Fit Texting with Aerosols Am. Ind. Hvg. Asmx I. 42.121-1 .AS (19HI I * General Indusirx suind.irds. Oxie ol Federal Regulations Title 29, Fan 1910, Subpan 7.. pp <>~>-93~ Is Government Printing OH ice. VC asli- inginn, IX't pWi 9 Hvatt, F. C.; Priich.ird, IA. Richards, CP Respirator Ffliciencs Mea surements l'sing Quanmativc OOP Man Tests Am Ind Hvg. Asmx J. 3.3:63S-h43 (19-2). 10 American Nanonal Standards Insiitule: American National Standard tor Respirator. Protection. /.KX2-19H0 ANSI. New York (19S0) 11 Live. R. hvaliut'ion ol a .Minialun/.cd Condens.uion Nucleus C< miner lor Measurement of Respirator Fit Factors i hit Six Rcspir Prot S:J-" yiulv/SeptemlxT ]98") 12 F.rnxiherger, I I G . Gall, R B; Turok, C.W. Experiments Supporting the I'se of Ambient Aerosols for Quantitative Respirator Fit Testing. Am. Ind Hvg Ax.mx\.| 49:613-619 (1988) 1.3. American National Standards Institute: American National Standard for Respirator.- Protection. Respirator Fit Methods: Draft ANSFZ88. 10 ANSI. Neve York l 1989> Is Gustafson. T.L: True Epislat, 2nd ed. Epistai Services. Richardson. TX f 198"') I s. Luchin. J.M . Introduction to Sample Size Determination and Power .Analysis for Clinical Trials. Control. Clin. Trials 2.93-113 < 1981). Received 6/16/89. review decision 11/27/89; revision 7/2/90. accepted 7/16/90 APPL 0CCUP. ENVIRON. HYG. SOU NOVEMBER 1990 000011536 797