Document oMB2okONvrRpDokGg3KRZEYQo

Measurement of Dust Exposures in the Asbestos Textile Industry JfEREMlAH R. LYNCH and HOWARD E AYER pivirio of Occupational Health, U. S, Public Health Service, 1014 Broadway, Cincinnati, Ohio ^ Data obtained from environmental surveys of nine asbestos textile mills, which represent the baseline for the textile segment of the Public Health Service epi demiological study of asbestos processing industries, are presented. From these data concentration ranges are derived which yield significant differences between typical sample groups. Variance ratio tests of different methods of counting and analysis were made and count weight ratios based on magnesium analyses for asbestos were calculated. Induction characterizing the exposure had to be studied THE DIVISION- of Occupational Health simultaneously. In order to accomplish this, of the U.8, Public Health Service has a large number of samples were required Igen conducting an epidemiological study (Table I). 4 the asbestos products industry since Janu0j of 1964, This study was prompted by gsr need to refine and update earlier work done in these industries in relation to asbstosis1-1 and to appraise the association of mbestos with other health hazards suggested 'Hj a number of investigators.*"* The proOBClures and objectives of this study have |een discussed in earlier papers*'* and the relationships between impinger counts and fter concentrations on membrane filters have The analysis of these data by manual methods would have involved an excessive burden and, consequently, the major por tion of the data were transcribed onto punch cards for computer processing. In addition to facilitating routine calculations of dust concentrations, computer processing made possible detailed statistical analyses which would otherwise have been prohibitively time-consuming. been presented.* The environmental field study phase com menced with an asbestos textile plant in January, 1964, and the last of the textile plants to be included in the study was sur veyed in June, 1965, thus completing the nbestos textile baseline. The nine plants wrveyed account for over 80% of the work- This paper summarizes the baseline as bestos textile dust count data and gives some observations on sampling methods and strategy and count-weight relationships. Oth er data, resulting from analyses for poly nuclear aromatic hydrocarbons, trace metals, etc., will be included in later papers. in in the industry and provided a cohort of over 2500 employees exposed to asbestos dust in varying degrees. The conduct of this study was somewhat more complex than previous studies of dusty radcF by the Public Health Service in that both the exposure itself and methods of Tablei Nymbers of Samples Collected in Asbestos Textile Plants Membrane Filter Impinger Breathing zone Person*! Genera! *ir Total 603 310 478 1,396 60S 2 508 1,115 431 432 Table ii Mean Oust Concentration by Plant and Operation Impinger. mppcf Operation Plant aa C D F Fiber ppf& H2 Carding AWhSTiapwitn*ivnidtinitnniinngggg P' J-J 1-5 ?? }0> .7* J 3 2.3 1.-3? 20.59 11 ** 2.6 3^.'47 2.5 02.40 a-J 2.5 3*.0* 5.4 31.77 0 A 3- o.l J0--65 * 4 SO 'f :* 10..7? 32..25 00.63 {.jij i.r 3.9 t.f 9.4 f3.0 0.4 044 ,`:J ?;f II Ali< 23 it ?-? Lo f;| The <<*1717---------------- --LLJLi: s'= 0-3 1.3 h **n for an September-,clob,>er. Table rv Ratios of Fiber CumS by teeth- bnptfigtr Flbt" MieromcQp* >3^ Electron Micro- cop# Ts?*j P jFoibtaeJr* Ooo, n Dost Concentrations pinger samples. That fibers were not mm Tables II and III present the mean dust concentrations by plant and principal opera tion in all the plants surveyed. In all cases, the median was also calculated, and was was not because they wens absent, bat ^ cause the 100X light field counting used with the impinger did not reveal them. The ratios of fiber counts by different rneth- ods are shown In Table IV. usually in close agreement with the mean. The impinger concentrations include both The ratios given in Table IV were ob grains and fibers, and the counts were made in accordance with the American Confer ence of Governmental Industrial Hygienists (AGGIH) procedure. Thus, these values tained by dividing the concentration of fibers determined by the method given in the head ing, by the fiber concentration as determined bv the method listed at the side for a large may be compared to the existing Threshold number of the same or comparable samples, Limit Value of 5 million particles per cubic For example, the total fiber concentration foot (mppcfl. The membrane filter (MF) as determined by optical membrane filter samples were counted using a procedure de counts at 430X phase contrast was eight scribed in detail by Edwards.1* times that found on comparable impinger Early workers in the study of asbestos ex* samples counted at 100X light field. TbW* posure elected to count grains as well as ratios are not necessarily meant to indicate fibers since too few fibers were seen in im any degree of equivalence, but only to show the relative magnitude of the numbers of Ope rati Table in Mean Dust Concentration by Jiam w cv. Membrane Fiber CperaUotI Plant EF Si Afl Note; A tal libera; B = 5 ft libers; ~ ' ^0 fi fibers <mb^ocl0bei> : iv 1unts by Method ooo( I) fibers were not Men vere absent, but beeld counting method did not reveal them, is by different meth- IV, Table IV were ohmcentration of fibers id given in the headration as determined the side for a large ,-omparabk samples, fiber concentration ;al membrane filter contrast was eight imparable impinger >i light field. These v meant to indicate .e, but only to show of the numbers of All 1. 8 26 21 .2 7.6 1.2 i.i 144.,29 2.Q 7.0 3.9 3.7 \4 12,3 11 ..31 63 .2 1l ,l 14 6 ), 7 4 .6 9/ 34 ) 9 3.4 9 8 0 . 5 3.6 ).? 2.5 9 12.5 5.6 35 Jnau.:--std Hygiene Association Journal 433 Table v Sigr.i.jiictly Different Dust ration Groups 00F rr.?- " Or*!"* X Fib*'* ? 2 D""- 8 2 Membran* Filter ___ fkbtr*/cc____________ Total $ fi x so Fiber* Fiber* Fiber* i- $ 5-20 20-ft 0 .7- 3 3.0-12 12 .0--48 .3-- 2-5 S.S- 6.0 6.0--24.0 gbers detected h' the different methods. The ,,tios enclosed : parentheses wen* obtained ^ indirect com Arisons since electron microgpraphs of imp: ;ger samples were not pos- jjblo. |t is evident that most of the fibers, even j^jsc longer thin 3 /*, are not counted by ^jther the imptrarer or the optical membrane gfer procedure, probably because they are fpo thin to be detected by light microscopy. However, enough fibers are found on memforane filters a: 430X phase contrast to be a statistically useful index of the number of fibers present. Since it is probable that the pgnificant biological property of asbestos ii its morphology, only fiber counts on mem brane filters are presented. Regarding the interpretation of these data, tome caution must be used to avoid assign ing excessive significance to small differences. Some method of categorizing relative dusti* ness is needed for eventual correlation with health data to arrive at safe levels of ex posure. Given the inherent variability of the environment and the variability in dust counting, broad ranges or classes of dust concentrations will be required. From the standard deviations calculated for each plant and operation group of samples, it was de termined that a four fold difference in con centration was required to obtain a signi ficance level of p=0.05 on typical sample sets. Based on this, the data fell rather conveniently into three arbitrary groups such that the probability of any value in a group not being significantly different from the middle value of an adjacent group does not exceed p = 0.05 (see Table VI. The ratios between corresponding groups by different count methods are based on the equivalence between impinger and filter count The groups are designated by number rather than low, medium and high, to avoid the implication that they can be compared with any existing standard. Using the con centration ranges given in Table V, a chart of the group numbers by plant and opera tion was prepared (Figure 1). The upper left triangle contains the group number as determined by impinger and the lower right by membrane filter total fiber counts. In 759c of the cases, the two methods lead to the same estimate of relative dustiness, and in cases where the groups are different, the differences never exceed one group and show no consistent trend. Since our concern is with exposure, or body burden rather than with dust concen trations, a different grouping may be de rived by the method of Roach.'1 It can be shown that: -C--o--e-f-,--o--f--V--a--r-.--o--f--B--o--d-yU--B--u-r-d--e-n- <. 0.S9/{t-\) V> Coef. of Var. of Concentration " MY where t is the duration of the samples used in determining the concentration and T is the biological half-time of the contaminant. Since t never exceeded two hours and T is greater than six* months, the ratio of co efficients of variation cannot be greater than 0,026, If the Coefficient of Variation of Body Burden obtained in this manner is used to estimate the standard error of the differ ence, ranges of concentration can be cal culated sudh that the body burdens caused by sample means falling within one range are significantly different (p<0.0l) from those caused by concentrations at the mid point of an adjacent range (Table VI). The same procedure illustrated by Figure 1 was applied using the body burden groups. As would be expected from the narrower limits of the classes, the agreement was not as good, with only 50% of the sets of samples yielding the same group by both methods. However, in less than 10% of the sets did the appraisal of dustiness differ by more than one group, in no case by more than two groups, and again these differences showed no consistent trend. The general agreement between the groups gives one confidence that both methods are I f i r i 434 Sfpiember-October, IS ^ican Industrial Hygi< d that membrane ''0F,0Vcontrast might be dfTehhjtot f1ifell)d1fib1imepr icnog-uenrtscomuing total fiber counts sinl ^uld be more visible. To investigate this matt gf samples on which magr jjbestos were performed | later in this paper1 were a fiber counts made on t Since the precision of gnalvses {<5# error) i pared with dust counts, i' die variance ratio test wt ability in the asbestos cc mined by magnesium anal not the case, and ail between standard deviatil Figure 1. Comparison of results by concentration groups ficant. From this outco jumed that the sample pendent estimates of the variance. Thus, the vail measuring approximately the same thing in plant than for the it asbestos textile plants. However, if in the oat plants inc actually the variability c that is, the variability oil f case of either the body burden or concentra effort made toward^ group of workers doing I tion method of grouping, the membrane filter total fiber count arrived at exactly the same decision as to relative dustiness as the im- a plant is more of * . iiw>: .. the difference between -optmtttipm. job at different times an ability in dust counting, not be significant whei pinger count, there vvould be no basis for Sampling Methods and Strategy actual variability of the distinguishing between these dust measure ments by their relevance to disease param All of the usual and some of the less fre eters. Both would correlate, or not correlate, equally well with medical findings. Thus, the larger range of disagreement between methods in the body burden groups may be useful in establishing the relevance of the methods. By either the cooatiHmtien r bn4y biwien method, the result* uniform for different quently used methods of sampling for min eral dusts in industrial environments were used in this study, and the large numbers of samples of each kind permitted com parison of these methods. One question which could be answered from these data is the relative variability of different count methods of analysis. To ex amine this factor, the variance ratio test11 was applied to the sample groups by different Count-Weight Relations! Chrysotile asbestos, over 90fiy of the asbest try. is a hydrated ma; has a relatively constan of about 25^-f. With th veloped atomic absorpi possible to analyze foi sensitivity in the ordel Table vi Significantly Different Dust Concentration Groups count methods in each plant and operation. These methods included impinger grains and fibers versus membrane filter total fibers, total (pgm). Consequently may be performed samples which are lig Body Burden Method fibers versus >5 p fibers and total fibers and this magnesium n Group Imp inc^r mppcf Grain* & Fiber* Membrane Filter __ __ tiber*/cc__________ Total > 5g > 10 p Fiber* Fiber* Fiber* versus > 10 (t fibers. In all cases, the variance ratio test indicated that the difference be tween the standard deviations by the differ weight of asbestos pres< the count and weight a direct count-weight i 1 0,1-0.2 1- 2 .6-- 3.2 ,3- .6 ent methods was not significant (p>0,05i 2 3 0.2-0. 4 0.4-0.ft 2-- 4 4- B 1,2- 2.5 2.5-- 5.0 .6-- 1.2 1.2- 2.5 and that none of the methods could be con 4 5 0.8-1 .6 1.6--3.2 8-16 16-32 5.0-10.0 10.0-20.0 2.5- 5.0 5.0-10.0 sidered less variable than the others. This 6 3 .2-6.4 32-64 20.0-40.0 10.0--20.0 finding is interesting in that it had been Several potential soi mediately apparent fibrous parent mineral tile asbestos developed ''pttmbcT-Octobc lSl jjrrii'rtVan Industrial Hygiene Association Journal 435 operation in differ* tat the engineering dust control within rmining factor than perations. ategy une of the less fresampling for minenviranments were the large numbers d permitted com- :mld be answered itive variability of f analysis. To exi-iance ratio test14 roups by different nt and operation, pinger grains and r total fibers, total and total fibers ases, the variance ie difference he rns by the differficant (p>0.05) ds could be conthe others. This at it had been ^mjied that membrane filter counts under uase contras! might be less variable than light field impinger counts, and that long ,-,]() is fiber counts might be less variable ^ar, total fiber counts since the long fibers jjicmld be more visible. To investigate this matter further, groups pf samples on which magnesium analysis for j^bestos were performed {to be discussed later in this paper! were compared with total fiber counts made on the same samples. Since the precision of the magnesium gjudyses (<5% error) is very good, com pared with dust counts, it was assumed that (jje variance ratio test would show less vari ability in the asbestos concentrations deter mined by magnesium analysis. However, this was not the case, and again the differences .between standard deviations were not signi ficant. From this outcome, it may be mpamed that the sample variances are inde pendent estimates of the same environmental variance. Thus, the variability observed is actually the variability of the environment; that is. the variability of the exposures of a group of workers doing nominally the same job at different times and places. The vari ability in dust counting, large as it is,fl8y not be significant when compared to the actual variability of the workers' exposure. Count-Weight Relationships Chrysotile asbestos, whidk for over 90% of the asbestos txM& ttr try, is a hydrated magnesium silicate and has a relatively constant magnesium content of about 25%. With the aid of recently de veloped atomic absorption techniques, it is possible to analyze for magnesium with a sensitivity in the order of Q~i mirrr%TiM (ggm). Consequently, magnesium analyses may be performed on membrane filter samples which are light enough to count, and this magnesium may be related to the weight of asbestos present on the filter. From the count and weight on the same sample, a direct count-weight ratio may be made. Several potential sources of error are im mediately apparent. Serpentine, the nonfibrous parent mineral from which chryso tile asbestos developed, is usually present in varying amounts with the fibrous asbestos and will contribute some magnesium but will not contribute to the fiber count. A second source of error is the ubiquitousness of mag nesium and magnesium-bearing minerals. Magnesium is found in general air pollution, and significant amounts could be present in industrial dust from other magnesium-rich process materials. Little can be done to eliminate the first source of error, and a tendency to over estimate the mass of fibrous asbestos is un avoidable. Regarding the second source of error, air pollution survey data15 revealed that the usual atmospheric concentrations of magnesium are several orders of magnitude lower than in-plant concentrations. Other magnesium-bearing process materials can be eliminated as a source in textile plants since the other materials do not contain signifi cant amounts of magnesium. Membrane filters from which a small wedge had been removed for counting were sliced in half with a sharp blade against a 436 f<t<mbcr-October, l9aflELri(rt''an lnliust,ial Table vii um analysis as appnwitnatcly equivalent b 0^tes of relative di Count-Weight Ratios 1 irsppcf by impinge* I) should be einph*, samples were ii Type Count Total fiber* ' 5 Lt flbtrs ^ 10 )X fiber* Fiber* pet It ,Q0Q 5,000 3,000 to sized that this relatiomhip was developed b* a very indirect method. However, it M&ta 4tajt.. *Sbesfti* .iifflais, may whigher than made from imj jpncentration or bci groups _ * Variance ratio tesj A number of sets of simultaneous sample, j measurement ind: brass die. From analyses of magnesium on were taken with and without size selective vprianre seen 3S due these filter halves, the count-weight ratios presamplers. Thirty-minute samples on a pa;r in terms of weight of asbestos, for each type- of membrane filters, otic of which was prc. population and that ^punting exhibit ami count. were calculated for each of the 250 ceded by a horizontal cSutriator with a i,, terms of a lessenec samples analyzed. The frequency distribu 7.3 ft unit density sphere rut-off, were taken Count-weight ratios tions of these sets of ratios are given in in each operation in each plant. While not analyses of membr Figure 2. all of the magnesium analyses are completed fibers > 10 ji/pgm < Given the sources of error discussed above, preliminary results indicate that the respir and the natural variability of dust counts, a able fraction of asbestos as determined by this nr 11.000 total fibers nf ratios exhibited considerable spread in the count-weight ratios would be expected. This spread did occur, in fact, but a central tendency in each of the groups of data is apparent, and yield the approximate values shown in Table VII. eiutriation is approximately 50 to 60 G, Ad ditional samples were taken in each plant using two pairs of membrane filters run for five to eight hours, with cyclones conform ing to the Atomic Energy Commission cri teria attached to one pair of filters. From the others. A mass j 50 /is. obtained fronl vielded a count-weig J*gm which was of th rude as the ratio o analysis, Based on til It would appear from Figure 2 that the count-weight ratios of the longer fibers ex hibited less spread. However, the ratios were found to follow a log normal distribution with the same standard geometric deviation of 2.3 for all three type counts. Thus, each of the groups of ratios is equally disperse, these samples an estimate of respirable frac tion of from 30 to 40G was obtained. In both cases the figures given are approximate: -however, they do follow the usual perform ance of these presamplers. In that the cyclone passed a smaller fraction. If these respirable percentages are applied to the gross airborne borne mass concentrr lent to one mppcf of 1 mg/m\ and res tions. as determined clone, would be in tl nr1 and 0.3-0.4 mg/ Hygienic criteria s!| and no one type of count is to be preferred as an estimate of mass. An independent estimate of the countweight ratio was obtained from size distribu tions on electron micrographs. Since the number of large fibers visible on these slides is small, though their contribution to the weight is very large, it was necessary to poo) a number of electron micrographs of dif ferent samples, to obtain a statistically use asbestos by magnesium results, 1 mg/nri of the total dust would probably contain Q.5-0,6 mg/m3 of dust passing an elutriator and 0.3-0.4 mg/m3 passing a cyclone. Additional magnesium analyses are being performed so that mean mass concentrations by plant and operation can be compared with the groups established for impinger and membrane filter counts, to determine if mass concentrations yield the same relative hazard estimates. disease-producing me curate assessment of convenient to use. Tf venient as any other ( the large variance fo no Jess accurate the However, it is diffic\ ing of the hygienic mated by grains whi to be the causes of the counting of fibe1 ful size distribution. From these data a typi cal mass median fiber volume of 50 cubic Conclusions appears to have an industries where asbe1 microns (>3) was obtaineda The reeuks of environmental surveys txf tion of the airborne count-weight on a density baaed nine asbestos textile pjaatt rwod tilltt ^K-. asbestos. concentrafir*** as weswwf fry* impinger are magnesium analysis and the method invc f * This result is of the same order of magni generally below the Thi willlf fmwfr- VUG* tude as fiber ratios obtained from the mag Due to the natural variability of the environ nesium analyses. ment, categories of significantly different dust By applying the impinger-membrane filter measurements must be broad. Three signifi equivalents obtained earlier to the weight cant groups were obtained from concentra count ratios given in Table VII, it w'as tion data and six significant groups were de found that 3 mg/m3 of asbestos by magnesi rived from bod v burden considerations. Es- -.4iaturn Journal 437 cated and expensive laboratory equipment not generally available in state, local and plant hygiene laboratories. However, it is hoped that the intensive study of total and respirable mass concentrations, made pos sible by this method, will lead to the develop ment of a relevant and convenient mass sampling method. Acknowledgments The authors wish to acknowledge the valuable contribution made by Mr. Richard E. Kirtser who developed the method for magnesium analysis and by Mr. Thomas A. Brown, who performed a targe part of the dust Counting- References i. fan* , B- , By v AA i HBML t an*d4 Dhu-s^en. W. C., j. M. Dalla Vaixz, T. I, Eowarus, j. M. MsitER, ad R, R- Sayers; A Study of As- in ike Asbestos: Tcxtile industry. Public Health Bulletin No. 24!, L.S. Government Printing Office, Washington. IL C. (1938) _ ^....; _ /AaStiCTnFSt^T VifcT'. 1 $A! 5, KM m __ ,...... X**k & $5 fWftW L Colley, L- J.; Ob'sffui-n and Central Plan for Or* cupational Health Study of the Asbestos Products in- dustr%. Division of Occupational Health, Public Health Service. Department of Health, Education, and Wel fare (19621. 8, Lynch. J. R.: Asbestos Study--Procedures and Find- !$$. Transactions of the 27th Annual Meeting of the American Conference of Coiernmental Industrial Hy- gitnists (1965). 9. Ay*. H. E., G. H Epwaups, I, L Fanney. J*,. and J, R. Lynch: Relationship of Irapinser Counts to Fiber Concentrations by Membrane Filter in Asbestos Textile Plants. Delivered at the annual meetins; of the American Industrial Hyjfienc Association, Houston, Texas (1963). ID. Epwaro*. G. H.. and J. R. Lynch: The L`.S, Public HeaJth Service Method for Membrane Filter Enumera II. tRioonacoh.f Asbestos S. A.: A Dust. U npublished (19661. More Rational Basis for Air Sampling Programmes. Amer. Ind. Hy. Assoc. J, 27: 1 (Jan.- Feb. 1966: 32. Sn&decOr, Ct. W.: Statistical Methods, The Iowa Slate CoMre Press. ^ Ames. Iowa {19561. _ |3. Divjjiop of Air Pollution: Air Pollution Measurements of the National Air Sampling Network* 1953-1957. \J .5. Department of Health, Education, and Welfare, Public Health Service. Cincinnati, Ohio (19581, \