Document DKpq0EQnn10z75vbjBZ8BXeQ

University o f Texas Medical Branch IL L (TXUTGA/TMB) Delivery type: Regular Delivery method: Ariel Special Instructions: Prefer arid delivery... 192.68.30.164.. ro ho-si * IS) 00 ro 00 O) a> oo ooo 0o5 o oX < 5' <D X "O CD to to 7J Date: 6/22/2009 Call #: Location: MaxCost: $100.00 Delivery notes: Prefer ariel delivery... 192.68.30.164.. Fax: 713-790-7056 Ariel: 192.68.30.164 E-Mail: tmcariel@library.tmc.edu Journal Title: American journal of industrial medicine Volume: 4 Issue: 3 Pages: 421-33 Month/Year: 1983 Article Author: Dement JM;Harris RL;Symons MJ;Shy CM Article Title: Exposures and mortality among chrysotile asbestos Patron: Yana Posey TN;197732 ILL Number: 27288660 TN: 266502.0 Notice: This material may be protected by copyright law (Title 17 U.S. Code) American Journal of Industrial Medicine 4:421-433 (1983) Exposures and Mortality Among Chrysotile Asbestos Workers. Part II: Mortality John M. Dement, PhD, Robert L. Harris, Jr, PhD, Michael J. Symons, PhD, and Carl M. Shy, md, DrPH A retrospective cohort mortality study was conducted among a cohort of 1,261 white males employed one or more months in chrysotile asbestos textile operations and fol lowed between 1940 and 1975. Statistically significant excess mortality was observed for all causes combined (standardized mortality ratio [SMR] = 150), lung cancer (SMR = 135), diseases of the circulatory system (SMR = 125), nonmalignant respiratory diseases (SMR = 294), and accidents (SMR = 134). Using estimated fiber exposure levels in con junction with detailed worker job histories, exposure-response relationships were investi gated. Strong exposure-response relationships for lung cancer and asbestos related non malignant respiratory diseases were observed. Compared with data for chrysotile miners and millers, chrysotile textile workers were found to experience significantly greater lung cancer mortality at lower lifetime cumulative exposure levels. Factors such as differences in airborne fiber characteristics may partially account for the large differences in expo sure response between textile workers and miners and millers. Key words: asbestos, chrysotile, lung cancer, asbestosis, exposure-response INTRODUCTION The companion paper in this volume described the facility and presented meth ods used to reconstruct exposure levels for an asbestos textile operation using chryso tile. Using linear statistical models to account for textile processes and controls, worker exposures between 1930 and 1975 were estimated. These data were combined with an assessment of mortality to study exposure-response relationships for lung cancer and nonmalignant respiratory diseases. Each employee's detailed employment history pro vided the necessary link with the exposure estimates to allow analyses of exposure-re sponse. This manuscript presents the overall mortality assessment and the observed exposure-response relationships. University of North Carolina, School of Public Health, Chapel Hill. Address reprint requests to John M. Dement, National Institute of Environmental Health Sciences, P.O. Box 12233, Mail Drop 19-02, Research Triangle Park, NC 27709. Accepted for publication September 17, 1982. 1983 Alan R. Liss, Inc. 422 Dement et al MATERIALS AND METHODS Although the plant under study began production of asbestos products in 1896, detailed personnel records were first maintained beginning in approximately 1930. The record system has remained remarkably unchanged since that time. For each worker, an employment card was completed at initial employment giving name, date of birth, sex, race, social security number, marital status, and address. This same card also con tained the detailed work history giving exact dates of employment by plant department and specific jobs. All information from these cards was entered onto a computer data file. The cohort was limited to 1,261 white males employed one or more months in textile production operations with at least one month of plant employment between January 1, 1940, and December 31, 1965. The cohort was followed through December 31, 1975. The 1965 cut-off date for cohort entry was chosen to insure that all workers would have a minimum "latency" of ten years as of the study end date. The one-month entry criteria was used to allow comparisons with other mortality studies of chrysotile workers [McDonald et al, 1980]. An attempt was made to determine the vital status of all cohort members as of December 31, 1975. The primary sources of information used for this follow-up in cluded the Social Security Administration (SSA), Internal Revenue Service (IRS), US Postal Mail Correction Service, state drivers license files, and state vital statistics of fices. Individuals not located through these primary sources were traced using local records such as telephone listings, Polk directories, property records, voter records, records of funeral homes, and v arious other local sources. Cause-specific standardized mortality ratios (SMRs) were calculated using a lifetable analysis based on the technique developed by Cutler and Edercr [1958]. Personyears at risk of dying were distributed by five-year age, calendar time, and time since initial employment (latency) groups. Person-years were accumulated for each cohort member beginning when all requirements for cohort entry were met until the date of death or December 31, 1975. Those whose vital status remained unknown were as sumed alive as of the study cut-off date, thereby contributing their maximum possible person-years to the analysis. The follow-up period for this study spans the fifth through eighth revisions of the International Lists of Diseases and Causes of Death (I CD A). Death certificates were coded by a qualified nosologist according to the ICDA revision in effect at the time of death. All death codes were then grouped into 89 death categories based on the seventh revision for purposes of standardization. Individuals known to be deceased but for whom no death certificates were available were assumed to be deceased, cause unknown. The number of expected deaths, standardized for sex, age, race, and calendar time, were calculated by application of cause-specific death rates for the total United States to the person-years at risk of dying. Death rates specific to the 89 seventh-re vision death groups were calculated from yearly tallies of deaths and census data. For evaluating exposure-response, cumulative exposures were calculated for each worker using detailed work histories contained in plant personnel records combined with the estimated level of exposure for each job held. A worker's cumulative dust ex posure at any time during the follow-up period was expressed as the cumulative product of the estimated average dust concentrations for particular jobs held by the worker and the time duration in those jobs. Mortality Among Chrysotile Workers 423 The estimated fiber concentrations for the period 1930-1975 are presented in the companion manuscript in this volume and are expressed as fibers longer than 5 /mi in length per cubic centimeter of air (fibers/cc). Time spent in each job was calculated as the difference between dates of job changes and expressed in days; therefore, the cumu lative exposure was expressed as fibers/cc x days. Using this method, weekends and holidays are not eliminated; thus the true number of "work days" in each job was over estimated. This was done to provide conservative estimates of exposure and to account for periods of work longer than eight hours. A few workers began employment before 1930. Pre-1930 exposure levels for these workers were estimated by assigning exposure lev els prior to implementation of control measures for each job held before 1930. Mortality in relation to exposure was investigated by creating cumulative expo sure strata through which a worker was moved as his cumulative exposure increased during the follow-up period [Breslow, 1976; Lundin et al, 1971]. This method allowed full use of each cohort member's survival experience for the entire follow-up period. Cause of specific SMRs were calculated for each exposure stratum using appropriate age, race, sex, and calendar time specific death rates. Statistical significance of observed excess or deficit mortality was evaluated using the Poisson distribution [Pearson and Hartley, 1958]. RESULTS Overall Mortality Results of the follow-up efforts are summarized in Table I. Vital status was deter mined for all but 26 (2.1 /o) of the 1,261 cohort members. Of the 308 deaths, all but 17 death certificates were obtained. A total of 33,141 person-years at risk were experienced by this cohort between January 1, 1940, and December 31, 1975. Observed and expected deaths by cause are given in Table II. A total fo 308 deaths were observed, whereas only 205.66 were expected (SMR = 150, p < 0.05). Observed and expected deaths by time interval since initial employment are shown in Figure 1. No statistically significant excess mortality was observed until after 15 or more years since initial employment, a finding consistent with other occu pational mortality studies. Examination of cause-specific mortality in Table II shows significant excess mor tality for malignant neoplasms (SMR = 168, p < 0.05), diseases of the circulatory sys tem (SMR = 125, p < 0.05), nonmalignant respiratory diseases (SMR = 294, p < 0.05), and accidents (SMR = 134, p < 0.05). Increased mortality was also observed for TABLE I. V ital Status for White Males With One or More Months Textile Employment Vital status as of Dec 31, 1975 Known alive Known deceased Certificate obtained Certificate not obtained Unknown vital status Total No. 927 308 (291) (17) 26 1261 Percent 73.5 24.4 (94.5) (5.5) 2.1 100 `7> 350 300 h to --EXPECTED DEATHS -A OBSERVED DEATHS YEARS SINCE INITIAL EMPLOYMENT Fig. 1. Observed and expected deaths for all causes by time interval since initial employment. TABLE II. Observed and Expected Deaths by Cause for White Male Asbestos Textile Workers 1940-1975 Cause of death 1CDA 7th list no. Observed Expected SMR All causes Malignant neoplasms Digestive system Trachea, bronchus, lung All other sites Vascular lesions affecting the central nervous system Diseases of the circulatory system All tuberculosis Nonmalignant respiratory diseases Acute upper respiratory infection Influenza Pneumonia Bronchitis Other respiratory disease Accidents Other violent deaths All other known causes Unknown causes including 17 missing death certificates 150-159 162-163 330-334 345 400-468 001-019 470-475 480-483 490-493 500-502 510-527 800-962 963-964 970-985 308 205.66 150" 59 35.06 168" 13 9.89 131 35 11.10 315" 11 14.07 78 15 10.97 137 105 83.74 125" 6 3.48 172 28 9.53 294" 0 0.03 -- 0 0.04 _ 4 4.19 95 0 0.55 _ 24 4.35 552" 34 25.38 134" 9 9.37 96 29 26.91 108 23 2.55 _ Mortality Among Chrysotile Workers 425 diseases of the central nervous system and tuberculosis; however, these excesses were not statistically significant. The elevated SMR for malignant neoplasms shown in Table II is largely ac counted for by cancer of the trachea, bronchus, and lung, and cancers of the digestive system. A total of 35 lung cancers were observed and only 11.10 were expected (SMR = 315, p < 0.05). Tables 111 and IV show lung cancer mortality by time interval since first employment (latency) and duration of employment. No lung cancers were ob served prior to 10 years latency and 17 of the 35 lung cancers occurred after 30 or more years latency. Table IV demonstrates an increasing trend in the lung cancer SMR with increased employment with an SMR of 976 for those employed 20-29 years. Of the 28 deaths attributed to nonmalignant respiratory diseases, 24 fell into the category "other respiratory diseases" (ONMRD) which includes asbestosis. Of the 24 deaths in this category, asbestosis or pulmonary fibrosis was the underlying cause for 17. ONMRD mortality by latency and duration of employment is given in Tables V and VI. The SMR for ONMRD was not statistically elevated until greater than ten years employment but increased dramatically for those employed more than ten years. Increased mortality due to cardiovascular diseases is a consistent observation among asbestos workers and represents a combined stress on the cardio-pulmonary system. A review of death certificates for the 105 deaths found that six mentioned as bestosis or pulmonary fibrosis as a contributory condition. TABLE III. Lung Cancer (ICDA 162,163) Mortality by Time Interval Since Initial Employment Years since initial employment Total < 10 10-19 20-29 2>30 Observed 0 6 12 17 35 Expected 0.47 2.08 4.76 3.79 11.10 SMR -- 288a 252a 449a 315a ap < 0.05. TABLE IV. Lung Cancer (ICDA 162,163) Mortality by Duration of Employment Years employed < 10 10-19 20-29 s>30 Total Observed 15 5 12 3 35 Expected 8.09 1.05 1.23 0.73 11.10 ap < 0.05. SMR 185a 4763 976a 410 315a TABLE V. Mortality Due to "Other Nonmalignant Respiratory Diseases" (ICDA 510-527) by Time Interval Since Initial Employment _______________________________________ ________ Years since initial employment Observed Expected SMR Total < 10 10-19 20-29 ^30 1 0.26 - 4 0.74 541a 10 1.77 565a 9 1.58 570a 24 4.35 552a 426 Dement el al TABLE VI. Mortality Due to "Other Nonmalignant Respiratory Diseases" (ICDA 510-527) by Duration of Employment Years employed Observed Expected SMR < 10 10-19 20-29 >30 Total 6 3.13 192 5 0.39 1282* 9 0.51 1765* 4 0.32 1250* 24 4.35 552* ap < 0.05. Only one mesothelioma was observed among this cohort. This was a peritoneal mesothelioma confirmed by autopsy. The interval (latency) between initial employ ment and death was 34 years. There were several other deaths which mentioned "cancer of the abdomen" that may be suspect; however, no autopsy or other confirmatory data were available. Exposure-Response Relationships Both lung cancer and asbestosis require lengthy periods from initial exposure to become clinically evident. For exposure-response studies based on mortality it is im portant to restrict the analyses to those achieving sufficient latency to be "at risk" of dying from lung cancer or asbestosis. For this reason, exposure-response analyses were restricted to those achieving 15 or more years since initial employment (latency). This was accomplished by beginning accumulation of person-years for each worker after the 15-year latency period was satisfied; however, cumulative exposures began at employ ment. Those dying before reaching 15 years latency were excluded. Results of the exposure-response analyses for all causes, diseases of the circula tory system, lung cancer, and digestive system cancer are given in Tables VII and VIII. Significant excess overall mortality was observed for all cumulative exposure strata except the highest, where observed numbers were small. Although increased numbers of circulatory system deaths were observed for each exposure stratum, statistically sig nificant excesses were observed in only one. There appeared to be no consistent increas ing trend of circulatory system mortality with exposure. Table VIII demonstrates a strong exposure-response relationship for lung cancer. Statistically significant excess lung cancer mortality was observed for all but the lowest exposure stratum where five lung cancers were observed versus 3.58 expected. Lung cancer SMRs increased with increasing cumulative exposure with an SMR of 1,818 in the highest exposure stratum. A plot of lung cancer SMRs by cumulative exposure is given in Figure 2. A linear 1 unction appears to adequately describe these data. Al though digestive system cancers were not excess overall, an increasing trend in the SMR was observed with exposure; however, none of the exposure strata demonstrated a sta tistically significant excess. Exposure-response relationships for ONMRD are given in Tabic IX. Significant excess mortality was observed in all except one exposure stratum with the SMR increas ing consistently with increased exposure. Shown in parentheses are those deaths with an Mortality Among Chryscrtik1 Workers 427 Fig. 2. Exposure-response for lung cancer among white males achieving 15 or more years latency. TABLE VH. Exposure-Response Relationship for All Causes and Diseases of the Circulatory System Among White Males Achieving 15 or More Years Latency Cumulative exposure fiber/cc x days Observed All causes Expected SMR Diseases of the circulatory system (ICDA 400-468) Observed Expected SMR < 1,000 1,000-10,000 10,000-40,000 40,000-100,000 > 100,000 Total 79 55.01 144 34 67 48.62 138a 24 60 32.77 183a 24 33 13.59 243a 8 6 2.50 240 2 245 152.49 161a 92 24.40 21.52 15.19 6.50 1.27 68.88 139 112 158a 123 157 134a *p < 0.05. TABLE VIII. Exposure-Response Relationships for Lung Cancer and Digestive System Cancer Among White Males Achieving 15 or More Years Latency CV'uUmiiiuulmattiivVeV evAv|nAoZO^uUrleV fiber/cc x days < 1,000 1,000-10,000 10,000-40,000 40,000-100,000 > 100,000 Total Lung cancer (ICDA 162,163) Observed Expected SMR 5 3.58 140 9 3.23 279a 7 1.99 352a 10 0.91 10993 2 0.11 18l8a 33 9.82 336s Digestive system (ICDA 150-159) Observed Expected SMR 2 2.85 70 1 2.54 - 4 1.78 225 3 0.77 390 0 0.14 -- 10 8.08 124 ap < 0.05. Jjenienil* *t m. mndteeMimiig sauafcfct' afftarriiib\ustw adaxtsamu- av ribr^-u^. Ob nftxt bii^e uiteacfti* iim aft*? IdweM 'saratttiimiib atmT* r.we were Mzr:h\Xi$ r.n- *or ^xaikmomiir? rellanuoe^ftvp^ for :a:foernor;u-i and pul;.T.^ti.aa?> foibevonitf were im- traced ^paraffleih 8ja catoxlani^a' inuTder.ce der^vr:% i.ea^er per 0 jOiOiO;*per^onr^etajir*)) tfioxr eaethi e^cmiJire 'HiLT'affnimm.. Tlhere dana are rner, m lark X and do'wmi ^aipikiai%^ iim Fmw1 3.. A orand uo iimrdta' demrin* wiinlhi <g^mmxt'Z a omttiH*ar*<sdi wiln afoe iiimridmpe fiim tifoe tegiraesm e^pmiLure' cranium app^oAimaceito SO* oiimne*' attnac to>tf altoe Ifomesu.. Tftwf paninermi vfew n hod T'$gm&: 3 ixsi^enis athcai aliiit1 imiidimcg<Gpf arad puitei^^ ffinmiis ineaeasev mm& rapidly wmHn e^p/omure nftairo pncdikaifidl toy a Ifiimeaar tfiucomaniiko>im.. TTteis m m$y a aerccafoe ^ISter^aaiADtm ivmee erf* aw<a> eatie^ wtsnetoftv- serocdiimataeMgteHt ca&ggmy wiiaihi a resting,wideaiowifudeoeerarer* all foiri dem&mr DISCUSSION Sc.attMki2llIl>' 5%mi5lfkanm e%ress mociiallm mas i^toser^ed for lloum^ oirar amdi nan- inmaliipiiaM r^pmr"aar%'dttiemic^ amoroig afmstOxiDle astoe&iQ'H- tie^tilk wovlkisirk. FuamJtoaiOT^ mmg. exposure m Elbe e%po's:mre * artaMe and atoe- SMR a* tilme a^tf'dfo^ estie risk, Mrcra^ CTpiMmireHreipQ'ffliie' adiarfomMp* were ed?eir*etd foe Humm^ eamgmr, A r* meiHlSM'csibolWl mpAfl appears tt adk^waceik' desoifcc nine forrani f nffaf* cspoMunMC- spoaiHC tfiiMMiicim for tang eanoeer. T'ABUE IX, ^ tlw OnlfetTf VortunuiSujii AMMn^ Maira; Adbik^iii^ 115 <m Mminr \& bMXMMIDI CiMis'Jidaciir^ti1 c^psmmir^ Kfiite^r siir * oEara*)) o^cro jcHiite i(GCtd>a 5ii(m>srr)) OlbitsrAtiidJ1- E^ptianitfsjil SMB& c U jOtOlO) vjwfti-mjsm > nwjofio Tmsi 5 tcr D UD 2 (CD ZCHllT)) 11.3# H.J^ COI^S (0L.3S (OUOlf JJUfc JK -- MR* n*4CS*' 238* 57* ilEE *jp> < (C'U5.. iMXdiixnMif' nciIinnoBTitCir <0>tF <5fijWCn^d] (iflteailflB* 3nn.lrilifelM!cfull H<j? 2I$i(ift^t0rjicf UJU' ]pnillnnrttOircjttr& ffiillrinffTflHv.. T %BLt X- InriMfltHnra ftawitty tfiow w PhAhmomB7 KAMraiw^ at* Vindwi^uqi: Caam lirMltc Mailkri. Atflut^n^ 15 mr JkAnvr Y*su% LaHcmc? (ri'iiiTnnniillaiiTii'xif- tf^sptQisffiiiine tfiifeor: 'ck: ,< Casas PtoTMKn-gj'icsiini, m irmlk < D jDlOf}.' a,Q)-i]fQ>J0M Da0OO)~43rX(O.TO jji0.'jD!0iQi-l]fDDMQ0i0 .> DflJMffl Inwall "Ti 11 6 6 Ti 117 SC54 .5J0CT luwoe IC6 !36j0m0> "BatHsdl <ou3 IBouk^osud dlii:iraiife[iiin.flo.. Catstc* pfsr Q JOKKO) perwcn^^caiirf (Oi 32 (OLD* \um 5.m 115..6T II.,1.0(5 traiwitfrirfbmng* iirKiejnnaJ^11 .m- u..116 C0U03- 5.577 (0X.73- 4.32 330MJ5i0r5 I3.-5T..2^ (01/52- 11 ..7X0) Mortality Among Chrysolite Workers 429 CUMULATIVE EXPOSURE (Thousand Fiber/cc x Days) Fig. 3. Incidence density for asbestosis and pulmonary fibrosis mortality among white males achieving 15 or more years latency. TABLK XI. Comparison of Lung Cancer Mortality Rates (ICDA 162,163) Between County in Which Study Plant Was Located and Contiguous Counties, 1950-1969 United States State in which plant was located County in which plant was located Contiguous counties (range) Counties one remoxed (range) Age-adjusted deaths/100,000 (white males) 37.98 37.83 66.5 40.1-53.4 25.9-44.1 observed for nonmalignant respiratory disease mortality including asbestosis and pul monary fibrosis. There are several factors which need to be considered in evaluating the occupa tional contribution to observed mortality patterns. The most important of these are the choice of standard population death rates to estimate expected deaths and cigarette smoking patterns among the cohort. Other potential confounders such as age, race, sex, and calendar time period were dealt with in the study design. The standard population death rates chosen for this study were those for white males for the entire United States. Table XI gives a comparison of lung cancer (ICDA 162,163) age-adjusted mortality rates for 1950-1969 for counties in the same area as the study plant with state and US rates [Mason et al, 1975]. Lung cancer death rates for the state in which the plant was located were nearly equal to US rates. On the other hand, rates for the county where the plant was located were 75% higher than US rates for white males. The choice of an appropriate comparison population for mortality analyses is difficult, and arguments could be made for using rates tor a set of counties contiguous to the county in which the plant was located. However, there are serious limitations to 430 Dement et al this approach which were considered in this study and resulted in rejecting the use of local county rates. First, the county in which the plant was located is the site of a large shipyard industry. Employees for this industry were largely drawn from the local popu lation. Many of these workers arc thought to ha\e been exposed to asbestos during ship construction and repair. In an ecological study. Blot et al [197X] demonstrated an as sociation between county lung cancer rates and shipyard employment. In a more re fined case-control study. Blot et al [1979] demonstrated a summary odds ratio of 1.6 for shipyard employment and lung cancer alter adjusting for smoking, other occupa tions, age, race, and county of residence. These data suggest that lung cancer death rates in the area in which the plant was located are likely to be elevated by local shipyard employment. In addition, the effect of the plant under study on county lung cancer mortality must be considered. The effects of shipyard and asbestos plant employment make the use of local death rates inappropriate for this study. Howes er, e\en if rates for contiguous counties had been used (Table XI), the expected lung cancer rates tor white males would have been increased by only approximately 150 o; not nearly sufficient to account for the ob served excess lung cancer risk among the studs cohort. Detailed plant svork histories as svell as prior occupational histories (collected by the I S Public Health Service in 1964 and 1971) svere reviesved for lung cancer and non malignant respiratory disease cases. No association svith either prior shipyard employment or plant employment in rubber operations svas observed. Cigarette smoking is a known risk (actor for respirators cancer, and smoking and asbestos exposures have been shown to act in a synergistic manner to greatly increase the risk of lung cancer [Hammond et al, 1979]. Respiratory-symptom questionnaires including questions on smoking history were administered by the US Public Health Service (USPHS)to active svorkers in this plant in 1964 and again in 1971. In addition, smoking data available from plant medical records were also collected. While smoking histories are not available on all cohort members, these data were useful in estimating the prevalence of smoking in this plant for comparison with smoking patterns among US males who were used as the standard population for estimation of expected lung cancer deaths. The prevalence ot cigarette smoking habits among the asbestos study cohort is given in Table XII. These data largely represent the smoking prevalence found by the 1964 USPHS survey since the cohort was limited to those achieving 1 month of employ ment before 1965. The USPHS 1971 data and company data were used only for those missed in 1964. Among white males, 52.4% were found to be current smokers, 25.3% nonsmokers, and 22.3% past smokers. Table XII also compares smoking prevalence among the study cohort members with comparable data for US adults [USPHS, 1979]. These data show the prevalence of smoking among white males in the study cohort to be nearly identical to that of US white males. The 22.3% prevalence of past smokers is also identical to US figures. Available smoking data for this cohort suggest that the observed lung cancer and nonmalignant mortality excess among white males cannot be explained by cigarette smok ing independent of asbestos exposure. This conclusion is also supported by the expo sure-response data. While smoking cannot explain the observed lung cancer excess, an interactive effect with asbestos exposure is likely. Although mortality among asbestos workers has been extensively studied, there are a few studies of populations exposed to only chrysotile. Mortality among Quebec Mortality Among Chrysotile Workers 431 1 AB1.K XII. Summary of C igarette Smoking Patterns for White Male Asbestos Textile Workers and Comparison With Data for US White Males Asbestos workers (N = 292) US white adult males (1965) Current smoker (Co) 52.4 51.5 Past smoker (Co) 22.3 22.1 Nonsmoker (Co) 25.3 26.4 chrysotile miners and millers has been extensively studied [McDonald et al, 1980]. The most recent report for this cohort included 10,939 men who had been employed one or more months and followed between 1926 and 1975. An overall SMR for lung cancer of 125 was observed; 42 deaths ( 1 .3%) were due to asbestosis and 11 (0.3%) due to meso thelioma. Increased mortality was also observed for cancer of the stomach and esopha gus but no other gastrointestinal sites. Similar patterns of lung cancer and asbestosis mortality have been reported for Italian chrysotile miners and millers where an SMR for lung cancer of 206 was observed among those with sufficient latency [Rubino et al, 1979]. The McDonald et al studies demonstrated a relatively modest increase in lung cancer risk even in the highest exposure group. Nicholson et al [1979] reported larger excesses for lung cancer and asbestosis in their study of chrysotile miners and millers in Quebec. This latter study cohort consisted of 544 miners and millers with at least 20 years seniority followed between 1961 and 1977. A total of 28 lung cancers were ob served versus 11.1 expected (SMR = 252). There were 30 deaths due to noninfectious respiratory diseases, whereas only 6.7 were expected. Of these 30 deaths, 26 were due to asbestosis. Only one mesothelioma (pleural) was observed. Studies of factory populations exposed to only chrysotile are rare. Weiss [1977] studied a small cohort of 264 workers in a plant producing asbestos millboard and re ported no excess cancer mortality. However, there were only 66 deaths (two of which were due to asbestosis). There are a few other published reports with which to compare the exposure-re sponse data obtained in this study. In fact, there are no other reports of exposure-re sponse using exposures expressed as fibers/cc by the phase-contrast method; all other reports have used impinger (MPPCF) data [Henderson and Enterline, 1979; McDonald et al, 1980]. For comparison with other published data, approximate impinger expo sure values, expressed as MPPCF x years, were calculated for data from the current study using the impingcr-membrane filter conversions. Estimated exposure-response for lung cancer based on these estimates are given in Table XIII along with other pub lished data. The data in Table XIII show the SMR for lung cancer at a given cumulative expo sure for the present study to be much higher than other published values. However, there are differences in the designs of the three studies which may account for some of this apparent discrepancy. For example, the McDonald et al [1980] study included per sons exposed to extremely high airborne-tiber levels, thus competing asbestosis risk may be important. The study by Henderson and Enterline [1979] consisted ot retirees 65 years or older. In the present study, only eight ot 35 lung cancer deaths were 65 or older. The Henderson and Enterline study may be a survivor population with less lung cancer risk for those surviving to age 65. 432 Dement et al TABLE XIII. Comparison of Exposure-Response Relationships for Lung Cancer With Other Published Data Present study Approximate MPPCF x yrs <0.9 0.9- 9.1 9.1-36.5 36.5-91.3 >91.3 SMR 140 279 352 1099 1818 Henderson and Enterline [1979] MPPCF x yrs SMR < 125 125-249 250-499 500-749 >750 197.9 180.0 327.6 450.0 777.8 McDonald et al MPPCF x yrs SMR 30 104 100 114 300 500 1200 142 170 268 o o00 aBased on cumulative exposures until age 45 years. SMRs calculated from regression line provided by authors. TABLE XIV. Comparison of Lung Cancer Mortality by Duration of Employment for Chrysotile-Exposed Cohorts Duration of employment Current study Observed Expected SMR McDonald (19801* Observed Expected SMR 1 mo-5 yr 5-20 yr >20 yr 11 5.32 207 76 83.39 91 3 1.22 246 50 36.50 137 15 2.06 728 104 64.60 161 aData in this table represents mortality after 20 or more years latency. Differences in lung cancer exposure-response relationships between this study and that reported by McDonald et al [1980] are significant. Estimation of historic ex posure levels is a difficult task, and it is possible that part of the apparent disparity be tween the two studies reflects imprecision of these estimates. However, large differ ences are also noted using duration of employment as a measure of exposure. Table XIV shows such a comparison for the two studies. In each duration of employment stratum, much larger lung cancer SMRs were observed in the current study. These dif ferences were very large for those achieving more than 20 years employment. These data suggest that imprecision of exposure estimates does not account for observed dif ferences in exposure-response. Other factors such as differences in airborne-fiber char acteristics (length, diameter, etc) may be important. Compared with other asbestos processing operations, textile processing has been shown to produce a greater airborne fraction of long (>5 fim in length), thin (<1.5 /nn in diameter) fibers [Dement and Harris, 1979]. Animal studies have shown these fibers to be more capable of producing tumors upon pleural implantation than are shorter, thicker fibers [Stanton et al, 1981]. The current Occupational Safety and Health Administration asbestos exposure standard of 2.0 fibers/cc is based on an allowable lifetime cumulative exposure of 100 fibers/cc x years (ie, 2.0 fibers/cc for a 50-year working lifetime). Based on data from this study, significantly elevated mortality risks are predicted for lung cancer and for asbestosis at cumulative exposures of 100 fibers/cc x years in the textile industry. This observation is based on use of cumulative exposures as a summary exposure measure to account for both exposure level and duration. Further analyses of these data are planned to investigate the separate effects of exposure level and duration. I ACKNOWLEDGMENTS Mortality Among Chrysotile Workers 433 Support for this research was provided by the National Institute for Occupa tional Safety and Health (NIOSH). The authors express their appreciation to Judy Bachmann, Joyce Ayersman, and Janet Dement for their assistance with data coding and cohort follow-up; to David Brown, Jay Beaumont, and Paul Watkins for their as sistance with computer analysis; and to Martha Devone for manuscript typing. REFERENCES BU>( WJ. Harrington M. Toledo A, Hoover R, Heath C, Fraumeni JF (1978): Lung cancer after employ ment in shipyards during World War II. N Engl J Med 299 (No. 12):620-624. Blot \\ I. Stone BJ. Fraumeni JF, Morris LE (1979): Cancer in U.S. counties with shipyard industries durini! World War II. Environ Res 18:281-290. Hreslow N (1976): Some statistical models useful in the study of occupational mortality. In Whitmore A (cd): "Environmental Health Quantitative Methods, Proceedings of a Conference on Environmen tal Health. Ana, Utah." Philadelphia, Pennsylvania: Society for Industrial and Applied Mathe matics. pp 88-103. Cutler S.l. I derer BS (1958): Maximum utilization of the life table method in analyzing survival. J Chronic Din 8:699-712. Dement IM. Harris Rl (1979): Estimates of pulmonary and gastrointestinal deposition for occupational liber exposures. DHEW Publication No. (NIOSH) 79-135. Hammond 1C, Selikoff IJ, Seidman H (1979): Asbestos exposure, cigarette smoking and death rates. Ann NY Acad Sci 330:473-490. Henderson YU, Interline PE (1979): Asbestos exposure: Factors associated with excess cancer and respira tors disease mortality. Ann NY Acad Sci 330:117-126. 1 undin IF. Wagoner Jk, Archer VE (1971): Radon daughter exposure and respiratory cancer: Quantita tive and temporal aspects. NIOSH-N1EHS Joint Monograph No. 1. Mason TJ. Mackay FW, Blot WJ, Fraumeni JF (1975): Atlas of cancer mortality for US counties. DHEW Publication No. (NIH) 75-780. ,, McDonald JC. Liddell FDK, Gibbs GW, Evssen GE, McDonald AD (1980): Dust exposure and mortality in chrysotile mining, 1910-75. Br J Ind Med 37:1 1-27. Nicholson WJ. Selikoff IJ, Seidman H, Uilis R, Formby P (1979): Uong-term mortality experience of chrvsotile miners and millers in Thetford Mines, Quebec. Ann NY Acad Sci 330:11-31. Pearson I S, Hartley HO (eds) (1958): "Biometrika Tables for Statisticians," 2nd ed. Cambridge: Cam bridge University Press. Ruhino GF Piolatto G, Newhouse MU, Scansetti G, Aresin GA, Murray A (1979): Mortality of chrysotile asbestos workers at the Balangero Mine, Northern Italy. Br J Ind Med 36:187-194. Stanton MF, Uayard M, Tegeris A, Miller E, May M, Morgan E, Smith A (1981): Relation of particle di mension to carcinogenicity of amphibole asbestoses and other fibrous minerals. J Natl Cancer Inst 67 (No. 5):965-976. r. U.S. Public Health Service (1979): Smoking and health, a report of the Surgeon General. DHEW Publica tion No. (PHS) 79-50066. Weiss W (1977): Mortality of a cohort exposed to chrysotile asbestos. J Occup Med 19 (No. 11):737-740.