Document baQeBDwV7KrzvBaejNmgBke5Z

CANCER MORTALITY AMONG RUBBER WORKERS: AN EPIDEMIOLOGIC STUDY * A. J. McMichael. D. A. Andjelkovic, and H. A. Tyroler Department of Epidemiology School of Public Health Unirersie of North Carolina Chapel Hill. North Carolina 27514 ISTRODUCTION Chemical carcinogenesis is toda)- an accepted phenomenon, based partly on a variety of occupational epidemioiogic studies. In 1895, an increased risk of bladder cancer was reported for workers in the dyestuffs and chemical industries1 In 1974, excess deaths from angiosarcoma of the liver were reported in the polyvinyl chloride manufacturing industry.' In the intervening eighty years, similar epidemiologic studies have been carried out in many other "chemical" industries, including the rubber industry. The rubber-tire manufacturing industry uses a vast and ever-changing variety of organic and inorganic chemicals. Further, because of the high temperatures involved in several of the processes. there are many chemical reaction byproducts produced, most of uhich remain unidentified. Among the chemical ingredients now or previously used are various proven or suspected carcinogens, such as .&naphthylamine. benzzne. asbestos (in talc dust), and various nitrosamines. Previous epidemiologic studies of the rubber industry, in Great Britain and in the United States, have identified excess deaths from several types of cancer. Associations of some of these cancers with specific jobs have also been reported. For the period 1919-31. the British Registar General reported that rubber workers showed a 10% total mortalit). excess. compared with national mortality data, and that this excess derived from cancer deaths3 The United States Department of Health, Education. and Welfare has reported that, for male rubber workers dying in 1950. when compared with death rates in male workers in all manufacturing industries. there was an 8% excess of cancer deaths. The greatest excesses were for cancers of the large bowel (30% ), respiratory system ( 18% ). and the lymphatic ana hematopoietic system (54% ) .I Mancuso, in 1949. reported an elevated proportion of deaths due to cancer for workers in the rubber and plastics industry in Ohio (the center of the U.S. rubber industry) .5 In particular. he reported a proportional mortality excess from respiratory, genitourina?. and central nervous system cancers. Subsequently, Mancuso et al. anal!zed mortality in a cohort of 1877 rubber workers between 1940 and 1968, finding. firstly. excess cancer death rates in workers 1 .:j I26 Annals New Y o r k d c a d e r n y of Sciences other associations between specific cancers and jobs within the rubber industry have been reported by Mancuso, based on data from four separate rubber corn- panies.s In two recent, separate studies, McMichael et al. reported. firstly, e x c w death rates from cancers o the stomach, cdon, prostate, and lymphatic hematopoietic system in a cohort of 6678 rubber workers. and. secondly, an association between lymphatic leukemia in rubber workers and a work-history of solvent exposure.'' Io Simultaneously, Monson and Nakano. in a preliminary analysis of a large cohort of rubber workers, reported excess death rates from leukemia, and cancers, of the bladder, stomach, pancreas and lar Certain of these and Other cancers were associated with exposure to specific work areas. In Great Britain, in 1954. Case and Hosker confirmed a suspected excess of bladder cancer in rubber workers.lz Despite elimination of the presumed car- cinogen (,B-naphthylamine) from the production process, some persistence of a bladder cancer mortality excess was subsequently reported by Veys l3 Adelstein." In 1974, however. more than twenty years after the change, report of a large mortality study in the rubber and cable-making indust stated: "No evidence is found of a continued excess risk of neoplasms of bladder in people who entered the industry after 1949." In that same st Fox et al. found an excess total cancer mortality among rubber workers in tire-manufacturing section, including particularly an excess for cancer of bronchus and lung which. In turn, was highest for workers exposed to ei the tire curing (vulcanization) process. or to jobs in the stores. packaging, despatch area. Despite the differences in location, time period. and methodology, th various epidemiologic studies of the rubber industry, demonsrate a gen problem of excess cancer mortality. This present study has sought to determ the recent cancer mortality experience of U.S. rubber workers and, ultimat to discover the specific jobs (and, where possible, the specific chemical physical agents) associated with increased risk of cancer mortality. . MEMODS The Occupational Health Studies Group. within the Schooi of Publlc Hea University of North Carolina. is under contract jointly- to the United Rub Workers Union and four of the six major U.S. rubber c o m p m e s to carry a comprehensive occupational health research program. The need for s research was agreed upon by labor and management in their 1970 contr negotiations. Foremost within the research program to date has been th analysis of mortality during the past decade. Four cohorts of male rubber workers were identified. at the major manufacturing plant in each of the four rubber companies (desi C, and D). Three of the plants are located in Akron, Ohio, and one in consin. Working from payroll, pension, life insurance, and other co records, the four historical populations of hourly (i.e., production) male workers-active or retired-and aged 40 to 84 at 1.1.64 were identified. company D, the cohort was identified at 1.1.63.) The ten-year mortality experience of these four cohorts uas determined fro death claims filed with the companies under the group life insurance policy McMichael et al.: Rubber Workers 127 the nineteen fifties. The completeness of death ascertainment was by cross-checking internally against other company records. Within cohorts, an external check was made against the files of the Internal evenue Service, confirming that a random sample of apparent cohort survivors ere in fact still alive. Approximately 1% of cohort workers were lost to follow-up, being initially tive workers who subsequently terminated their employment and went to work sewhere. Of the remaining study subjects, death certificates were obtained for 98% of those who died. These were coded by a certified nosologist, according $$to the Eighth Revision of the International Classification of Diseases. '* The mortality experience of these four cohorts was evaluated by the calcula- tion of race-age-standardized mortality ratios (SMR). (Because of the decline in the validity of cause-of-death recording at very old ages, the few deaths above age 84 were excluded from the analysis.) The 1968 U.S. male population provided the ``standard'' race-age-specific death rates, for calculating the expected deaths. Choice of this particular standard population introduces an underestimation of mortality. because an active industrial workforce. not exposed to significant work-related health hazards, is normally healthier than a similarly aged sample of the general population.'`. In a hazard-free occupational environment, an industrial population, when compared to the general population, could reasonabIy be expected.to have an SMR of about 80. (Indeed, according to Social Security actuarial data, the sumival experience of the general industrial working population may be 40% better than the general population.) li One procedure Ily eliminate this underestimation of mortality is to compare each ShfR to the underlying all-causes SMR. ing the association of specific cancers to specific work exposures, histories of individual workers were obtained from the companies. diversity of job types. combined with the individual's job mobility, presents dable task I f the full range of work experience is to be taken into account osed to using a single-classiiication approach, based on "most representa- tive job," "most recent job." or "job at cohort-identification") . The techniques in this study to code and analyze an individual's total work experience have been described e1sewhere.l` The job classification scheme entails grouping jobs according to type of materials handled and/or production stage. 1 RESL`LTSAND DIsCUSSION I mposition of the study population is shown in T AB L E 1. The two (companies A and C) were also of older average age. Overall. rs here aged 10-54 at cohort identification, 30% were aged and 29% were aged 65-84. Retirement normally occurs at age 65 the rubber industry. Of the total of 18,903 workers. 98% had worked - r at least ten years in the rubber industry at the time of cohort identification, and the average employment duration was 25-29 years. For cohorts A, B, and c the Proportion of nonu hites ranged from 5-1 3 CC. Cohort D contained white only. CURE 1 shows the S3IRs for deaths from all for the four cohorts. 12s Annals New York Academy of Sciences 8For each cohort, three SMRs. re calculated: for ages 40-64, for ages 65- and for the full age-range 40-84. Three observations can be made: 1. Mortality is relatively greater at the older ages. 2. The individual cohort SMRs (for the age-range 40-84) vary bet 82 and 98.f 3. The combined-cohorts SMR, fpr ages 40-84, is 94-suggesti 10-15% excess of mortality in a population whose SMR cou wise be expected to be in the range 80-85 if its members had exposed to work-place health hazards. Of the total of 5106 deaths, 1014 were due to cancers. FIGURE 2 shows individual and cgmbined cohort SMRs for deaths due to cancer. Again, cancer mortality relatively greater at older ages, and there is approx 10% variation among the four cohorts. For each of the 15 cancer SMRs the corresponding 'all-causes SMR is lower, suggesting that in this popul of rubber workers there is an excess of cancer deaths. TABLE1 AGE COMPOSITION, B Y PERCENTAGE, OF T H E FOUR MALERUBBER-WORKER COHOR AT 1/1/64 (COLUMN PERCENTAGE) Age 40-54 55-64 65-84 Total Number A 38% 30% 32% 6,678 Company BC 59% 24% 17% 1,339 37% 3 1% 32% 8,938 D 58% 39% 13% 1,948 '&.18,903I"%&$ The cancer mortality experience of each of these four cohorts analyzed in detail. (Further, proportional mortality analyses have bee out at a number of other smaller plants, for which the underlying p risk have not been identified.) The results of these total-population presented before considering the relationship of specific cancers to categories within the rubber industry. The SMRs for four major cancers of the digestive system are show FIGURE 3, and are calculated for the four cohorts combined. For each can three SMRs are shown: for ages 40-64, 65-84, and for the full age-r 40-84. The four individual cohort SMRs, for ages 4SF-84,are also indica For stomach cancer, there is a clear excess mortality, which is relati greater at younger ages. Overall, the stomach cancer SMR of 148 repr j I. 1,- I 'j 125 40-64 g65-84 40-84 - All-Causer SMR NO OF DEATHS. 392 61 482 79 1014 I 3 0 Annals New T F k Academy of SC;m c e s fCancers of the liver. gall bladder, and bile ducts, not shown h deaths. representmg a deficit of approximately 25% be!ow evpe With respect to the excess stomach cancer death rates. the approx 50% excess shown here has also been found repeatedly in proportional analyses at other plants. A detailed study of the specific work-exposu lates of stomach cancer is currently under c a y , and is pas ing particular to the ingestion of carbon black (with adsorbed carcinogens suc pyrene) ,asbestos, and nitrosamines. A case-control study of 7 2 colorectal cancer deaths and 71 trols, from company A (with an SMR of 173), found no statisti association with specific jobs, although the cases tended to h rubber-batch preparation, milling, and maintenance jobs more th troIs.l'' FIGURE 4 shows the combined-cohorts SMRs for respirato bladder, and central nervous system cancers. With the exception of the prostate. with an overall SMR of 119, no excess deaths from the ot cancer types were consistently observed. Respiratory cancer mortality sistently at o r slightly below the expected figure. which is particular in view of the recent report of an SMR for bronchogenic cancer population of 16,035 British rubber-tire workers (also calculated w national population as standard) .I5 lModerate proponional excesses of lung cancer, however, were observed at a number of other plants. underscoring the important consideration that working environments and their attendant healthhazards may differ significantly between plants. Bladder cancer mortality. although mildly elevated in two of the coho& ( B and C ) . is also slightly below expected in the combined-cohorts SMR. 200 m40-64 m65-84 m a 0 - 8 4 - Company-Speclfic SMR CANCER SITE STOMACH (Is11 C Z (1531 RECTUM 11541 PANCREAS (1571 . FIGURE 3. Standardized mortality ratios (SMR) for deaths due to selected Can cers of the digestive system; white male rubber workers, all 4 companies, 1964- Figures in parentheses after cancer site indicate ICDA (International Classification I Diseases, Adapted) code, Eighth Revision. (See FIGURE 1 le,.end for calculation expected deaths.) x 132 Annals New York Academy oi Sciences -m 4 0 - 6 4 m 6 5 - 8 4 -40-84 Cornpony-SpecificSMR * *n 200 -SMR ! ! 100 NO.OFDEATHS: 33 76 109 NEOPLASMS : ALL L.H.S. NEOPLASMS (200-209) 13 19 32 15 31 46 LYMPHOSARCOMA, L N K E M I A 8 HODGKIN'S DS. (Zc4-zO7) ROO- 200 9 II 20 LYMPHATIC LEUKEMIA I2041 FIGURE 5. Standardized mortality ratios (SMR) for deaths due to neoplasms of the lymphatic and hematopoietic system: white male rubber workers, all 4 companies, 1964-73. (See FIGURE 1 legend for calculation of expected deaths.) control study of leukemia deaths in all plants of company A was camed out.10 For each death from, or with, leukemia, three individually matched controls (matched on plant, sex, race, and age, and known to have died of some other noncancer cause) were selected, and complete work histories were obtained for each study subject. Work histories were coded, blind. according to the general job-type classification system, and were also coded according to the presumed level of solvent exposure in each job, as assessed a priori by the research group's industrial hygienists. The major production stages in the manufacture of rubber tires are shown in FIGURE 6. A number of jobs within the middle stages (boxes 3 through 8) I Compounding Mixing Milling 3 Treads 6 "Cemenr' Mixing 5 I 13 Warehwx . ' .L...." , . ': " < ..- . . , , , L' FIGURE 6. Diagrammatic representation of rubber tire production stages.;.i$--&. McMichael et al.: Rubber Workers 133 rganic solvents, as do some latter jobs in finishing, repair, ese jobs were grouped into high, medium, and low solventJobs involving no apparent solvent exposure were grouped 1 constituents, such as Compounding and Mixing, Milling, myeloid leukemia with solvent exposure did not achieve However, in the case-control analysis of lymphatic leung association with solvent exposure, as summarized in 7. The job categories are shown along the horizontal axis, and the axis is the difference in exposure duration between cases and controls. h of the three solventexposure categories, the cases on average were -I v) a +2 0- E% %` ww O v)Y UU LE z w -2 :e :a 0 -4 W OLVENT EXPOSURE^ NO APPARENT SOLVENT EXPOSURE 1 ESTIMATED as\*,* - RELATIVE RISK^ 5 0 20 2 5 < I G U X 7. Mean difference in duration of exposure between 17 lymphatic leukeCases and 5 1 individually matched controls, to job categories by level of solvent -e. Note: the estimated relative risk of lymphatic leukemia for solvent ed workers was calculated by "discordant matched sets" analysis. sed for several years more than their controls. Overall, &ses experienced rage of eleven years of solvent exposure, compared to four years for S. For all other jobs (except General Service), the controls had had ~ b ? e r average exposure. The association between cases and exposure to General is quite plausible in terms of the solvent-exposure hypothesis. since many this category are itinerant. and-some involve cleaning and degreasing m. estimates of relative risk shown beneath FIGURE 7 are based on dis- nt matched sets analysis, with exposure categorized as a "yes-no" variable. !%.workers in solvent-exposure jobs overall, there was a more than threefold e- in the risk of lymphatic leukemia. For those who worked in the high t-exposure area. the estimated incre-se in risk is fivefold. u" 134 IOC Annals N e v Y o r k Academy of Scfences FIGURE 8. Comparison of 61 cases of lung cancer (ICDA. 9tb Revision, codes 1 163) and 61 matched controls for differences in exposure 10 a n n g process. An association between bronchogenic cancer and the recently been reported in the above-mentioned study of Biirish Although the four individual cohort SMRs for lung GECC'T were 1 tionaI mortality analyses at other plants within company -4identified with 50-100% proportional excesses of lung cancer dezths. From plants, 61 lung cancer cases were identified, and their aork hi pared with those of 61 controls, for differences in expos-are to curing (as we1 as to tire-building, with which an association has also k x recently reported*) Controls were matched on sex, race, age, and plant. Tire-b association, but curing clearly did, both in terms of the controls exposed and in terms of the average duration o cases and controls who had worked in curing (see FIGLX current smoking habits by work area in another tire-rna cates that smoking was unlikely to have been a confa analysis. (The association between lung cancer and curhlg bec when two of the nine exposed controls, first exposed less rhan te death, are eliminated.) Rzplication of this study at other plants i In addition to the above-mentioned case-control studes of 1 lung cancer, each with a specific hypothesis testable in mms of exposures, exploratory analyses were carried out for the job expos of stomach, bladder, and prostate cancers, and lymphosarcoma a disease. These preliminary analyses, done within company A. were comparison of work-histories between each cause-of-deah case the stomach cancer deaths) and an age-stratified random sarn 23 % of the cohort. Jobs in this particular plant were class similar to those in FIGURE 6. Each case group was corn sample for differences in the proportion that had spent each job category. The results are summarized in TABLE 2. Stomach cancer is associated with compounding and mixing, milling, with jobs involving contact with "green rubber" (postmilling but precur i 136 Annals York Academy of Sciences SUMMARY Several previous epidemiologic studies of the rubber industry-an ind that uses many chemicals-have identified excess mortality from certain s cancers. In this study, four cohorts pf active and retired workers, at four rubber-tire plants. were identified historically and followed for the te period from 1964 to 1973. The cancer mortality of these four populations was compared. separately and combined, with that of the general community. For all cancers combined, there was a slight excess above the expected number of deaths, whereas for some specific cancers (stomach, colon, prostate, and n e e plasms of the: lymphatic and hematopoietic system) there was a marked exof deaths. Prbportional mortality analysis at other small plants revealed similar excesses for these cancers, and some excess for lung, bladder, and CNS cancers. .Analysis of detailed individual work histories reveals an association of certain I cancers with specific job exposures; in particular, lymphatic leukemia and solvent exposure, and lung cancer and curing-room exposure. For both bladder cancer and stomach cancer, preliminary analyses indicate an association with * groups of jobs in adjacent production stages (handling and mixing raw ingredi- 1 ents, and processing the "green" precured rubber. respectively). Further analytic studies are currently under way to ,identify groups of rubber workers at increased risk of other specific cancers. ACKNOWLEDGMENTS Of the many individuals in the Occupational Health Studies Group whose - efforts were drawn upon, the authors wish particularly to acknowledge the assistance of R. Spirtas, J. D. Taulbee, D. R. Herman, P. hi. Tousey, and J. F. Gamble. We thank iMs. A. Colosi and her data-processing and programming staff, and also acknowledge the contributions of Ms. Mary Donelan, and Dr. F. S. Jones, Dr. F. Gruber. and Dr. M. J. Symons. REFERENCES 1. REHV, L. 1895. Blasengeschwiiltse bei Fuchsin-.kbeitem. Arch. Klin. Chir. 50: 588-600. 2. CREECH, J. L. & M. N. JOHNSON. 1974. Angiosarcoma of liver in the man facture of polyvinyl chloride. J. Occup. Med. 16(3): 150-151. 3. GENERARLEGISTER OFFICE. 1934. Census of England and Wales. 1931. OcC pations, Table 2. H.M.S.O., London. 4. U.S. DEPT. OF HEALTH, EDUCATION, AND WELFARE. 1961-1963. Mortality IIJ 1950 by Occupation and Industry. Vital StatisticsSpecial Reports. 53: n 1-5. 5. MANCUSO,T. F. 1949. Occupational cancer survey in Ohio. Proc. h b Health Cancer Assoc. Am. :56-70. 6. MANCUSOT, . F., A. CIOCCO& A. A. EL-AITAR. 1968. An epidemiological a proach to the rubber indnstry. J. Occup. Med. 10(5):213-232. 7. MANCUSOT,. F. & M. J. BRENNAN. 1970. Epidemiological consideratiom cancer of the gallbladder, bile ducts, and salivary glands in the rubber ind try. J. Occup. Med. 12(9): 333-341. McMichael el ai.: Rubber Workers 137 MANCUSOT,. F. 1974. Epidemiological investigation of occupational cancers in the rubber industry. Presented Int. Cod. on Occupational Cancers and Health Hazards in the Chemical and Rubber Industry. Geneva, Switzerland. *. October. MCMICHAELA,. J., R. SPIRTAS& L. L. KUPPER. 1974. An epidemiologic study of mortality within a cohort of rubber workers, 1964-72. J. Occup. Med. ',: 16(7): 4 5 8 4 6 4 . MCMICHAELA.. J.. R. SPIRTAS. L. L. KUPPER& J. F. GAMBLE. 1975. T h e relationship of solvent exposure to leukemia among rubber workers: an epi- demiologic study. J. Occup. Med. 17(4): 234-239. MONSON, R. R. & K. K. NAKANO. 1974. Mortality among rubber workers. Paper presented to SOC.for Epidemiol. Res. Berkeley, Calif. June. Unpublished. CASE, R. A. M. it M. E. HOSKER. 1954. Tumor of the urinary bladder as an occupational disease in the rubber industry in England and Wales. Brit. J. Prev. Soc.M e d 8: 39-50. VEYS, C. A. 1969. Two epidemiological inquiries into the incidence of bladder tumors in industrial workers. J. Nat. Cancer Inst. 43: 219-226. ADELSTEINA,. 51. 1972. Occupational mortality: Cancer. Ann. Occupat. Hyg. 15: 53-57. Fox, A. J.. D. C. LINDAR&S R. OWEN. 1974. A survey of occupational cancer in the rubber and cablemaking industries: results of five-year analysis, 196771. Br. J. Ind. lied. 31: 140-151. MCMICHAELA,. J., S. G. HAYNES & H. A. TYROLER. 1975. Observations on the evaluation of occupational mortality data. J. Occup. Med. 17: 128-131. 17. MYERS, R. J. 1961. Mortality of uorkers insured under OASDI. Social Se,: curity Bulletin 24: 11-17. 18. GAMBLE, J. F., R. SPIRTAS& M.Y. LOSGLEY.1975. Job classification and use .' of complete u ork-histories in epidemiologic studies. Paper presented Am. Occup. Med. Assn. San Francisco, Calif. April. Unpublished. 19. BUCK, C. R. 1974. Colorectal cancer in rubber workers: a case-control study. M.A. Thesis. Department of Epidemiology, School of Public Health. University of North Carolina. Chapel Hill, N.C. Unpublished. 20. MANCUSOT,. F. 1963. Tumors of the central nervous system, industrial considerations. Acta Union Internationale contre le Cancer 19: 488-489. 21. VIGLIANIE, . C. 6; G. SATTA1. 964 Benzene and leukemia. New Engl. J. Med. 271: 872-876. 22. VIGLIANIE,. C. This annal. 23. KIPLING,M. D. & J. A4.H. WATERHOUSE19. 67. Cadmium and prostatic can- cers. Lancet i: 730-732. s