Document 0DqJ5mMrvj9BQQQn1Kz4Kzdd

Mortality Among Rubber Workers: Relationship to Specific - A. J. McMichael, M.D., Ph.D.; R. Spirtas, Dr. P.H.; J. F. Gamble, Ph.D.; and P. M. Tousey, B.S.N. Jobs I n an earlier article, the nine year mortality experience of a population of 6678 male rubber workers, during 1%472, was described.' As a group, these active or retired production workers, from a single large tire-manufacturingplant in Akron, Ohio, experienced death rates from various specific causes of death that differed noticeably from those prevailing within the national male population. In particular, excess mortality occurred from cancers of the stomach, colon, prostate, and lymphatic and hematopoietic system, as well as from diabetes mellitus, arteriosclerosis, and (in the 40-54 age-range) ischemic heart disease and stroke. The question immediately arises: are these mortality excesses related to specific jobs within 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 which remain unidentified. Among the chemical ingredients now or previously used are various proven or suspected carcinogens, such as beta-naphthylamine, benzene, asbestos (in talc dust), and various nitrosamines. The primary focus of this article is the analysis of the association between excess mortality and specific jobs within this same plant. These analyses have been carried out for specific causes of death observed to be in excess in the study population, and also for two causes reported to have been in excess in earlier studies of the rubber industry (bladder cancer and lung cancer).? 3 4 The cases studied here come from deaths in the ten year period 1%4-73. The identification of the study population, and its mortality experience, has been described in detail.2 Thirtyeight percent of the population was aged 40-54 at 1/1/64, 30% was aged 55-64, and 32% was aged 65-84. Eighty-six percent were white males, and 14% were black males. Forty-eight percent began work in the plant at least twenty-five years before 1964, and all but 1% had worked for at least ten years by 1%4. The Figure summarizes the 1964-73 -SMR mortality experience of the total study population, presenting age-racestandardized mortality ratios (SMR), These ratios were calculated by comparison with the 1968 U.S. male population mortality. This particular comparison of a working population with the general community may introduce a conservative bias, resulting in some underestimation of the mortality attributable to occupational factors. However, this selection bias, known as the "healthy worker effect", usually a p plies less to cancer than to other causes of death. The equivalent mortality analysis within a second, smaller population of 1339 male rubber workers, during 196473 from another tire-manufacturing plant in Akron, also identified mortality ex- From the Occupational Health Studies Group, k h w l 01 Public Health, University 01 North CarcJina. Chapel Hill. Reprint requests to D ~ p lol Epidemiology. School ol Public Health. University 01 North Carolina, Chap1 Hill, NC 27514 (Dr McMicharl). 178 -F i t 1. Standardized Mortality Ratios' in population of 6.670 Male Rubber Workers, 1964-73. Ages 4064. Ex. petted deaths were calculated using 1968 U.S. race-i;e-spedlic death rates. CPSSes elevated popalati was 119 lymphati was 176 of 82, ex from de (SMR ec (SMR ec To ch parent c reflected cesses, I compare Ohio ar (which available in Table rates ten( state or I are not g for by thc County d minated during tt Method The r vironmer and che dividual time in working ber of q the plan major w Note t h i listed in producti In ord plete e> dividual were ob office. E different changinf with tim job clas, of genr studies hypothe# Objec specifiec dustrial jobs int Each 0 jobs thi material, requird jobs at OT's W I Journal c cesses from the two cancers most elevated (Figure). Within this second population, the SMR for stomach cancer was 119, and the SMR for cancers of the lymphatic and hematopoietic systern was 176. Relative to the all-causes SMR of 82, excesses were again also observed from deaths due to diabetes mellitus (SMR equals 102) and arteriosclerosis (SMR equals 145). To check whether any of these apparent cancer mortality excesses merely reflected local or regional mortality excesses, the national death rates were compared with the rates for the state of Ohio and those of Summit County (which includes Akron). The figures, available only for 1950-69, are presented in Table 1.5 While the Summit County rates tend to have been higher than the state or national rates, these differences are not greater than could be accounted for by the estimated 20% of male Summit County deaths contributed by active, terminated and retired rubber workers during that period. -Source: Mason, T.J.. W a y . F W. US. Cancer Moctalrty by County: 1950 69.5 Methods The rubber industry working en- vironment is very diverse in its physical and chemical exposures. Further, an individual rubber worker typically spends time in 15-20 different jobs during his working career, often embracing a number of quite different work areas within the plant. A summary description of the major work areas is presented in Table 2. Note that the first eleven categories are listed in the order they occur on the production line. In order to take account of the com- plete exposure experience of the individual worker, detailed work histories were obtained from the plant personnel office. Because of the several hundred different jobs within the plant, and their changing nomenclature and numbering with time, it was necessary to develop a job classification system that would be of general utility in epidemiologic studies of both thi exploratory and hypothesis-testing kind. Objective criteria were therefore specified by the research group's industrial hygienists, for the grouping of jobs into "Occupational Titles" (OT). Each OT contains functionally similar jobs that .entail working with similar materials and processes. Sixty 0 7 s were required, to embrace all present and past jobs at this particular plant, and these OT's were therefore used in classifying the contents of the work-histories in this study. OT codes were assigned blind to all jobs listed on each work-history, without knowledge of the vital status of the study subject. A detailed discussion of this job classification appears elsewhere in this issue. Because of the high costs and time delays involved in the collection, coding, and data-preparation of detailed work-histories, a study design was used that required work-histories for only a minority (about one quarter) of the study population. This design enabled each case group of interest (e.g. stomach cancer deaths) to be compared to a sample of the total study population for differences in work-histories. Once this initial "case-control" comparison had identified differences between the cases and the population sample, in frequency and duration of employment within specific job categories, the design then enabled an estimate to be made, in cohort analysis fashion, of the risk ratio associated with that particular job category. The rationale and theoretic statistical basis for this hybrid (ambidirectional) study design, and its risk estimates, have been discussed by Kupper et al.6 Potential confounding factors in this study (i.e. risk indicators of the cause of death under investigation, that are also likely to be associated with the kind of Journal of Occupational Medicineflol. 18, No. 3/March 1976 job done within the rubber industry) were sex, age and race. Sex was controlled by restriction of the study to male workers; age by age-stratification and age-adjustment procedures; and race by subsequent review of the data. This latter approach was necessary because information on race was not initially available on much of the study population, at the time of sampling. Smoking, known as a risk indicator for cancers of the lung and bladder, and for chronic cardiovascular disease, was not an obvious confounding factor, since recent cross-sectional studies in two other tire-manufacturing plants have demonstrated no strong association of smoking habits with job category. Unfortunately, satisfactory retrospective information on smoking habits was not available for study subjects in this present study. Using the sample size formulae of Kupper et al.,6 applied to predicted exposure frequencies and mortality effects, sampling fractions of 9%. 13%. and 14% were indicated for the three age strata in the study population (40-54, 55-64, 65- 84). As both a precautionary measure against erroneous predictions of exposure frequency, and knowing that the sample would be used for multiple comparisons, the actual sampling fractions used were 20%, 25%, and 23% (representing 500 in each age stratum). Com- 179 .. IC plete work-histories were siibsequently obtained for 1482 (99%) of this agestratified random sample. In the initial case-control comparisons, the OT's were grouped, a priori, into 16 broader groups of similar jobs, known as Occupational Title Groups (OTG) (Table 2). Frequency and duration of employment within these sixteen OTC's (henceforth referred to as "exposure rates", to accord with con- ventional epidemiologic terminology) were then calculated for each study subject. The exposure rates in each case group were age-adjusted (by the direct method) to the age-distribution of the population sample, thereby ensuring mutual comparability of the adjusted rates in each case group. In the analysis of exposure rates, certain criteria were stipulated with respect to time and duration of exposure. Firstly, cumulative exposures of less than two years to an OTG were excluded from this analysis. (In view of the chronic and complex etiology of the causes of death under investigation, this criterion might forfeit some relevant exposure data; yet preliminary analyses indicated that this criterion was desirable since i t eliminates considerable "noise" from the work-history data due to the many instances of brief job tenure.) Secondly, in the analyses of cancer deaths, discussion is here restricted to exposures that occurred during the presumed etiologic period, 1940-60. Current information indicates that for environmentally-induced cancers in adults, in general, the average latency period between first exposure and disease diagnosis i s in the range 1030 years.' The 1940-60 stipulation means that for the "average" death in this study (occurring in 1%8), the period between first exposure and death could range from 8 to 29 years. Analyses of exposures outside the 1940-60 period showed that in no instance did the strength of association of cancer with an OTG exceed those presented below. Results and Discussion Table 3 presents, for the seven cancer case groups, the ratios of their ageadjusted OTC-exposure rates to the rates within the population sample. These ratios provide a convenient preliminary measure of association between cancer type and OTC. Since the proportions of the population sample exposed to any one OTG are low (typically 5-IS%), hence the ratios presented in Table 3 are 180 an approximation for the more conventional odds ratio, which in turn i s an approximation for the risk ratio.8 Thus, the ratios in Table 3 provide a useful, age-adjusted approximation of the increased risk of death from a specific Mortality Among Rubber WorkerslMcMichael et al cause posure the ot Tables ship, e Two and 5 to dei relatior, ticular I a gener modest binatio than 2 ratio w For associa in the shippin mill-mi Table 2 the du, organic Journal I 'Tha rate (proportion) d prior upmure to a specific OTG. lor u c h case group. was ageadjusted directly to the agcdntfibution d the population sample. tColum figurer foc the popuhlim sample are ra&d exposure (rounded OH). fa example. 6% d the population vmpk had spent at lust Iwo years in the RKeiving and Shipping OTC. $Column figurrr f a case groups are-d exporure rates (Le. case group proporlion:population sampk proportion). Thus far OTC I.82 d stomach cancer cafes had spent two years there. wnus 6% 01 the popuhti sample. ginng a ratio ol 1.3. cause among workers with prior exposure to a particular OTG, compared to the other workers. (Comparison of Tables 3 and 5 illustrates this relationship, empirically.) Two minimum exposure durations (2 and 5 years, during 1940-60)were used to detect underlying dose-response relationships. However, using these particular exposure criteria, the evidence for a general relationship of this kind is only -modest of those case-OTG com- binations with at least one ratio greater than 2.0, 12 out of 20 show a higher ratio with the longer exposure duration. For stomach cancer, there is an association with prior exposure to jobs -in the first four OTC's receiving and shipping, compounding and mixing, mill-mixing, and milling. As indicated in Table 2, these OTG's entail exposure to the dusts of dry ingredients (including organic and inorganic pigments, carbon black, and talc dust), and, in mill-mixing and milling, to the fumes and vapors produced by heating the newlycompounded rubber batch. Stomach cancer was also strongly associated with extrusion and tread cementing, as was also colorectal cancer. A moderate association existed with building tubes and flaps (talc dust exposure), curing (fumes), and inspection and repair (rubber dust and solvents). A strong association with history of employment in the synthetic plant was also observed, although the numbers exposed were small (2-3% of the population sample had worked there). A stomach cancer mortalily excess in rubber workers has been reported in other plant populations in the U.S., suggesting that there is indeed a real work-related hazard involved.9 10 Two other studies have reported, respectively, a n association with rubber "processing" (compounding, mixing, milling and calendering), and with compounding, milling, calendering, and tube and tire building.10 11 The inhalation, and subsequent ingestion, of particulate matter is one possible mechanism of stomach cancer hazard contact in these work areas. Colorectal cancers showed fewer associations with OTC's than did stomach cancer. Strong associations existed with extrusion and tread cementing, and, to a slightly lesser degree, with maintenance jobs. A moderate association with inspection and repair jobs was also observed. Weaker associations (for exposures greater than five years) with compounding-and-mixing and milling resemble those reported by Monson and Nakano.10 Lung cancer mortality showed a moderate association with each of the first six OTC's. These associations Journal of Occupational Medicinehol. 18, No. 3/March 1976 181 - -. 8. -Table 4. Ralios of i;c-adiasted' rites of exposure to rprcllic`-g&h?r non.wnur uu irwps compircd to study population u m p l h `The rate (prwwtion) 01 priw exposure lo a s p ~ i l i cO K la uch u s e gfwp. was aleadlusted dlrectly to the apdatrlbutom 01 Ihe populrli~nsample :Column figures for the population sample a r c w o l expos~re(rounded dfi tColumn figures lor u s e pJups arc @of e z r e rates (i.e. use [roup propoclion: populrli sarnpk proportion). Thus kr OTG 1. 12% ol ischemicheart disaase uses b d spent two wrs there. versus 9% 01 the papulation umpk. giving I ratio of 1.3. `Ratios are crlculaled lor IHO deaths at age 40 54 only. where Ihe signihunl morlalily e m s s ~ccurred tended to be stronger for the fiveyear !nir?i~~m-C!uE!te!x~posnre. This observation is consistent with the notion of a lung cancer hazard from prolonged exposure to dusts, fumes, and vapors. However, there was a negative association with exposure to the curing process - an unexpected finding, in view of recent reports of a lung cancer mortality excess among curing workers, both in the U.S. and Britain.` 10 11 Nor was there any association with tire-building, as found by Mancuso.11 There was, however, a strong association with the reclaim operation, entailing some exposures similar to those in the compounding, mixing, and milling OTC`s. A separate case-control study of lung cancer deaths within three other plants belonging to the same company has revealed an association of exposure to curing with lung cancer.9 Twenly-five percent of 61 lung cancer cases had WML"' ,,,; n rL.u. r,:,n,,n5, rLeLm,,pared !a ? % cf their pair-matched (age, race, sex) controls. Further, the exposed cases had experienced more than twice the duration of employment in curing than that of the exposed controls. Cancer of the prostate gland showed few positive associations, being strongly associated with the calendering and the janitoring-trucking OTC`s, and moderately associated with compounding and mixing. This latter association is of interest since cadmium compounds are used as vulcanization (curing) accelerators. Cadmium has previously been found to be associated with prostate cancer.12 The significance of the excess prostate cancer mortality i s underscored by the finding that mortality from prostate cancer has been found to be elevated in three of four major U.S. rubber worker cohorts recently studied.9 Gnce: Gf the bladder ha: long k n recognized as a problem within the rubber industry.3 73 The early cause was probably an organic antioxidant added to the rubber mix, beta-naphthylamine, although subsequent concern has focussed on alpha-naphthylamine, benzidine, and four-aminobiphenyl (xenylamine). Bladder cancer, in Table 3, is associated with three of the first four OTC`s, where the possibility of contact with carcinogenic raw-ingredient chemicals and the vapors of reaction products i s high. It i s also of interest that associations with the receiving-shipping-andwarehouse group of jobs have previously been twice reported elsewhere.10 14 A strong association also existed between bladder cancer and employment in the reclaim OTC. It should be noted, however, that the number of 182 Mortality Among Rubber WorkerslMcMichael et al . bladdf thirteei Can hematc with fi ping, spectic plant, phatic of canc these s. -and shi cale cemenl phatic nantly here, k main fi, associa! history (includi industp jobs cit. mixing, tire-buil These F associai study. Table mia w i with wc the rati cancers strongl\ Althoug were fc this OTI of the predorr Hodgkii posure. leukemi OTC is (and th also), v rapidly ployed remissic have bc medical Table for t w c ischemi mellitu! not re "latenc posure Table 3 at any Deat I the agi particul Journal bladder cancer cases studied was only thirteen. Cancers of the lymphatic and hematopoietic system were associated with five OTG's - receiving and ship ping, compounding and mixing, inspection, finishing and repair, synthetic plant, and janitoring-and-trucking.Lym- phatic leukemia, a subset of this group of cancers, was associated with four of these same five OTG's (minus receiving and shipping), and also with two others - calendering, and extrusion and tread cementing. A case-control study of lymphatic leukemia deaths, based predominantly on the fourteen cases analyzed here, has been reported earlier.15 The main finding of that study was a strong association of lymphatic leukemia with a history of exposure to organic solvents (including benzene) within the rubber industry. The particular solvent-exposure jobs cited in that study included cementmixing, calendering, t read-cementing, tire-building, and finishing and repair. These probably account for some of the associations observed in this present study. Table 3 shows that lymphatic leukemia was also very strongly associated with working in the synthetic plant, and the ratios indicate that the remaining cancers in this group were even more strongly associated with this OTC. Although the other types of leukemia were found not to be associated with this OTC, there was a strong association of the lymphoma group (comprising predominantly lymphosarcoma and Hodgkin's disease) with this type of exposure. The association of lymphatic leukemia with the janitoring-trucking OTG is not readily explainable; however (and this could apply to prostate cancer also), workers with cancers that are not rapidly fatal, if remaining actively employed while in apparent disease remission or stabilization, could well have been transferred into this OTG for medical reasons. Table 4 presents similar information -for two non-cancer causes of death ischemic heart disease, and diabetes mellitus. Since these chronic diseases are not regarded as having a formal "latency" period, a third category of exposure has been added to those used in Table 3 (viz. at least two years exposure at any time). Deaths from ischemic heart disease, in the age-range 40-54, were associated particularly with a history of working in -Table 5. Age-standardized risk ratios (relative risk) lor death from selected -I . causer, among workers with at least fiw years expowre to a rpcalic work area, during 194060, relative to all other workers. .,C a r d DNU Canar I- I Bladder Cancer Lymphatic 6 ~topwtK Cancer Lymphatc Leukemia I~hmIC Hull hwse Diabetes Mettitur Occupational 7 t h boup ( a Number Code) Compounding miring cement mixing (2) YIiIImaing (3) Extrusion. tread amcnhng (6) Synthetic plant (15) Extrusion, tread cementing (6) Maintenano (12) Rwtving and shipping (1) Compounding miung cemcnt miung (2) Miilmiaing (3) Edrusion. tread cementing (6) Reclaim (14) Compatnding mung cement mixing (2) Catenduine (5) Janitorin& truckin& dc. (16) Compounding mixin& ament miaing (2) Milling (4) Reclaim (14) Recaving 6 shippa; ( I ) Compounding mung aannt miring (2) Inspection, finishing repair (1 1) Synlhchc plant (15) Compounding miung cement miaina (2) Eattuuon. tread amenling (6) Inspection. finishint repair ( I 1) Synthetic planl (15) lanita~ing. IruckinL etc. (16) Receiwng 6 shippint (1) Miltlng (4) Eatruuon. tread aawnbng (6) Maintmana (I?) Synthetic plant (15) Inspaction. h n i s h w repair ( I 11 MKhafliUI products (13) Janitaing trucktng etc. (16) i Punt Estinutr' 2.0 2.0 2.3 22 2.2 .. 1.8 , 1.9 1.4 2.1 1.1 23 1.6 24 3.5 1.3 96 2.5 1.8 14 2.0 6.2 1.3 3.2 37 39 2.2 1.5 14 3.3 I .8 3.0 28 1.5 4.3 , Ratio 99.9% CdidtlCB lnlerral+ 1.6.2.9 1.34.0 1.7.3.6 1.4.4.3 16.3.4 1.4.2.9 1.9.3 3 2.84 6 0.83.8 1.64.0 1.84 t 13.2.7 1.1 2.0 1.5.2.9 4.1.12.5 1.0.1.8 2 4.5 0 284 3 2640 18.2.9 .9-3 9 9.3 4 1 8.14 9 1.24 6 1.7-12 5 2.14 0 1.3-1.9 3.44 6 ~~~~ ~~~ 'lhesc r i l l ratio pcint eslimales are conrervalive since the "unexposed" cotnpaiison group includes workers eapoacd to other hiphiisk OTG's (loc that same cause d death) -1he 999% conhdence mlervals are relatirdy mall since the risk estimates apply only to this particular study pPpulalion 6 the extrusion and tread-cementing OTC, and in the synthetic plant. Moderate 'associations were observed with receiving and shipping, milling, and maintenance jobs. Since the deaths studied here were restricted to those in younger workers, i t seems unlikely that any of these associations could be accounted for by selective transfer (self- or company-initiated) of men with chronic cardiovascular limitations. Deaths from diabetes mellitus were strongly associated with working in the inspection and repair OTC and the janitoring-trucking OTC. The latter association could result from selective transfer of less fit individuals. Although moderate dose-response trends exist, in Table 4, for these associations for ischemic heart disease and for diabetes mellitus, yet no readv explanations are apparent. Since mortality rates within the U.S. male population, for many diseases, differ between whites and nonwhites, and since there have been differences in the types of jobs done by whites and nonwhites within the rubber industry, accordingly there is a possibility that some of the abovementioned associations are partially spurious. To determine the extent of confounding by race, Tables 3 and 4 (exposure duration greater than five years, during 1940-60) were recalculated for whites only. fewer than 10% of the exposure ratios underwent any significant change when restricted to white males. further, the majority of these 15 changes were in cells containing small numbers, pertaining to the Journal of Occupational MedicineNol. 18, No. 3/March 1976 183 minority of OTG's with low exposure rates in the population sample andlor to the less frequent causes of death (e.g. bladder). Of the other six changes, two examples will illustrate possible race confounding. In many of the sixteen OTC's, there are greater than twofold differences in the proportions of whites and blacks who worked there for at least five years. In particular, 3% of whites and 27% of blacks in the study population sample had worked in the compounding and mixing OTG. For lung cancer, the exposure ratio (in Table 3) for this OTG was 1.4,while it was 0.8 for whites only. For prostate cancer, these figures were 1.5 and 0.9, respectively. Both sets of figures indicate that, for blacks within this OTG, the ratios must be higher, and that the two ratios in Table 3 (1.4and 1.5) are clearly influenced by the racial composition of the workforce in that work area. Nevertheless, for true confounding to occur, it is necessary that the "background" disease frequency (discounting any occupational exposure effect upon the disease) should "naturally" differ between white and black rubber workers. Whether this difference exists is difficult to determine. Lilienfeld et al. report that, among U.S. males, 1959-61, the mortality ratio, for nonwhites compared to whites, is 0.98 for lung cancer, and 1.69for prostate cancer.16 However, these ratios are not controlled for race differences in socioeconomic class and occupation, and therefore may themselves be confounded by these two factors. Thus, whereas these two ratios from the general population suggest a likelihood of race confounding in the association of prostate cancer and the compounding and mixing OTC, and no such likelihood for lung cancer, yet within this particular occupational class of persons this likelihood of confounding could be quite different. The other four major changes between ratios in Table 3, and for whites only, each indicated an increased risk for whites. The exposure ratios increased from 1.4 to 2.7 for lymphatic and hematopoietic cancers in compounding and mixing, from 0.9 to 1.4 and 1.2 to 2.0 for colorectal cancer and prostate cancer, respectively, i n inspection, finishing, and repair, and from 0.9to 1.5 for colorectal cancer in the maintenance OTG. However, for lymphatic and hematopoietic cancers and colorectal cancers, the mortality ratios for nonwhites compared to whites in the U.S. male population are 0.9 and 0.8, respectively, suggesting little likelihood of confounding.16 Overall, the problem of race confounding appears to be minimal within these data. Howwer, additional racespecific mortality data are being collected at other similar plants, and it will then be possible to determine whether the several abovementioned differences in race-specific death rates associated with particular work areas are consistent across plants. Table 5 presents risk ratios (relative risks) for combinations of cause of death and OTC, selected from Tables 3 and 4. Selection was based on the magnitude of the exposure ratio and its consjstency across exposure duration-time categories. Point estimates and confidence intervals were calculated using the formulae discussed in Kupper et al.6 Further, since multiple intervals were calculated, the confidence level for each interval was set at 99.9% (i.e. a one-inthousand probability of the true parameter value lying outside the calculated interval), thus ensuring an overall a level (based on all 34 calculated intervals) of about 0.04. With this level of statistical confidence, Table 5 indicates that, in all except five instances, the true value of the risk ratio within this particular population of rubber workers exceeds 1.0. That is, in each remaining instance, there was apparently a real increase in risk of death from the indicated cause following five or more years working exposure, during 1940-60,to the listed job catqory. Summary This study has examined the ten-year mortality in a single-plant population of 6678 male rubber workers, in terms of the association of specific causes of death with a history of having worked in certain categories of jobs within the rubber industry. The work-histories of individual study subjects were analyzed, in detail, for all workers dying of selected causes of death. Comparison was made with the work-histories of a 22% age-stratified random sample of the total population. Age-adjusted exposure ratios (Tables 3 and 4) were calculated for all nine case groups in all 16 work areas, using differing exposure criteria (i.e. duration and calendar period), These ratios provide an approximation of the increased mortality risk associated with particular work areas. The risk ratios (with their associated confidence intervals), in Table 5, provide more rigorous estimates of these instances of increased mortality risk. For each cause of death studied, there were statistically significant associations with several work areas. For the cancers, the strongest associations tended to be with work areas at Ihe front end of the production line (especially compounding and mixing), where the likelihood of contact with dusts, chemical ingredients, and vapors containing the early reaction byproducts, is high. The reclaim operation and the synthetic plant were each associated with several cancers (respiratory and bladder, and stomach and lymphato-hematopoietic cancers, respectively). The lymphatic leukemias were associated with solvent-exposure areas, especially inspection, finishing, and repair. Ischemic heart disease deaths, at ages 40-54, were strongly associated with having worked in extrusion and tread cementing, and in the synthetic plant. Deaths from diabetes mellitus were strongly associated with the janitoringtrucking category, and with jobs in the inspection, finishing and repair area. These observed associations, calculated after controlling for the variables sex and age, were apparently not due to confounding by smoking and race differences between work areas. The role of selective transfer of sick individuals (into, say janitoring or trucking) warrants further investigation. However, this mechanism is unlikely to be involved in the great majority of the observed associations. Another possible source of spurious inferences of increased risk would be correlation, within workhistories, between two job categories, one of which actually involved increased risk, while the other did not. Such associations could occur if certain job sequence trends existed within this working population. (Preliminary analyses indicate that this likelihood is not great; howwer, further analysis is planned.) Detailed study of specific environmental agents, historically and crosssectionally, is currently underway, in an attempt to identify the workenvironment agents responsible for the associations reported here. 184 Mortality Among Rubber Workers/McMichael et al thi of1 Ki an of re1 lo1 RI eP co 16 eP dL ur Ih If Joui 2 The following deserve thanks for their assistance: occupational cancer in the rubber and 10. Monson RR, Nakano K K Mortality < the corporate medical director and other company cablemaking industries: results o f five-year officials for assistance in obtaining the data, Or. L. 1. analysis, 1967-71. Brir I lndusrr Med 31:140- among rubber workers: Preliminary report. Paper presented to Society for Epidemiologic r 1 5 Kupper for statistical advice, the data-proces,ing staff and Ms. Pandora Ray for tabulation and calculation of results, Dn. R. L. Harris and H. A Tyroler for reviewing the manuscript, and Ms. Wanda Dodson 151. 5. Mason TI, McKay IW US. Cancer Mor- tality by County: 1950-1969, DHEW Publica- Research, Berkely, California, lune, 1974. 11. Mancuso TF: Epidemiological in- vestigation of occupational cancers in the rub- I for patient typing of text a n d tables. tion No. (NIH) 74-615, 1975. 6. Kupper LL, McMichael AI, Spirtas R A ber industry. Paper presented to the Internalional Conference on Occupational Can- hybrid epidemiologic design useful i n cers and Health Hazards in the Chemical and estimating relative risk. 1 Am. Sraf Assoc Rubber Industry, Geneva, Switzerland, Oc- 1 70:351, 1975. tober, 1974. 7. Armenian HK, Lilienfeld AM: The distri- 12. Kipling MD, Waterhouse IAH: Cadmium References bution of incubation periods of neoplastic and prostatic cancers. Lancer 1:730, `1967. ? 1. McMichael AJ, Spirtas R, Kupper LL: An diseases. Am @id 99:92-100. 1974. 13. Cole P, Hoover R, Friedell CH: Oc- ? epidemiologic study of mortality within a 8. Cornfield I: A method of estimating com- cupational and cancer of the lower urinary cohort of rubber workers, 1964-72. /OM parative rates from clinical data. Applications tract. Cancer 29:1250-1260. 1972. ? 16~458-464,1974. to cancer of lung, breast and cervix. / Nar 14. Veys CA: Aetiology of tumors of the 2. Mancuso TF, Ciocco A, El-Altar A A An Cancer lnst 11:1269-1275, 1951. urinary bladder. Urol Inr 24:276-288, 1969. epidemiological approach to the rubber in- 9. McMichael AJ, Andjelkovic DA, Tyroler 15. McMichael AI, S1)irtas R, Kupper LL, dustry. IOM 101213-232, l%8. HA: Cancer mortality among rubber workers: Gamble IF: Solvent exposure and leukemia > 3. Case RAM, Hosker ME: Tumor of the An qiiclmmiologic 5lUdy. Annuals of the N.Y. atniing rubber workers: an epidemiologic urinary bladder as an occupational disease in Academy oi Sriencrs. iii lm'$s. Prestmld at study. /OM 17:234-239, 1975. 1 the rubber industry in England and Wales. Brir the N.Y. Acddcriiy o f Science\ Conlcrcncc on / Prev b c Med 8:39-50, 1954. Occupational Carcinogenesis, New York, 16. Lilienfeld AM, Levin ML, Kessler II: Cancer in the United States. Harvard University 4. Fox AI, Lindars DC, Owen R A survey of March, 1975. Press, Cambridge, Mass., l972. Journal of Occupational MedicinelVol. 18, No. 3/March 1976 185