Massachusetts mortality survey (Dubrow and Wegman 19841, it must be concluded that to the extent smoking effects were adequately controlled by statistical design, certain occupations appear to be associated with significant risks of cancer, independent of the role of smoking. Unavoidably, the occupational classes used for this kind of study are broad: the actual exposure circumstances within these classes that might be conducive to cancer can only be speculated. Viadana and colleagues (19761 studied 11,591 white male cancer patients and controls who were hospitalized at Roswell Park Memorial Institute between 1956 and 1965. They used occupational histories to classify cases and controls as being exposed to "chemi- cals" and "combustion products," and into specific groups within these broad classifications, e.g., operatives in the chemical industry or bus, cab, or taxi drivers. Operatives in the chemical industry had an increased relative risk of having cancer of the stomach (RR 4.25, p29 years, RR 10.5). The chemical nature of the dyes to which workers were actually exposed was not discussed. A case+control study of 212 cases of bladder cancer in rural Denmark drew upon general population controls matched to cases by sex, age, and region of residence (Mommsen et al. 1982, 1983). Multivariate logistic regression was used to identify factors (evalu- ated by questionnaire) associated statistically with the malignancy. Significant relative risk ratios were found for tobacco use (l&2.1), work with petroleum, asphalt, oil, or gasoline (2.9-3.8), industrial work (2.21, work with chemical materials (2.0), alcohol use (2.3), and previous venereal disease (2.9). Farmers, who were presumably exposed to pesticides, were at less than average risk of developing bladder cancer. The report did not explore the interactive effects of occupation and tobacco use. 392 A case-control study in the greater New Orleans area, wherein 82 patients with bladder cancer and 169 matched general population controls were interviewed by telephone, was used to identify smoking of filter cigarettes (but not of unfiltered cigarettes or other tobacco products) as a risk factor for bladder cancer (Sullivan 1982). Matching criteria were not specified. A large number of employment categories and chemical exposures appeared to involve risks, most prominently mechanical engineers and people exposed to paint thinners, coal, petroleum, metals, welding materials, office supplies, and industrial equipment. In a study in northern New Jersey (Najem et al. 1982), 75 cases were compared with 150 patient controls, matched by age, sex, race, place of birth (in New Jersey or elsewhere), current residence, and the clinic providing care. Occupations were recorded only when job tenure was more than 1 year. Smoking habits were characterized as never smoked, former smoker, or current smoker. Several criteria were used to test the significance of associations, and significant risks were analyzed to test for the confounding effects of smoking. Significant risk ratios were identified for cigarette smoking (2.0) and work in dye (3.1), petroleum (2.5), and plastics (3.4) industries, but not for employment in rubber, textile, printing, rodenticide, or cable industries, although some ratios did exceed 1. When the significant occupational risk ratios were analyzed within the three strata of smoking status, ratios for current smokers were essentially the same as those calculated without controlling for smoking. Curiously, the occupational risk ratios for nonsmokers in the dye, plastics, and petroleum industries consistently exceeded the ratios for current smokers, but the ratios for ex-smokers were consistently lower (1.3 to 1.5). Some ratios were based on only a single case in a smoking- occupation cell. In a recently reported case-control study at Turin, Italy, 512 male bladder cancer patients diagnosed from 1978 to 1983 were compared with 596 patient controls (225 urologic, 371 surgical) (Vineis et al. 1984). Smoking and occupational information was assembled by interview. Highly significant relative risk ratios were found for cigarette smoking, the magnitude depending on smoking intensity, on age when smoking started, and possibly on brand of cigarettes smoked. Occupational risk analysis was based on 64 patients classified as "exposed": 14 cases and 2 controls employed more than 6 months in dye production (said to include exposure to benzidine and betanaphthylamine), plus 28 cases and 20 controls employed in the rubber industry. From this, the authors calculated risk ratios strongly suggesting interactive effects of occupational exposure and smoking, most striking in workers less than 50 years of age, and dependent on smoking intensity (relative risk was 144.0 for hazar- dously employed workers who smoked). The relative risk for 383 hazardous occupation among nonsmokers of all ages (there were only five chemically exposed: two cases, three controls) was 3.7. The relative risk for smoking among the nonexposed of all ages was 5.2. The risk for the occupationally exposed who smoked was 11.6 relative to nonexposed nonsmokers. The risk ratio relationships based on all age groups are more suggestive of an additive overall effect than of synergy. The small number of occupat,ionally exposed nonsmokers in the study limits the confidence that can be placed in the analysis. Smith and colleagues (1985) recently examined relationships of occupational solvent exposure and smoking to incident cases of bladder cancer in regions of the United States served by the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and included in the National Bladder Cancer Study (NBCS) of 1978. Controls selected from the NBCS study were frequency-matched to the cases by age and sex. Among nonsmokers reporting at least 6 month's exposure to chlorinated and simple hydrocarbon solvents used in dry cleaning and in other industries, the relative risk of bladder cancer was significantly elevated only in women (RR 1.38), and this was not significantly related to duration of exposure. In both men and women chemically exposed, however, the risk of bladder cancer increased progressively and significantly from nonsmokers to former smokers to current smokers. There was no evidence of interaction between the effects of occupation and smoking. The evidence of a smoking effect was consistent and substantial in this study (highest RR 4.681, but indications of a solvent exposure effect were weak and of borderline statistical significance. Inhalation of combustion effluents is thought to pose a risk of cancer at various sites, including the bladder. For a case-control study of 81 male bladder cancer patients in Quebec (1970-19751, age- and sex-matched individuals living in the patients' neighborhoods were recruited to serve as controls (Theriault et al. 1981). Detailed histories were taken covering lifetime employment and residence, medications, personal habits, and water supply. Smoking habit was characterized qualitatively and quantitatively. Information on de- ceased cases was received from next of kin. These authors calculated a mean relative risk of bladder cancer of 5.70 among currently smoking workers in the electrolysis department of an aluminum reduction plant. Relative risk for smoking among other workers was 1.82, and risk associated with electrolysis department employment alone was 1.90. (It was necessary to base the latter estimate on ex- smokers because there were no nonsmokers so employed.) The independent risk ratios are not statistically significant, but the range of values for electrolysis workers who smoked extended from 2.0 to 12.30. To the extent that the ratio estimates can be relied 384 upon, they suggest a strong interaction between smoking and occupational exposure. Silverman and colleagues (19831, in a population-based case- control study of 303 white male bladder cancer cases in Detroit (1977-19781, explored occupational, dietary, and personal habit associations. Controls under 65 years of age were chosen from the Detroit population by random digit dialing, operating with a pool of 2,110 households. Controls of retirement age were selected at random from Health Care Financing Administration lists. A total of 296 controls were recruited for home interviews. Analysis of occupational associations relied on an "ever employed" query, and responses were classified according to the industry identified and by occupation. An apparent strong interaction involved truck driving as an occupation and cigarette smoking. Among people who had never been truck drivers, smoking one or two packs of cigarettes per day increased the risk of bladder cancer by a factor of 1.6; smoking more than two packs increased the risk ratio to 2.1. The ratio for truck drivers smoking less than one pack per day was 1.3; for smokers of one or two packs per day, the ratio was 6.8. The risk ratio for heavy smokers could not be calculated because none of the controls reported smoking more than two packs per day. A relationship with the inhalation of diesel exhaust was suspected, but could not be confirmed from the data. Not unexpectedly, the case-control studies used to test a relation- ship of pesticide exposure to urinary tract cancer are inconclusive. Pesticides are chemically and toxicologically diverse; worker expo sures to them are equally varied. It is unlikely that in epidemiologic studies based on broad occupational categories carcinogenic risk would be detected, if indeed it exists. Several case-control studies of bladder cancer have indicated that as an occupational group, people "working in agriculture" are at no more than average risk of urinary tract cancer, or are actually at less than average risk (Anthony and Thomas 1970b; Cole et al. 1972; Howe et al. 1980). In some studies of exposures to pesticides, specifically, nonsignificantly elevated risk ratios have been shown (Najem et al. 1982; McLaughlin et al. 1983). In Canada, bladder cancer was found to be significantly associated with crop spraying and nursery work as occupations (Howe et al. 1980). These associations were said to be unaffected by controlling for smoking. Other Specific Cancer Sites Cancer of the kidney and upper urinary tract has received less attention than bladder cancer. In a nationwide case-control study of 202 patients with renal adenocarcinoma (Wynder et al. 19741, the past personal and occupational histories of the patients were examined in relation to histories from other hospitalized patients. 385 Relative risk among heavy smokers (more than one pack per day) under 50 years of age was 8.0, but only 2.1 in patients over age 50. Moderate smokers (up to one pack per day) were at intermediate risk. Except for a tentative identification of employment in textiles as a risk factor, no associations with occupational exposures to metals, dyes, or other organic chemicals were found. In the Boston area, 43 cases of cancer of the renal pelvis and ureter were studied in relation to bladder cancer cases and randomly chosen general population controls (Schmauz and Cole 1974). Only among smokers of more than two and one-half packs of cigarettes per day did a significant risk appear (RR 10.0). Of the occupational categories identified, only leather working exhibited a suspect relationship to cancer at these sites. Occurrences of renal cell carcinoma (495) and cancer of the renal pelvis (74) in the Minneapolis-St. Paul area from 1974 to 1979 were studied for heritable and environmental risk factors (McLaughlin et al. 1983, 1984). Controls were chosen randomly from the metropoli- tan area population. Because half of the cases were already deceased, background data had to be obtained through next-of-kin interviews. With respect to cancer of the renal pelvis, the risk of disease increased steadily in both men and women in relation to smoking intensity (maximum RR 10.7 in men, 11.1 in women). The only links to occupation appeared in relation to exposures to hydrocarbons: coal, natural gas, and mineral and cutting oils. There were not enough cases to permit analyses for smoking risk within occupations. Renal cell carcinoma also appeared to be related to cigarette smoking, but relative risk ratios were much lower (2.3 in male, 2.1 in female heavy smokers), and the dose-response relationship was not as consistent as in the case of renal pelvis or bladder cancer. Analysis for "usual industry of employment" failed to identify any significant occupational associations. Musicco and colleagues (1982), in a case-control study of brain neoplasms in Italy, sought to associate the occurrence of gliomas (various types and grades) with occupations of victims prior to diagnosis during 1979 and 1980. Forty-two cases were matched with nonglioma patients at the Neurological Institute C. Besta of Milan. The controls were matched by age, sex, and area of residence. They suffered from a variety of chronic diseases, some probably character- ized by physical and or mental disability from a relatively early age. Smoking was defined as a minimum l-year usage, and total lifetime usage was estimated. More than 20 pack-years was considered heavy smoking. The authors found a significantly elevated risk ratio (5.0) for "agricultural work after 1960" when the data were analyzed without stratification. When stratified by sex, age, and residence, the ratio was 1.9 (not significant). The relative risk was 1.3 for smoking and 1.6 for heavy smoking, neither statistically significant. No occupational risk ratios for nonsmokers were calculated. Particular- ly with respect to occupational risk calculations, the appropriateness of neurologically afflicted patients as controls must be questioned. Nonetheless, the authors were inclined to indict modern pesticides and fertilizers as causal factors for gliomas. Austin and Schnatter (19831, in a followup case-control study of 21 patients dying from a brain tumor in a Texas petrochemical worker cohort, indicated that the tumors were of several different types. Efforts to identify unique past chemical exposures of brain tumor victims were not successful. Using case-control methodology in reviewing 142 cases of pancre- atic adenocarcinoma in several large U.S. clinical centers, Wynder and colleagues (1973) demonstrated that cigarette smokers were at increased risk of developing this disease. Risk ratios increased progressively with the number of cigarettes smoked per day. Controls for this study were patients in the same hospitals who had been interviewed for other epidemiologic studies. Controls did not include patients suffering from tobacco-related cancers (mouth, larynx, lung, esophagus, bladder, kidney) or other tobacco-related diseases (bronchitis, emphysema, coronary heart disease). Fifteen male cancer patients reported having been occupationally exposed to "dyes, chemicals, metal dust, saw dust, grease, oil, or gas fumes," but there was no difference between cases and controls with respect to the frequency with which this exposure was reported. A case-control study in New Jersey (Stemhagen et al. 1983) of 265 victims of primary liver cancer occurring from 1975 to 1980 was conducted by interview of family members. Controls were selected from hospital records and death certificates, and were matched by age, sex, race, and county of residence. No evidence was adduced to indict smoking as a factor in causing this disease. Significantly elevated risk ratios were derived for farm laborers but not for farm owners or farm managers or for people engaged in manufacturing pesticides. Other people apparently at risk were gasoline service station employees, those employed at eating and drinking establish- ments, and those providing laundry and dry cleaning services. It was not possible to identify specific past chemical exposures that might have contributed to the risk. A recent study of 102 cases of primary liver cancer in Sweden utilized controls matched by age, sex, race, year of death, and municipality where the decedent had lived (Hardell et al. 1984). No association with smoking history was found. Occupational exposure to solvents appeared to double the risk of liver cancer. No other occupations or chemical exposures were identified as risk factors, although a strong association with alcoholism was indicated. Investigation of 207 cases of large bowel cancer in a Quebec community explored several risk factors in cases and controls, the 387 latter selected from the communities where the cases resided, and matched by age and sex (Vobecky et al. 1983). Smoking was not identified as a significant risk factor, although a slightly elevated risk ratio (1.2) for smoking (alone) was calculated. Industrial exposure at a local synthetic fiber factory did appear to be a significant association (RR 2.2). When the risk of industrial exposure and smoking were considered in combination, a higher risk was evident (2.81, at a stronger level of significance. A moderate degree of smoking+ccupation interaction is suggested. Chronic Lung Disease The likelihood that exposure to dusts and fumes in rubber product manufacture plays a causative role in the chronic obstructive lung disease encountered in this industry was examined in two studies. Fine and Peters (1976) assessed symptomatology and pulmonary function in 65 white male workers engaged an average of 7 years in rubber processing at three Akron tire plants. Air sampling showed 1 to 3 mg/m3 of respirable dust in the work environments. Smoking habits were classified as never smoker, former smoker, current cigarette smoker, and current and former pipe and cigar smoker. Controls (189) were chosen from plant workers not exposed to polluted air. Processing workers reported a much higher prevalence of cough and phlegm than controls; this was true among nonsmokers as well as smokers in the various categories. Smoking nearly doubled the frequency of this symptom complex in the processing workers. However, dyspnea and wheeze, generally considered indicative of chronic obstructive lung disease, were no more prevalent among processing workers than among controls. Reported frequencies of bronchitis, pneumonia, asthma, and winter colds were not signifi- cantly greater among these workers than among controls. Pulmo- nary function testing yielded important findings. In comparing all workers with all controls, only the ratio of forced expiratory volume in 1 second to forced vital capacity (FEV,/FVC) was significantly lower in the particulate-exposed workers. However, the exposed group and the control group were subdivided according to whether they had been employed in their respective jobs for more or less than 10 years. Although the long-term processing workers were older and had smoked longer than the controls, decrements in flow rates and FEV, /FVC were not significantly correlated with years of cigarette smoking. Using appropriate adjustments for age, the long-term- exposed employees exhibited significant deficits in FEVI, FEV,/FVC, and flow rates at 50 and 25 percent of FVC. Multiple regression analysis confirmed that duration of employment in rubber processing was a significant predictor of reduced FEV, and FVC. Employment for more than 10 years appeared to cause a significant decline in FEV,/FVC and the FVC-standardized flow rate at 50 percent FVC. These results were independent of smoking variables, ethnicity, socioeconomic status, and age. The absence of a correlation between decrement in lung function and cigarette smoking and the small number of workers in this study raise questions about the generalizability of the data in this study. Lednar and colleagues (1977) examined the work history and smoking habit backgrounds of 73 former rubber workers who were retired prematurely between 1964 and 1973 with medically docu- mented, disabling pulmonary disease. They were members of a cohort of 4,302 workers employed in 1964 at an Akron plant. Thirty- nine were retired with emphysema, 10 with lung cancer, 8 with asthma, and 16 with other pulmonary conditions. Work background and likely exposure to dusts and fumes were determined from company records; smoking histories were obtained from question- naires mailed to retirees or relatives. The investigators utilized two control groups, the first consisting of disabled employees retired because of diseases other than pulmonary (disabled controls) and the second of currently employed workers and early retirees free of acknowledged pulmonary disease (nondisabled controls). Relative risk ratios were calculated for smoking and for occupational exposures to dusts and fumes. Relative risks for pulmonary disability retirement in relation to smoking and various occupational titles were also calculated. Risk ratios for smoking alone (based on smokers and nonsmokers at worksites not otherwise contaminated) were consistently greater than 1.0, but they were significant only in the case of maintenance workers. For all workers combined, the smoking risk ratio was 2.95 (p