The Copenhagen case-referent study on bladder cancer. Risks among drivers, painters and certain other occupations.

KNUDSEN .IB, SORENSEN BL. The Copenhagen case-referent study on bladder cancer: Risks among drivers, painters and certain other occupations. Scand J Work Environ Health 13 (1987) 129-134. Occupational risk of bladder cancer (including papilloma) was investigated as part of a case-referent study in Copenhagen. Occupational histories were obtained on 389 cases and 790 referents drawn at random from the general population of the study area. When persons with missing information were excluded, a total of 371 cases (280 men, 91 women) and 771 referents (577 men, 194 women) were left for analysis. After adjustment for tobacco smoking, age, and sex, significantly increased relative risks were observed for occupation in land transport, in particular bus, taxi, or truck driving. A statistically significant trend was seen with duration of employment in these trades. Based on a logistic regression analysis, a relative risk (RR) of 1.3 was determined for 10 years of employment in the trade. An association was also found for employment in trades undertaking painting (RR = 1.4 for 10 years' employment), and a significant trend emerged for duration of employment. A positive association with employment in the textile and leather industry disappeared after adjustment for tobacco smoking, and no association with duration of employment emerged. No association was found with employment in the chemical, rubber, iron and metal industries or in health services.

The incidence of bladder cancer in Denmark is among the highest in those parts of the world where papillomas of the urinary bladder are included with the frankly invasive tumors in cancer incidence statistics (22). Within Denmark the incidence is particularly high in Copenhagen (8), where tobacco smoking has been identified as the most important risk factor in the past (17), and also in the present study (15). This study was conducted to shed light on additional exogenous factors, which may be associated with the high bladder cancer incidence in Copenhagen in particular.
Among such factors, occupational exposures are known to increase the risk of bladder cancer. 2-Naphthylamine, benzidine, auramine, and magenta produced from aniline, as well as 4-aminobiphenyl are recognized bladder carcinogens (13). Increased risks have been recorded among persons in industries using Reprint requests to Dr OM Jensen, Danish Cancer Registry, Institute of Cancer Epidemiology, Danish Cancer Society, Landskronagade 66, DK-2100 Copenhagen, Denmark. 4 aromatic amines, such as among rubber workers in Great Britain and among dyers in the textile industry. Increased risks have also been reported among hair dyers, machinists, and truck drivers (4,7,18,21,23).
In Denmark, Lockwood's case-referent study, published in 1961 (17), is suggestive of an increased risk for workers in the skin, leather, rubber, iron and metal industries, in construction, and in work on railways and tramways. The appropriateness of the comparison between cases and a national survey is, however, doubtful, as pointed out by the author, but similar associations have also been found in the United States (7). Exposure to gasoline and oil products has been suggested to be of etiological importance in Denmark (19), and an analysis of a linked data set from the Cancer Registry and the Danish Supplementary Pension Fund indicates a significantly increased risk among men ever employed in printing, the manufacture of paint and glue products, the manufacture of electricity and gas, and among persons working in pharmacies and paint shops (20). The present paper reports on risks associated with previous employment iii certain occupations, as recorded in a case-referent study conducted in Copenhagen between 1978 and 1981.

Cases
The investigation includes tumors located in the urinary bladder only. As the clinical distinction between noninvasive (papilloma) and invasive tumors of the urinary bladder is difficult (5), the two have been grouped together under the term "bladder cancer. " Ascertainment of the cases and referents for the study and the interviewing techniques have been described in detail elsewhere (14);to be eligible for study the persons representing the cases and the referents had to be alive. From May 1979 to April 1981 a total of 412 patients living in the municipalities of Copenhagen and Frederiksberg and the county of Copenhagen were reported to the study , and 389 of these, or 94.4 l1Jo, were interviewed. A comparison with the records of the Cancer Registry for the population and time period in question shows the case group to be a representative sample of all bladder cancer cases arising in Greater Copenhagen with regard to age, sex, place of residence by area, and broad occupational groups. For some 99 l 1Jo of the men and all of the women the bladder tumors were histologically verified; 89.4 l 1Jo of all the tumors were of the transitional cell type, of which 66.6 l 1Jo were Bergquist grade 0 to 2 (2). Altogether 61.2 l 1Jo of the tumors were in stages Ta without infiltration or TI with infiltration of subepithelial connective tissue only; 3.6 l 1Jo were T4 tumors, ie, with fixation. The majority of cases was thus investigated in early stages of disease development.

Referents
The referents were selected at random in April 1979 among the residents of the municipalities where the cases were living, covering a total population (below the age of 75 years) of 552 071 men and 577 382 women on I January 1980. The sample was stratified to group-match the cases with regard to sex and age in five-year age groups . As all the referents were drawn at the beginning of the study, some had died or moved out of the area when approached for interview during the following two years. Such persons were replaced with new persons drawn at random from the list of residents in the same age and sex groups . Among the I 052 referents approached, the overall participation rate was 75.1 l1Jo.

Inter views
Interviews were usually conducted in the persons' home by a trained interviewer, who was not informed of the interviewee's statu s as a case or referen t. A structured questionnaire was used to obtain information on known or suspected risk factors for bladder cancer. The questions concerned tobacco smoking, drinking of coffee, tea and other beverages , use of artificial sweeteners, use of drugs, other medical procedures, previous medical conditions, and occupation.
Questions were asked about specific occupational exposures , and an occupational history was then constructed for each person , from systematic requests for information about single employments (company, industry , type of work performed, place of work, periods of employment). The information was coded according 130 to industry by means of an extended Dani sh version of the International Standard Industrial Classification (ISIC) of all economic activities (10). Jobs were classified according to the Dan ish occupational classification code used for the 1970 census (9).

Statistical analysis
After initial scrutiny of the specific occupations, the occupational historie s of the cases and referents were summarized as the total duration of emplo yment in certain occupations. Thi s pro cedure permited both simple anal yses comparing persons ever with tho se never employed in a given indu stry and investigations of the association of bladder cancer with the dur ation of employment in a given industry.
The stati stical analysis was carried out by unconditional maximum likelihood estimation of the parameters of a logistic regression model (3) using the generalized linear interactive modelling (GUM) package (I). As cigarette smoking was found to be the most important dete rminant of bladder cancer, it was fully controlled by including the following three smokingrelated variables: two dichotomous variables qualifying whether a person (a) had ever smoked and (b) was currently smoking and a continuou s variable quantifying the lifetime cigarette exposure by the logar ithm s of pack-years, as described elsewhere (15). All the models investigated accounted also for age and sex. The confidence limits were based on the asymptotic theory of maximum likelihood estimates.
Scrutiny of the data did not lead to the identification of other potentially confounding variables. Individuals for whom single pieces of information were missing were excluded from the analysis, the result leaving 371 cases (280 men, 91 women) and 771 referents (577 men, 194women). For the logistic regression analysis the maximum number of individuals with complete information on the variables of interest was used. Table I shows the number of cases and referents ever and never employed in a given industry , as well as the estimated relative risks (RR) adjusted for age and sex. Significant increased relati ve risks were observed for emplo yment in the textile and leather industry (ISIC 32), in painting [ISIC 50150 (painting firms), 39097 (sign post factories and sign post painters), 95133 (automobile painters), 33209 (furniture lacquering and painting), 38196 (industrial painting)], and in land transport (lSIC 711). Table 2 gives the results of the logistic regression analysis when duration of employment in land tran sport, driv ing, painting, and the textile and leather indu stries, respectively, were regarded as continuous variable s, adjusted for age, sex, and the three smoking variables. As effect measures the relati ve risk estimates associated with 10 years of employment have been derived as exp (to b), where b is the estimated regression coefficient. The reduction of deviance, as well as that of the confidence intervals, showed a significant association (5 070 level) between bladder cancer and employment in land transport (RR = 1.28), bus, taxi or truck driving (RR = 1.29), and painting (RR = 1.39). By contrast, duration of employment in the textile and leather industry was not a significant risk determinant in this smoking adjusted analysis. In addition, there was no clear trend in the relative risk with duration of employment in this industry when examined in a categorical fashion by to-year intervals. Model 6 in table 2 shows that there was little potential for two single occupations to confound each other, as there was no substantial change in the coefficients when occupation as a painter (model 4) and occupation as a bus, taxi or truck driver (model 3) were included jointly in the regression model.

Results
The association between bladder cancer and employment in land transport trades comprising railroads (lSIC 7I1tO), bus services (lSIC 7112), taxi services (ISIC 7113), trucking (lSIC 7114), and other transport services (lSI C 7115) is shown in further detail in table 3. There was a significant trend of increasing relative risks of bladder cancer with increasing duration of employment in land transport in general, and, when restricted to bus, taxi and truck services, ie, excluding employment on railroads (lSIC 71II X) and other transport services (ISIC 7115X).
Regression coefficients and relative risks were also estimated for the employment of 24 cases and 57 referents in sea transport (lSIC 712) (RR 0.9, 95 070 confidence interval (95 % CI) 0.5-1.4) and for the subgroup (2 cases, 9 referents) of machinists and stokers at sea (ISIC 712, partly) (RR 0.4, 95 % CI 0.1-1.7). Neither of these risk estimates were thus significantly different from I. There were 13 cases and II referents who had been employed in painting trades, and, among these, five cases and seven referents had a duration of employment between one and nine years, while the remaining eight cases and four referents were employed for 20 years or more. Table 4 shows the clearly increased relative risks as well as the significant positive trend (P = 0.017, two-sided).  Z·value for tr end 2.38 P = 0.017 c a See t able 1. b Ad justed f or age , sex , and th ree smokin g varia ble s in the lo gist ic regress io n. c Two-sided tes t f or tre nd .

Discussion
Th e present study indicates that the risk of bladder cancer in Copenhagen is increa sed amo ng person s who have been employed in the lan d t ra nsport services or as a pa int er. By contrast , no increas ed risk was found for persons in the chemical industr y, the iron and metal industry, and the rubber industry or among health personnel (table 1). The indication of an increased risk for employment in the textile an d leat her industry did not persist when smoking was acco unted for , and there was no association with durat ion of employment.
Fo r a ll occu patio ns associated with bladder ca ncer risk in the present study, est imat es o f risk associ ated with 10 years of occupat ion , as well as their con fidence intervals (tabl e 2) were almost iden tical with the est ima tes and confidence inte rvals determined for males and smo kers. In oth er words , the effec t of occupation was ba sically found for male smokers, fo r whom it ho wever represent s a significa nt increase over the smoki ng-related risks for which the model was adj usted .
The increased relative risks among persons emplo yed in land transport services persisted a fter thorough adju stm ent for smoking (tables 2 a nd 3). The risk was prim ar ily confined to bus, tru cking, and taxi services 132 with a two-to threefold increase for persons employed for 10 or more years. The association could not have been biased by interviewer or inte rviewee suspicion of such an associa tion from other studies, as the data collection a nd cod ing were performed prior to reports emerging fro m North America. T he associa tion was very specific, as no increased risks were ob served among persons wor king in the railway services, in oth er tr an spo rt serv ices (eg, car rental) or in transpo rt services at sea . T he increasing risk with duration o f emp loyment poi nts to a biologically meanin gful associatio n .
Th e present findings corroborate results of recent report s of an incre ased risk of bladder ca ncer among truck dri vers in the United States. Th e bladder ca ncer risk has been associated with exposur e to diesel fume s (12, 2 1). An increased risk with such expos ure a lso emerged from a British mortality study of bladd er cancer (6). It has however been poin ted out th at exposure to diesel fumes is unlik ely to be th e sole explana tion , since the risks have also been determin ed for nondiesel expos ed driv ers (II), and expo sur e to traffic exhaust has been offered as another po ssible explanation (II). Most truc ks and buses, as well as a lar ge proportion of taxis in Denmark, have diesel engines, but no informati on on diesel exhaust expo sur e was collected in the present investigation . The increased risk associated with employment in land transport services in Co penhagen might be explained in this way . As an oth er possibility, person s in the tran sport services may have an increased expos ure to gaso line and grease, which have been suggested to increase the risk of bladder ca ncer in Denm ark (19).
In anot her pa per from th is study (16), we repo rted on the resu lts in relation to beverage co nsumption . Total liqu id co nsumption was fou nd to be relat ed to bladder ca ncer risk, and it was speculated tha t thi s find ing might be related to the regular ity of bladd er emptying, which had also been brought up as a possible explana tio n of incre ased bladder ca ncer risk among truck dr ivers (I I). Th erefor e, in th e present analysis we investigated the possibility that total liquid consumption and duration of employment as a driver might be confounding each other, but we did not find any such effect.
A fourfold increased risk was noted among persons with long-term employment in the application of paint in various trades (table 4). In a study of the total Danish Cancer Registry material for the years 1970-1979 (20) an increased risk of lower urinary tract cancer (ie, bladder, renal pelvis, ureter, and urethra) was found among persons in paint and glue manufacturing, and among those with a work history in paint shops and pharmacies. An increased risk of bladder cancer among painters has previously been described in the United States (23), whereas other studies have not reported such an association. Persons in these trades are likely to be in contact with a large variety of chemical and complex mixtures, and the nature and composition of these mixtures may vary from one country to the other and thus influence the consistency of the observation.
In conclusion, the present study shows an increased risk of bladder cancer among persons who have been employed in the transport services, including buses, trucks and taxis, in Copenhagen. The explanation for this increased risk, primarily among men, is unclear, but the reality of the finding is strengthened by an association with duration of employment and by similar results emerging from studies elsewhere. If truely causal, as much as 5 010 of all bladder cancer cases in Copenhagen can be considered attributable to employment in land transport services. The examination of urinary concentrations of compounds associated with diesel fuels, traffic fume exposure in general, and gasoline or grease exposure in these trades may be rewarding, both with regard to the understanding of bladder carcinogenesis and with a view to prevention. Similar studies of biological samples and of the work environment should be pursued among painters. Confirmation of the present results by studies of cohorts with well-defined occupational exposures should also be attempted.