Identifying possible bladder cancer ocupational carcinogens via a case-control study and JEM

A significant proportion of cancer development is attributable to exposures to certain chemicals in the workplace. However, examining these occupational exposures is often a difficult and challenging task. In this thesis we use the relatively new approach of applying an extensive JEM (Job Exposure Matrix) to estimate the occupational exposures of 1,062 bladder cancer cases and 8,057 matched other cancer controls. The subjects are all male, and were at least 20 years old and resident in British Columbia when diagnosed with cancer between 1983 and 1990. A self-administered questionnaire provided the occupational histories and confounding information on the study subjects. The cumulative exposure (expected work-years of considerable exposure) to each of 11,132 occupational agents was estimated. The bladder cancer cases were matched to cancer controls of other cancer sites (excluding lung) on age at diagnosis and year of diagnosis. The analysis was performed via conditional logistic regression and the following confounders were taken into account: ethnicity, who completed the questionnaire, smoking duration, and alcohol drinking status. Of the 5,699 agents with at least 3 bladder cancer cases exposed, a significantly increased (5% level) odds-ratio was seen for ever exposure to 646 of them. Of the 3,450 agents with at least 9 bladder cancer cases exposed, 350 exhibited a significantly (5% level) increasing dose-response relationship. After adjusting for multiplicity, a subset of 30 agents was selected that demonstrated sufficient evidence of bladder carcinogeneity. Principal components analysis was performed on the cumulative exposures of these selected agents and 10 independent groups of agents were identified. The groups were mainly distinguished by job. The cumulative exposures to these 30 agents were mainly due to employment in logging, ship and boat building, and construction industries, and in occupations involving motor vehicles (e.g. gasoline service station attendant, mechanic, and truck driver). The selected 30 agents seem to mainly be of petroleum or mineral oil base.Science, Faculty ofStatistics, Department ofGraduat