Objectives This study's purpose was to conduct a more in-depth analysis of the potential association between lung cancer, occupational exposures and smoking using data on cohort members from a Canadian petroleum company and refined statistical analyses.

Methods Information on various exposures including asbestos and petroleum coke dust, as well as job type and operating segment were collected via manual and computerised company records. We performed life-table analyses, Poisson regression and restricted cubic splines to model exposure–response patterns while controlling for smoking status and age. Model diagnostics included the assessment of dispersion and offset parameters.

Results These analyses show that lung cancer risk is strongly related to age and smoking, and to a lesser extent to province of last residence. When controlling for these covariates, there is suggestive evidence that maintenance work may also be related to lung cancer risk. Some analyses also indicate that asbestos exposure may be associated with lung cancer risk, although a clear exposure–response trend is not seen. Other exposures, including petroleum coke dust, were not strongly related to lung cancer risk, particularly when expressed as a continuous measure.

Conclusions These data suggest that maintenance work may be associated with lung cancer incidence, although exposures to the single agents studied did not emerge as strong predictors of lung cancer incidence. Maintenance work may be a surrogate for general exposures to several agents (eg, polycyclic aromatic hydrocarbons, metals, welding fumes, radiation, etc), although these results may be affected by residual confounding due to smoking or other socio-demographic factors.