When to be skeptical of negative studies: pitfalls in evaluating occupational risks using population-based case-control studies.

This study investigated arsenic and lung cancer incidence in a community setting in the Montreal area. Job histories and sociodemographic factors were collected by interview from 857 lung cancer cases, 533 general population controls, and 1,360 controls with other cancers. Chemist-hygienists assessed each subject's life-time occupational exposure to 294 substances. Logistic regressions yielded arsenic/lung cancer odds ratios of 1.1 (95% confidence interval = 0.60, 1.7) based on cancer controls, and 0.82 (95% confidence interval = 0.41, 1.6) based on population controls. Risk did not rise with increasing level or probability of exposure. Worksite studies consistently show lung carcinogenicity from arsenic. Since confounding from other chemicals was well controlled, the most likely explanation is substantially lower exposures than in previous studies. The lack of association in this study demonstrates the need for caution in interpreting negative findings from population-based case-control studies, particularly when exposures are low or rare, as well as the difficulty in generating hypotheses from such studies.