The reliability of passive smoking histories reported in a case-control study of lung cancer.

A test-retest design has been used to examine the reliability of passive smoking histories reported in personal interviews. A total of 117 control subjects initially interviewed in a lung cancer case-control study conducted in metropolitan Toronto, Canada, between 1983 and 1984 were reinterviewed on average six months later. Responses to initial screening questions used to detect a person's exposure to passive smoke were more reliable for residential than for occupational exposure. Respondents also more reliably reported residential exposure to spouse's passive smoke than to the passive smoke of others at home. Quantitative measures of exposure to passive smoke, i.e., number and duration of exposure, were even less reliably reported. Nonsmoking respondents gave the most reliable information. The low reliability of self-reported duration of exposure to passive smoke is consistent with the inability of several studies to detect a significant dose-response relation with lung cancer risk when measures of dose that depend solely on duration are used.