Reliability of passive smoke exposure histories in a case-control study of lung cancer

Despite the growing number of studies on the health effects of passive smoke exposure, few data exist on the quality of questionnaire data on passive smoking. To measure the reliability of passive smoking histories, re-interviews were conducted for 110 subjects (37 cases and 73 controls) as part of a larger study of lung cancer among non-smoking women in Missouri. Agreement was high both for parental smoking status (94% concordance; kappa = 0.82) and for spousal smoking status (84% concordance; kappa = 0.67). Concordance also was relatively high for cigarette pack-years of exposure due to the parents or spouse. Reliability tended to be somewhat higher among controls than among cases, and for exposure due to a parent or spouse than for that due to other household members. Questions on the perceived harmfulness of passive smoke exposure showed no differences between cases and controls. These findings indicate a high degree of repeatability in responses regarding passive smoking, but also suggest the potential for misclassification of passive smoke exposure status, the desirability of standardized questions on passive smoking, and the need for additional studies of reliability and validity.