Objectives: The purpose of this study was to validate, in a case-control study, the reporting by lung cancer cases and controls of their own lifetime smoking habits and of the smoking habit of the spouse.
Methods: In a multicenter (Sweden, Spain, Italy) case-control study of environmental tobacco smoke (ETS) and lung cancer, subjects were screened by repeated probing to exclude regular smokers of one cigarette/day or more for one year or more, and to quantify any occasional smoking. We then performed a short validation interview with next-of-kin in three centers.
Results: Only five of 408 index subjects who had never smoked regularly (1.7 percent) were reported by next-of-kin to be former regular smokers. These subjects had a cumulative lifetime consumption of cigarettes below 1.1 pack years. Among 351 subjects with quantitative smoking information from both sources who reported ever smoking 400 cigarettes or less (the definition of never-smoker used in the multicenter ETS study), nine subjects (2.6 percent) had smoked more than this amount occasionally according to next-of-kin. Misclassification was not higher for cases than controls. Relative risks for lung cancer associated with indicators of ETS exposure were not substantially altered by excluding the nine possibly misclassified subjects. The reports from 223 pairs of index subjects and next-of kin regarding the cumulative amount smoked by the spouse agreed quite well (Spearman's rank correlation 0.75 for reported smokers, 0.92 for all subjects). Only one index subject failed to report a spouse who had smoked regularly (99 percent sensitivity).
Conclusions: Smoking status and exposure to spousal ETS as reported by lung cancer cases and controls agreed strongly with reports by next-of-kin. Overall, our results suggest that bias from smoker misclassification is likely to be insignificant, and they contribute to the evidence linking exposure to ETS with an increased risk of lung cancer.