Abstract
In studies of the health effects of exposure to environmental tobacco smoke (ETS), misclassification of active smokers has the potential to bias the estimates of disease risk. Biochemical validation of exposure to ETS can provide objective evidence of current smoking status in epidemiological studies. Intrinsic to this effort is the establishment of appropriate cut-off points for the measurements of tobacco biomarkers. Within a collaborative study on ETS co-ordinated by the International Agency for Research on Cancer, questionnaire data and urine samples were collected from 1,369 women at 13 centres in 10 countries. Forty seven of these women had urine cotinine levels above 50 ng.mg-1 creatinine, a level used to discriminate smokers from nonsmokers in previous studies. The distributions of the subjects across cotinine values and self-reported exposure to ETS was consistent with the association, at one extreme, of moderate cotinine levels (50-150 ng.mg-1) with very high exposure to ETS, and, at the other extreme, of very high cotinine levels indicating actual use of nicotine-containing products in women with low ETS exposure. Using the cut-off point of 150 ng.mg-1, only 1.5% of the alleged nonsmokers were reclassified as current light smokers. Potential bias due to smoker misclassification is very unlikely to be responsible for the increased health risks observed in epidemiological studies on ETS.