Accuracy of self-reported smoking status among participants in a chemoprevention trial

Background. Exposure to tobacco products is readily assessed through self- or interview-administered questionnaires. Degree of misreporting among participants in chemoprevention trials is unknown. We assessed the level of discrepancy between self-reported smoking exposure and plasma cotinine among participants in a chemoprevention trial. Methods. Analyses were conducted among 824 men and women who participated in a dietary trial of adenoma recurrence. Smoking exposure was ascertained through self-administered questionnaires at three time-points. Plasma cotinine was measured by gas chromatography among 283 never, 446 former and 95 current self-reported smokers. Sensitivity and specificity were assessed using various plasma cotinine cut-points. Results. Degree of misclassification for self-reported current smokers was minor (0-3%), regardless of cotinine cut-point used. Using a cut-point of 20 ng/ml, which takes into account exposure to environmental tobacco smoke among nonsmokers, sensitivity and specificity were 98.9% and 80.2%, respectively. Conclusions. These data indicate that degree of misreport for current smokers is extremely low; however, approximately 20% of self-reported never smokers misreport their exposure, suggesting that validation of self-report is needed for these individuals. (C) 2004 The Institute For Cancer Prevention and Elsevier Inc. All rights reserved.