Sex related differences in cigarette smoking nondisclosure and serum cotinine levels

Introduction: Cigarette smoking self-reports have been regularly questioned for social desirability response bias, thus challenging the validity of epidemiological studies that use self-reports as exposure. Methods: The authors analyzed self-reports, serum cotinine and tobacco consumption data on 493 men and 489 women recruited in a 2005-2009 prospective pregnancy cohort study using receiver operating characteristic curve, linear ANCOVA plot and Wilson score intervals. Nondisclosure was defined as self-reporting as non-smoker but having serum cotinine level higher than cut-off. Results: Nondisclosure was significantly higher in male smokers (20.88%; 95% CI: 13.79, 30.32) than female smokers (5.26%; 95% CI: 1.81, 14.37) using already established serum cotinine cut-offs (1.78 ng/mL for males and 4.47 ng/mL for females). Compared with male smokers, female smokers had lower average serum cotinine levels [218.42 (185.29,252.89) vs 140.21 (110.29,171.78); p<0.01] with no significant difference in average number of cigarettes smoked per day [9.78 (7.90, 11.67) vs 8.80 (6.52, 11.07); p=0.51]. The optimal cut-offs for differentiating smokers and non-smokers were calculated for men and women as 15.16 ng/mL (88.9% Sensitivity, 72.7% Specificity) and 7.14 ng/mL (92.6% Sensitivity, 88.5% Specificity) respectively. Analyzing with new cut-offs showed a lower estimate of nondisclosure but it still remained significantly higher in male smokers [15.79% (9.27,25.60)] than female smokers [5.66% (1.94, 15.37)]. Conclusion: Cigarette smoking self-report is not an accurate measure of tobacco exposure. Additionally, significant sex related differences exist in serum cotinine levels among smokers suggesting that self-reports may overestimate the tobacco exposure and risk of associated diseases in women compared with men.