Abstract
Future mothers tend to underreport intentional or unintentional tobacco smoke exposure to avoid social stigma, and urine levels of cotinine, the major metabolite of nicotine can serve as a valuable biomarker of tobacco smoke exposure to resolve such uncertainties. When establishing the cut-off level to discern non-smokers from active smokers, however, one should bear in mind the shorter half-life of cotinine in pregnant than non-pregnant women to avoid misclassification of maternal smoking status. The aim of our study was to determine the urinary cotinine cut-off level to objectively detect active smokers and to see if any participant characteristics were associated with underreporting. To do that, we collected spot urine samples from pregnant women with normal vaginal delivery at term, self-reported as non-smokers (n=123) or smokers (n=33), in the maternity ward before delivery. We also took their sociodemographic, cigarette smoking, and clinical data, as well as clinical data on their newborns using a questionnaire. To ensure objective classification of participants by smoking status, cotinine levels were quantified in urine using gas chromatography-mass spectrometry. The receiver operating characteristic (ROC) curve analysis showed that the optimal cut-off value of urinary cotinine to discern pregnant non-smokers from smokers at term were 120 µg/L or 144 µg/g creatinine, yielding 94 % sensitivity and 96 % or 97 % specificity. Respective to these cut-off values, 4.1 % and 3.3 % of self-reported non-smokers were classified as current smokers. Our findings support the use of urinary cotinine, whenever feasible, to reduce reporting bias in pregnancy studies, and, given the altered nicotine metabolism during pregnancy, future epidemiological studies should bear in mind that urinary cotinine cut-off values may depend on the stage of pregnancy.