[Estimation of optimal levels of tobacco biomarkers to distinguish active and passive smokers using ROC analysis].

Many epidemiological studies on health consequences of tobacco smoke exposure require classification of examined subjects either as active or passive smokers. Receiver operating characteristics (ROC) curves are useful for organizing cut-off values of tobacco biomarkers and visualizing their performance. The cut-off values might be applied to distinguish cigarette smokers and persons involuntary exposed to second-hand tobacco smoke (SHS). Aim of the study was estimation of optimal levels of three biomarkers (cotinine, 1-hydroxypyren, and carboxyhemoglobin) to distinguish active and passive smokers using ROC curves. 98 subjects (62% females) were qualified to the study. Mean age was 40 +/- 12 years. Active smokers (n = 38) had an average smoking history of 9 +/- 8 years and declared smoking at least 5 cigarettes per day (mean 17 +/- 7). Passive smokers (n = 60) declared being exposed to environmental tobacco smoke either at home or work (n = 18) or other indoor microenvironments, where they spent some time during their daily activity (n = 42). Cut-off values were determined for each biomarker using ROC curves. Optimal cut-off values were: 327 microg/g creatinine for cotinine, 47 ng/g creatinine for hydroxypyren, and 1.27% HbCO for carboxyhemoglobin. Among three studied biomarkers, cotinine showed the best sensitivity of 97.4% and specificity of 90.0%. Carboxyhemoglobin showed sensitivity of 89.5% and specificity of 93.3%, whereas 1-hydroxypyren 76.3% and 78.3%, respectively. Analysis of ROC curves appears to be a way to distinguish active and passive smokers using various tobacco biomarkers.