This study aims to compare the prevalence that measured from self-reported adolescent non-smokers with the prevalence findings that obtained from the biomarker (serum cotinine level) in the National Health Interview Survey (NHANES) dataset from 2001 to 2012. We also discuss (i) what is the criteria of selection of optimal cut-offs of serum cotinine level, as a biomarker; (ii) whether cut-off points should vary by gender and race; and whether categorizing the biomarker (serum cotinine level) is a good idea?

Methods

This study used NHANES data from six studies between 2001 and 2012. The focus is on adolescents aged from 12 to 19 years old. Participants who lived in passive smoking environments and pregnant girls were excluded. Serum cotinine level (as a biomarker) is used as a gold standard measure to distinguish between smokers and non-smokers. However, there is no criteria to determine the cut-off point of serum cotinine level. In this study, we initially use cut-off points which were recommended by Benowitz et al. [11] and compare the findings with other cut-off points based on previous studies.

Results

Among self-reported non-smoking adolescents, non-Hispanic white participants who reported that they are non-smokers; 271(19.6%), 294(21.9%) and 163(15.1%) had a serum cotinine level greater than 3 ng/ml aged 12 to 15, 15 to 17 and 17 to 19, respectively. For non-Hispanic white, misclassification was slightly higher compared to non-Hispanic-white and Mexican; while Mexican participants yielded the lowest percentage of misclassification. The highest percentage of misclassification was for adolescents aged 12 to 15; while the lowest level was for adolescents aged 17 to 19. The percentages of misclassification changed based on the selected cut-off points. At age 12 to 15, the percentages of misclassifications were higher for boys compared to girls and this relationship was maintained for all other age groups. Classification by poverty ratio yielded the lowest percentages of misclassification. The analyses was repeated using another survey question ‘have you smoked nicotine or tobacco last five days’. We found the same findings for non-Hispanic white, and the cut-off point of 3 for SC level again underestimated the findings.

self-reported non-smoking, cotinine level, cut-offs points, adolescents and NHANES dataset.