Abstract

Aims: To reassess saliva cotinine cut-points to discriminate smoking status. Cotinine cut-points that are in use were derived from relatively small samples of smokers and non-smokers 20 or more years ago. It is possible that optimal cut-points may have changed as prevalence and exposure to passive smoking have declined. Design: Cross-sectional survey of the general population, with assessment of self-reported smoking and saliva cotinine. Participants: A total of 58,791 respondents aged 4 and older in the Health Survey for England for the years 1996-2004 who provided valid saliva cotinine specimens. Measures: Saliva cotinine concentrations, demographic variables, self-reported smoking, presence or absence of smoking in the home, a composite index of social disadvantage derived from occupation, housing tenure and access to a car. Findings: A cut-point of 12 ng/ml performed best overall, with specificity of 96.9% and sensitivity of 96.7% in discriminating confirmed cigarette smokers from never regular smokers. This cut-point also identified 95.8% of children aged 8-15 smoking 6 or more cigarettes a week correctly. There was evidence of substantial misreport in claimed ex-smokers, especially adolescents (specificity 72.3%) and young adults aged 16-24 (77.5%). Optimal cut-points varied by presence (18 ng/ml) or absence (5 ng/ml) of smoking in the home, and there was a gradient from 8 ng/ml to 18 ng/ml with increasing social disadvantage. Conclusions: The extent of non-smokers’ exposure to other people’s tobacco smoke is the principal factor driving optimal cotinine cut-points. A cut-point of 12 ng/ml can be recommended for general use across the whole age range, although different cut-points may be appropriate for population subgroups and in societies with differing levels of exposure to secondhand smoke.