Cross validation of the prognostic and diagnostic utility of tobacco craving in a general and a pregnant sample of treatment-seeking smokers

Abstract

BackgroundValid and reliable brief measures of tobacco craving are necessary for research and clinical purposes. However, comparisons of the utility of single-item and brief multidimensional craving measures are scarce. MethodsWe analyzed two tobacco craving measures, the French versions of the 12-item Tobacco Craving Questionnaire (FTCQ-12) and the single craving item on the Minnesota Nicotine Withdrawal Scale, in pregnant and non-pregnant (females and males) French smokers from two independent smoking cessation trials. Using data from pregnant smokers, we estimated first a prognostic model for each measure and compared their ability to predict relapse from one visit to the next. We next used cutoff values for both measures to determine how well craving distinguished current smokers from abstainers. For cross-validation, we tested how well the same cut-off values generalized to non-pregnant smokers. ResultsTwo FTCQ-12 factors (Expectancy and Purposefulness) predicted smoking status; scores were higher in pregnant non-abstainers than abstainers. The sum of these two factors, FTCQ-12 risk score (RS), yielded higher sensitivity both in prognostic and diagnostic models than single-item MNWS craving. FTCQ-12 RS had higher sensitivity among pregnant than non-pregnant smokers. Specificity of both tobacco craving measures was similar among pregnant smokers, but higher with MNWS craving than with FTCQ-12 RS in non-pregnant smokers. ConclusionsFTCQ-12 RS and MNWS craving can be used together in clinical practice to rapidly predict smoking status. Interventions targeting Expectancy and Purposefulness may result in higher likelihood of abstinence, especially among pregnant smokers.