The Fagerström test for nicotine dependence: a comparison of standard scoring and latent class analysis approaches.

Abstract

The classification of being tobacco dependent obtained via the established scoring method of the Fagerström test for nicotine dependence (FTND) is compared to a method that bases classification on the pattern of item responses. Young adults participating in a longitudinal study, who indicated they had ever smoked, were asked six standardized items (n = 962; mean age 21 years). By standard scoring, the mean FTND score was 1.9 (S.E.= 2.3): 66% of the smokers qualified for a very low level of dependence, 17% low, 9% moderate, and 9% a high level of dependence. Response patterns detected by latent class analysis (LCA) indicated class differences based on severity gradations and of qualitative content. Three profiles of tobacco dependence were found: a non-dependent class (50%), a class manifesting a moderate number of dependence features (31%), and more severely affected class (19%). The vast majority of smokers (three-fourth) were classified congruently by these two methods. Discrepancies involved LCA classifying smokers into a higher level of dependence when compared to the conventional scoring classification. Patterns of dependence features obtained from population samples that include a wide range of smokers may provide insight into possible phenotypic differences among tobacco smokers, particularly when LCA methods are used to complement standard scoring methods.