Time Varying Trends in the Association Between Smoking and Nicotine Dependence

Introduction: Given the continued public health burden associated with smoking, despite substantial environmental restrictions, a “hardening hypothesis” has been posited in which largely hard core or dependent smokers remain in the wake of population-based intervention. That is, although there has been a successful reduction in overall population prevalence, when conditioned on established smoking patterns, those who reach established smoking more recently have been shown to be at higher risk for becoming dependent on cigarettes than those reaching established smoking during prior decades (Breslau, 2001). Given emerging evidence that the development of nicotine dependence may not require heavy, long-term smoking, but rather, for some of the most sensitive individuals may instead occur at relatively low levels of smoking, there may have been a corresponding change in the strength and significance of the association between smoking and nicotine dependence over time. In other words, if these “sensitive” individuals now make up a larger proportion of current smokers than in previous decades when the behavior was considered normative, we would hypothesize that the strength of the association would be stronger in the earliest cohorts and weaker in those that are more contemporary. 

Method: Data from 24 individual years of the National Survey on Drug Use and Health (formerly National Household Survey on Drug Abuse) (spanning 3 decades) were pooled to obtain a large multi-cohort sample of daily smokers with considerable variability in the number of cigarettes smoked per day. We used moderated nonlinear factor analysis (MNLFA), which produces study equivalent ND scores after accounting for study differences, followed by time varying effects (TVE) models, to evaluate the strength of the association across time. Though other modern methods may be able to evaluate the average association between time varying covariates and substance use outcomes, TVE models are able to test how these effects may change over time. In this way, TVE models provides the opportunity to evaluate previously unanswerable questions such as how the relationship between risk/protective factors and smoking may change over time as well as the direction and strength of these association.

Results: Preliminary findings suggest the presence of some TVE differences that may be further moderated by individual characteristics such as age, and gender.

Conclusions: Time varying associations in this context suggest that currently, individuals at risk for becoming smokers may require targeted intervention based on an understanding of individual differences.