ABSTRACT

Introduction: The addictive nature of tobacco makes smoking cessation a dynamic and challenging process. Smokers who attempt to quit transition through different phases based on their withdrawal symptoms, motivation, outcome expectancies, and self-efficacy. Relapse risk is determined collectively by the above constructs, and the effect of these constructs varies over time. Our study aims to identify underlying classes (or profiles) among smokers participating in a pharmacotherapy cessation program based on withdrawal symptoms and motivational processes. Our interest further lies in examining changes in class membership over 4 weeks since the target quit day (TQD) and its association with long-term abstinence outcomes.

Methods: Our case study includes Ecological Momentary Assessment (EMA) data from the first 4 weeks post-TQD of a smoking cessation study which compared the efficacy of three pharmacotherapies. We apply Latent Profile Analysis to classify 1086 participating smokers into subgroups based on their responses to EMA prompts that indicate latent quitting behavior at four different time points (day 0 post-TQD, and end of weeks 1, 2, and 4  post-TQD). We then determine the changes in membership over time using Latent Transition Analysis (LTA). This analytical approach was executed separately for morning and evening prompts. The associations between baseline covariates, smoking profiles, the transition trend, and long-term abstinence outcomes were also examined to characterize key predictors of relapse risk.

Results: A four latent class model solution is selected through a holistic assessment of model fit-statistics (BIC, VLMR LRT p-value, entropy) and interpretability of profiles. The transition probabilities estimated via the LTA model with no assumed model structure across time show a moderate to high percentage of neutral/motivated subjects tend to improve on their motivation or stick to the same subgroup and the demotivated subjects transition to improved subgroups over 4 weeks. The findings are consistent across the morning and evening prompt-specific analyses.

Conclusions: Change in behavioral patterns due to fluctuating withdrawal tendencies and motivational processes during cessation attempts are useful in identifying vulnerable subgroups and to target interventions to prevent relapse risk in smoking cessation trials.