Patterns and predictors of change in a smoking intervention study: latent growth analysis of a multivariate outcome model.

Objective: This study reports the outcome of a smoking intervention study in a general population setting in Germany.

Design: A randomized, controlled trial, with a sample of 719 current daily cigarette smokers.

Main Outcome Measure: The authors tested a criterion measurement model that is related to the transtheoretical model and that proposes multivariate outcome measures with positive and negative attitudes about smoking and habit strength. The authors analyzed patterns and predictors of change in self-efficacy and in the pros and cons of nonsmoking using latent growth models.

Results And Conclusion: Participation in the intervention groups did not predict average mean levels or growth trajectories in the multivariate outcome model. Accounting for attrition in multisample missing-data procedures and accounting for demographic and smoking behavior variables did not alter this result. Thus, the intervention was ineffective. However, endorsement of self-efficacy and the pros of nonsmoking increased linearly over time in all study groups, which may be attributable to historical or societal changes or to repeated, comprehensive smoking assessments. In addition to reporting the outcome of a smoking intervention trial, the current study also illustrates a modeling approach to the measurement of change.