Is stage-of-change a useful measure of the likelihood of smoking cessation?

We compared two stage-of-change models that differentiate smokers by their level in the quitting process. The original 1983 model by Prochaska and DiClemente (1) divided smokers first by relapse status and then by intention to quit; their revised 1991 model (2) reversed the primacy of these factors. No published data justify whether the revision improves prediction of cessation. We used data from a population-based panel of 1,921 smokers interviewed in 1990 and 1992 for the California Tobacco Surveys. Model variables (quitting intention and recent quitting history) were used in a logistic regression to predict 30-day or longer cessation at follow-up and quit attempts made during the year preceding the survey. Predictive power of the revised model was not better than predictive power of the original model. Newt approaches to differentiating smokers on likelihood to quit should emphasize quitting behavior rather than intention to quit.