How to obtain long term projections for smoking behaviour: A case study in the Dutch population

A state-transition model is presented to project the prevalence rates of never, current and former smokers within the Dutch population. Changes are determined by the transition rates among these three classes: start rates, quit rates and relapse rates. Both the initial prevalence rates and transition rates are calculated from cross-sectional data using a restricted quadratic multinomial regression spline and a restricted quadratic logistic regression spline, respectively. Through a Monte Carlo experiment an uncertainty analysis was performed to assess the level of reliability of the results, while a sensitivity analysis was carried out in order to detect the input variables that mostly contribute to the output variability. The results obtained through this model show a considerable but rather slow decrease of smokers up to year 2050.