CMStatistics 2022: Start Registration
View Submission - CMStatistics
Title: A compartmental model for smoking habits in Tuscany (Italy) Authors:  Alessio Lachi - University of Florence (Italy)
Cecilia Viscardi - Univeristy of Florence (Italy)
Giulia Cereda - ()
Giulia Carreras - ISPRO (Italy)
Michela Baccini - University of Florence (Italy) [presenting]
Abstract: Investigating smoking habits in the population is crucial to plan appropriate tobacco control policies. We develop a compartmental model to describe the smoking dynamics in Tuscany (Italy) from 1993 to 2019 and forecast them until 2043. The model, which allows for mortality and new births, assumes that at each time, the population is divided into the following compartments: Never, Current, and Former smokers. Never smokers can become smokers, smokers can become ex-smokers and ex-smokers may relapse into smoking. Calibrating the size of the compartments on the estimated percentages of smokers, ex-smokers, and never-smokers arising from annual surveys, we estimate flexibly through regression splines the probability of starting and quitting smoking by age, and the probability of relapsing smoking as a negative exponential function of time from cessation. We also estimate the evolution of the number of individuals in the compartments in the past and we predict it in the future, under the assumption of unchanged transition parameters. We obtained the confidence intervals besides estimates through a parametric bootstrap procedure. Separate analyses were performed by gender. The predictive performance of the compartmental models, evaluated via cross-validation, was good. In a context of an overall decrease in smoking prevalence, the transition rates between compartments were quite different among males and females, with an evident change of them over the calendar period.