A Markovian model for predicting the impact of prevention interventions on a population with multiple behavioral risk factors

Over the years, an increasing number of studies have focused on the impact on public health of non-transmissible diseases. These conditions (which include, e.g., cardiovascular and oncological diseases) are characterized by the fact that they are driven by the exposure to behavioral risk factors (such as smoking, sedentary lifestyle, alcohol consumption) and not by human-to-human interactions. In this paper, we focus on the impact of three behavioral risk factors: smoking, sedentary lifestyle and alcohol consumption. We construct a Markovian model that characterizes each agent by an independent Markov chain, whose states describe age, gender, exposure to risk factors and health of the individual. Importantly, we assume that the evolutions of the subjects are independent, and keep track of the dynamics of the population over a finite time horizon. The ultimate goal of this work is to compare a baseline scenario, where no intervention is implemented, with intervention scenarios, in which the exposure of the population to the risk factors is decreased. In order to evaluate the benefits, the impact is measured in terms of DALYs. To obtain a more realistic representation of the population, the model is constructed and calibrated based on real data and evidence published in the literature. The main novelty introduced lies in the inclusion of three risk factors. In particular, the main focus of this model is on the coupling between smoking and alcohol consumption, whose effects are known to be deeply correlated. In order to assess the impact of selected public health measures, we compare the model’s outcomes under different parameter assumptions. In particular, results indicate that while differences between independent and dependent models can be modest for some quantities of interest, accounting for dependencies among behavioral risk factors is crucial for interventions targeting single factors, since they generate complex interactions.