Forecasting frequent alcohol use among adolescents in HBSC countries: a Bayesian framework for making predictions.

(1) Aim: To summarize alcohol trends in the last 30 years (1985/6-2017/8) among 15-year-olds in Health Behaviour in School-aged Children (HBSC) countries (overall sample size: 413,399 adolescents; 51.55% girls) and to forecast the potential evolution in the upcoming 2021/22 HBSC survey. (2) Methods: Using 1986-2018 prevalence data on weekly alcohol consumption among 15-year-olds related to 40 HBSC countries/regions, a Bayesian semi-parametric hierarchical model was adopted to estimate trends making a clusterization of the countries, and to give estimates for the 2022 HBSC survey. (3) Results: An overall declining trend in alcohol consumption was observed over time in almost all the countries. However, compared to 2014, some countries showed a new increase in 2018 and 2021/22 estimates forecast a slight increase in the majority of countries, pointing out a potential bounce after a decreasing period in frequent drinking habits. (4) Conclusions: The clusterization suggested a homogenization of consumption habits among HBSC countries. The comparison between 2022 observed and expected data could be helpful to investigate the effect of risk behaviour determinants, including the pandemic impact, occurring between the last two waves of the survey