Can we predict adolescent cannabis use? A Bayesian semi-parametric approach to project future trends

Abstract

Despite its decrease in many Western countries, cannabis remains the most used illicit substance among adolescents. This study aims to summarize cannabis consumption during the last two decades and project trends among 15-year-olds in the 2021-22 HBSC survey. A Bayesian semi-parametric hierarchical model was adopted to estimate the trend of cannabis consumption using data of about 287,000 adolescents from the 2001/2002 to the 2017/2018 HBSC wave and the 38 countries that met the inclusion criteria. Data show an overall decline in most countries for both boys and girls. However, in 22 countries of 38 cannabis use is expected to increase again in our projection. The discussion of these findings should take into account cultural, policy, social factors and unpredictable events such as the Covid-19 pandemic, that can significantly impact future trends leading to discrepancies between the projected and observed values. However, these discrepancies can provide insight into understanding the potential impact of preventive strategies and the underlying processes responsible for changes in cannabis use over time.

Keywords: Adolescents; Bayesian longitudinal model; Cannabis; Forecast; Risk behaviour; Trend.