Mental health and cannabis use among Canadian youth: Integrated findings from cross-sectional and longitudinal analyses

Background and Purpose: Using data from two methodologically independent youth research studies in Canada, the Health Behaviour in School-aged Children (HBSC) study and the Cannabis, Obesity, Mental health, Physical activity, Alcohol, Smoking, and Sedentary behaviour (COMPASS) study, the objective of this study was to compare associations between youth's mental health and cannabis use across samples. Using similar indicators in both studies, our goal was to affirm the potential for nationally representative cross-sectional analyses (HBSC) to replicate findings found in a longitudinal non-representative data source (COMPASS), enhancing opportunity for causal inferences.Methods: Data were collected from grade 9 and 10 Canadian students participating in the 2017/18 HBSC ( n = 8462) and 2017/18 to 2018/19 waves of COMPASS ( n = 32,023). Using multivariable Poisson regression models, the strength and statistical significance of the effects of mental health indicators on cannabis use outcomes were estimated within both studies and compared for consistency. Using a 2-year linked sample of students partic-ipating in COMPASS, models examining the impact of mental health indicators on cannabis use initiation and maintenance over time were similarly fit using Poisson regression to estimate relative risk.Results: Similar associations between mental health problems and cannabis use were observed in both data sources. The direction, magnitude, and precision of the estimates for restless sleep, loneliness, poor wellbeing, and cannabis use were highly comparable across both studies. Worse mental health was consistently associated with current and lifetime cannabis use among youth.Discussion: Cross-sectional and longitudinal findings from two large methodologically diverse studies in Canada demonstrate a replicable association between indicators of mental health and youth cannabis use. Similarities were identified and two generalizations may be concluded: 1) potentially causal etiological relationships inferred from HBSC data were supported in longitudinal findings based on COMPASS, and 2) longitudinal COMPASS data aligns with nationally representative data from HBSC.