Nonmedical cannabis legalization and adolescent inpatient cannabis use disorder, 2008–2020: A quasi-experimental analysis

Background Evidence on the impact of nonmedical cannabis legalization on adolescent cannabis use disorder (CUD) is inconsistent, with limited data on inpatient encounters. This study assesses whether state nonmedical cannabis legalization is related to inpatient discharges for adolescent CUD. Methods We used a quasi-experimental design with generalized synthetic control and interactive fixed effects applied to annual state-level aggregates from 2,861,324 inpatient discharge records for adolescents aged 10–17 years across 13 U.S. states (2008–2020) with staggered legalization timelines. The primary outcome was a CUD diagnosis (ICD-9/ICD-10 codes) in any discharge position. Sensitivity analyses incorporated dispensary opening dates and subgroup stratification. Results Legalization was not associated with a statistically significant change in adolescent CUD-coded inpatient discharges (average treatment effect on the treated [ATT] = 0.0051; equivalent to a 0.51 percentage point increase; 95% CI, –0.0059-0.0160). Dynamic estimates showed no changes in the first six years post-legalization, but a statistically significant increase was observed in year seven (ATT = 0.0251; 95% CI, 0.0011-0.0491), driven by the earliest policy adopters. No consistent effects were found across later-adopting states. Results were robust to alternative specifications, including dispensary-based exposure timing (ATT = 0.0081; 95% CI, –0.003- 0.0194) and individual-level models. Stratified analyses by sex, age, race/ethnicity, and urbanicity showed no statistically significant associations. Conclusions Findings did not detect a statistically significant overall association between state nonmedical cannabis legalization and adolescent CUD-coded inpatient discharges. Estimates from the observed follow-up were consistent with no statistically detectable association, though potential delayed or cohort-specific effects cannot be ruled out.