Adult-use cannabis policy stages linked to lower heroin involvement in…

INTRODUCTION: Evidence linking cannabis-policy liberalization to opioid outcomes is inconsistent, and the accumulation of successive policy stages has received limited attention. We estimated differences in opioid involvement among reported substance use treatment admissions across an accumulated cannabis-policy pathway distinguishing legal authorization, operational access, and preceding policy history. METHODS: Data were drawn from approximately 55.2 million admission records in the Treatment Episode Data Set-Admissions covering 1992 through 2023. Cannabis policy was classified into five mutually exclusive state-year regimes: no coded cannabis policy, medical authorization without operational sales, operational medical access, adult-use legalization without operational retail, and operational adult-use retail. A staggered-adoption difference-in-differences design estimated regime-specific average treatment effects on the treated (ATTs), representing accumulated pathway effects relative to continued no-policy status. Event-time and equal-state-year analyses assessed temporal patterns and sensitivity to alternative weighting. RESULTS: In the primary equal-state analysis, heroin-related ATTs were not statistically significant during medical-policy stages. Any heroin involvement was lower during adult-use legalization without operational retail (ATT = -107.01 per 1000 reported admissions; 95% confidence interval [CI] = -190.91 to -23.12) and operational adult-use retail (ATT = -127.71; 95% CI = -225.80 to -29.63). Corresponding primary-heroin ATTs were -102.81 (95% CI = -182.39 to -23.24) and -126.45 (95% CI = -221.47 to -31.44), respectively. Estimates for other opiates and synthetics were statistically non-significant. Later-stage heroin estimates remained negative but were attenuated and statistically non-significant under equal-state-year weighting. CONCLUSION: The accumulated-pathway framework identified lower heroin involvement during both adult-use policy environments relative to continued no-policy status in the primary analysis. The findings demonstrate the value of distinguishing successive policy environments and interpreting later-stage effects in the context of preceding policy history.