Trends and disparities in cannabis use disorder-coded inpatient discharges across 18 U.S. states, 2005-2023.

Azagba S.; Ebling T.; Roy B. · Public health · 2026

**Background & Methods** This repeated administrative dataset analysis examined trends in cannabis use disorder (CUD)-coded inpatient discharges across 18 U.S. states from 2005-2023, analyzing 216,203,471 total discharges. Logistic regression with cubic time trends and two-way interactions assessed CUD coding probability by sociodemographic characteristics (age, sex, race/ethnicity, insurance type, state), with secondary analyses restricted to principal diagnoses. **Key Findings** - CUD-coded inpatient discharges increased substantially from 1.02% of all discharges in 2005 to 3.15% in 2023, representing a threefold increase over 18 years. - Significant sociodemographic disparities persisted in 2023: males (4.19%) had nearly twice the prevalence of females (2.32%); adolescents (10-19 years) had the highest prevalence at 9.33%; and non-Hispanic Black patients (5.43%) demonstrated higher rates compared to other racial/ethnic groups. Government and other insurance types were associated with higher CUD prevalence compared to private insurance. - Marked geographic variation existed across states, with New Mexico showing the highest prevalence (4.39%) and South Carolina the lowest (1.89%) in 2023. Time-varying changes were most pronounced by age and race/ethnicity intersections. **Dosage & Administration** Not reported. **Safety & Adverse Effects** Not reported. **Evidence Quality** This large administrative dataset study provides robust population-level trend data but has notable limitations. The analysis relies on coded diagnosis data, which may underestimate true CUD prevalence if cases go undiagnosed or uncoded. Restricting analysis to 18 states limits generalizability to the entire U.S. population. The study is descriptive/observational rather than causal, and differences in hospital coding practices across states and time periods may influence apparent disparities. Secondary analyses showed estimates were lower when restricted to principal diagnoses, suggesting considerable variation based on diagnostic position definitions. The findings represent Level III evidence suitable for surveillance and epidemiologic characterization but cannot establish causation or patient-level mechanisms underlying observed disparities.

More recent cannabis research