Sunday Azagba, Todd Ebling, Bijit Roy
CUD-coded inpatient discharges increased substantially over time, with persistent sociodemographic and geographic disparities. These findings highlight the importance of screening and linkage-to-care efforts for disproportionately affected groups and of using consistent diagnosis-position definitions in surveillance.
OBJECTIVE: Prior studies document rising cannabis use disorder (CUD) prevalence and disparities, but less is known about trends in CUD-coded inpatient discharges across populations and intersecting sociodemographic groups. We examined trends and variation in CUD-coded inpatient discharges by sociodemographic characteristics.
STUDY DESIGN: Repeated administrative dataset analysis of inpatient discharge records.
METHODS: We analyzed 216,203,471 inpatient discharges from community hospitals in 18 U S. states (2005-2023). Logistic regression with non-linear (cubic) time trends and prespecified two-way interactions across age, sex, race/ethnicity, insurance, and state estimated the probability of CUD coding in any diagnosis position. We derived population-weighted predicted prevalence, prevalence ratios, and average marginal effect risk ratios. Secondary analyses restricted outcomes to principal diagnoses and excluded "cannabis use, unspecified" codes.
RESULTS: The proportion of discharges with CUD coding increased from 1.02% in 2005 to 3.15% in 2023, with substantial and persistent disparities. In 2023, prevalence was higher among males (4.19%) than females (2.32%), highest among adolescents (10-19 years: 9.33%), and higher among non-Hispanic Black patients (5.43%). Compared with private insurance, government and other insurance were associated with a higher prevalence. State-level prevalence varied markedly (New Mexico: 4.39%; South Carolina: 1.89%), indicating geographic heterogeneity. Time-varying changes were most pronounced by age and race/ethnicity. Estimates were lower when restricted to principal diagnoses.
CONCLUSIONS: CUD-coded inpatient discharges increased substantially over time, with persistent sociodemographic and geographic disparities. These findings highlight the importance of screening and linkage-to-care efforts for disproportionately affected groups and of using consistent diagnosis-position definitions in surveillance.