Association of inpatient outcomes with cannabis use disorder in head and neck cancer.

e18112 Background: Cannabis is the most commonly used illicit substance worldwide and has recently been identified as a risk factor for head and neck cancer (HNC), joining established risk factors such as HPV, tobacco, and alcohol use. Whether cannabis use disorder (CUD) affects inpatient outcomes is unknown. We evaluated clinical characteristics associated with CUD among hospitalized HNC pts and hypothesized that CUD is associated with worse inpatient outcomes. Methods: We queried the National Inpatient Sample from 2016 to 2022 to identify adult hospitalizations for HNC with or without CUD using ICD-10 codes. Baseline characteristics were compared using chi-square and t-tests. The primary outcome was mortality; secondary outcomes included length of stay and major complications. Multivariable logistic regression assessed the association between CUD and mortality, adjusting for demographics, insurance, income, hospital characteristics, Charlson Comorbidity Index (CCI) excluding cancer, metastatic disease, alcohol and tobacco use, and an HPV-associated anatomic site proxy based on oropharyngeal tumor location. Sensitivity analyses stratified by HPV-associated site were performed. Results: Among 108,344 HNC hospitalizations, 1,596 (1.47%) involved CUD. Compared with non-CUD pts, those with CUD were younger (mean age 57.1 vs. 64.6 yrs), more often male (80.7% vs. 72.0%), Black (20.3% vs. 12.0%), and Medicaid insured (39.1% vs. 15.7%) (all p<0.001). Tobacco use (50.6% vs. 21.3%) and alcohol use disorder (26.8% vs. 7.9%) were more prevalent in the CUD cohort (both p<0.001). Unadjusted in-hospital mortality was lower among CUD pts (2.01% vs. 4.73%, p<0.001), while LOS was longer (7.48 vs. 7.05 days, p=0.041). Rates of major complications were similar, with lower pneumonia and acute kidney injury in the CUD group and no differences in sepsis, respiratory failure, aspiration, or venous thromboembolism. After adjustment, CUD was independently associated with lower in-hospital mortality (aOR 0.47, 95% CI 0.33-0.68; p<0.001). Metastatic disease, higher CCI, alcohol use disorder, and non-Medicare status were associated with increased mortality, while younger age, female sex, tobacco use, and HPV-associated tumor site were associated with lower odds in HNC pts. Stratified analyses showed no significant association between CUD and mortality among HPV-associated HNC, while a significant association persisted among non-HPV-associated HNC. Conclusions: Despite a higher burden of substance use and socioeconomic disadvantage, CUD was not associated with increased inpatient complications and was independently associated with lower in-hospital mortality, particularly in non-HPV-associated HNC, potentially reflecting younger age and lower comorbidity burden at hospitalization. Limitations include possible incomplete ascertainment of CUD use. These findings are important given the growing prevalence of CUD.