Abstract
BACKGROUND: Treatment discontinuation is common in outpatient alcohol use disorder (AUD) care, and attendance-related missingness can complicate interpretation of early symptom change in naturalistic cohorts. METHODS: We followed 72 adults newly entering outpatient care for moderate-to-severe AUD at a tertiary hospital in Taiwan (baseline and 1, 3, and 6 months). Discontinuation was defined as ≥30 consecutive days without documented clinic contact. Symptom trajectories were modeled using generalized estimating equations with and without baseline severity adjustment; baseline correlates of 1-month discontinuation were estimated using Firth bias-reduced logistic regression, and early symptom change was evaluated in a landmark analysis restricted to participants retained to the 1-month visit. RESULTS: The demoralization-by-time interaction for Alcohol Use Disorders Identification Test (AUDIT) reduction was significant in unadjusted models but attenuated after adjusting for baseline AUDIT (P = .385). Regular employment was independently associated with higher odds of 1-month discontinuation [adjusted Firth odds ratio (OR) 7.67, 95% confidence interval (CI) 1.19-49.24]. In the landmark cohort (n = 50), early symptom improvement at 1 month was not significantly associated with subsequent discontinuation by 3 months (Firth OR 0.58, 95% CI 0.16-2.04). CONCLUSIONS: After accounting for baseline severity, demoralization was not a robust determinant of symptom trajectories. Regular employment showed a consistent association with early discontinuation, suggesting potentially modifiable time- and access-related barriers; early symptom change should be interpreted as an attendance-conditioned correlate rather than a standalone prognostic marker.