Who leaves early? Factors associated with against medical advice discharge during alcohol withdrawal treatment

Objective: To determine if certain patient, clinical, and disease factors are associated with against medical advice (AMA) discharge among patients admitted for treatment of alcohol withdrawal. Methods: Data from admissions to a dedicated unit for treatment of substance withdrawal were collected over a 6-month period. Patients with AMA and planned discharge were compared with regard to demographics, clinical data, and substance use disorder disease characteristics. A stepwise logistic regression was used to find the best model. Results: The study population included 655 patient encounters. A total of 93 (14%) discharges were AMA. Bivariate analysis showed patients with AMA discharge were younger (mean age 43 vs 46 years; P < 0.05), more likely to leave on a Tuesday to Thursday, and to have an initial withdrawal score at or above the median (AMA 69% vs planned 56%; P ¼ 0.02). Emergency department (ED) admissions had an AMA discharge rate of 21% compared with 10% of community admissions (P < 0.05). Regression analysis found AMA discharge was significantly associated with admission from the ED (odds ratio [OR] 2.03, confidence interval [CI] 1.27 - 3.25) and younger age (OR 0.97, CI 0.95 - 0.99). There was no significant difference in discharge disposition among patients with concurrent opioid use disorder who were on opioid agonist therapy. Conclusions: AMA discharges occurred in 1 of every 7 admissions. Being admitted from the ED and younger age was associated with AMA discharge. No other patient or clinical factors were found to be associated with AMA discharge.