Background: Although alcohol withdrawal (AW) is common in hospitalized patients, few studies describe contemporary alcohol withdrawal management in hospitalized settings or compare practices to recent guidelines published by the American Society of Addiction Medicine (ASAM). We sought to describe patient characteristics and determine factors associated with complicated withdrawal and treatment duration in hospitalized patients with alcohol withdrawal.

Methods: We used the Alcohol Withdrawal Study (AWS), a cohort of hospitalized patients with AW from 19 Veterans Affairs (VA) hospitals across the US. Demographic and comorbidity data were obtained from the VA’s corporate data warehouse. Inpatient management and hospital outcomes were obtained by chart review. Treatment duration was the total number days of treatment, defined as last date minus first date of receiving AW medication. Complicated withdrawal (CW) was a composite of delirium and seizure. Using a Poisson model, factors associated with treatment duration were examined. Logistic regression was used to examine factors associated with CW. Both models were adjusted for demographic, clinical characteristics, treatment type, and medications.

Results: There were 594 patients with confirmed AW. They were mostly male (96%), white (74%) and unemployed (60%). More than half had an AW admission in the prior year, 28% had experienced prior alcohol withdrawal delirium, and 33% had a prior withdrawal seizure. When hospitalized, 56% were managed with symptom triggered therapy alone, 28% with fixed dose plus symptom triggered therapy, 13% with front load regimens plus symptom triggered therapy, and 6% with fixed dose alone. The most common medication given was lorazepam (87%) followed by chlordiazepoxide (33%), diazepam (14%), and phenobarbital (6%). Median duration of treatment was 3 days. Thirteen percent of patients developed delirium, 5% seizure, and 1% died. Factors associated with treatment duration included prior delirium, acute medical reason for admission, and prior intensive care unit admissions. Symptom triggered therapy alone and symptom triggered therapy with front loads were associated with reduced treatment duration. Lorazepam, gabapentin and phenobarbital were associated with increased treatment duration. Factors associated with complicated withdrawal included race (Black compared to White, AOR 3.1, 95% CI 1.5-6.4), prior delirium (AOR 2.3, 95% CI 1.2-4.4), prior seizure (AOR 2.1, 1.1-3.9), admission alcohol level (AOR 0.97, 95% CI 0.95-0.99) and thrombocytopenia (AOR 2.2, 1.2-3.8). Neither treatment mode nor choice of benzodiazepine were associated with complicated withdrawal.

Conclusions: Overall, our results support recommendations by the ASAM to manage patients with long-acting benzodiazepines using symptom triggered therapy alone or with front-loaded medications. We found that the most popular medication was the intermediate-acting lorazepam, which was associated with increased treatment duration. Fixed-dose regimens were also still commonly used and were associated with increased treatment duration. Symptom triggered regimens were also preferable to fixed dose regimens for preventing complicated withdrawal in univariate analysis, but the difference was non-significant in adjusted models.

IMAGE 1: Variables Associated with Treatment Duration

IMAGE 2: Variables associated with complicated alcohol withdrawal