Multicenter Evaluation of Phenobarbital in ICU Patients With Refractory Alcohol Withdrawal

Introduction: Alcohol withdrawal syndrome (AWS) significantly contributes to morbidity and mortality among ICU patients. While benzodiazepines are first-line treatment, resistant alcohol withdrawal (RAW) often requires adjunctive phenobarbital. Factors predicting phenobarbital response remain unclear. This study aimed to identify predictors of successful phenobarbital response (CIWA-Ar ≤10 within 24 hours) in ICU patients with RAW.


Methods: We conducted a retrospective cohort study involving adult patients admitted to ICUs across five Mayo Clinic locations between January 2018 and December 2024. Included patients had severe AWS treated initially with high doses of benzodiazepines (at least 24 mg lorazepam equivalent within 4 hours), followed by intravenous phenobarbital. We collected data on demographics, medical and psychiatric history, medications, and laboratory values. Statistical analyses using univariate and multivariable logistic regression were conducted to determine predictors of phenobarbital response.


Results: Of 163 eligible patients, 135 (82.8%) responded to phenobarbital, while 28 (17.2%) did not. Responders and nonresponders had comparable demographics except for a higher incidence of depression among responders (56% vs. 32%; p=0.023). Multivariable analysis revealed three significant predictors of phenobarbital response: history of depression (OR 2.98, 95% CI 1.18–7.53; p=0.021), higher initial phenobarbital dose per 10 mg increment (OR 1.03, 95% CI 1.00–1.06; p=0.050), and lower baseline CIWA-Ar scores (OR 0.91 per point increase, 95% CI 0.86–0.97; p=0.003). Model discrimination was acceptable (c-statistic=0.74) with good calibration (Hosmer–Lemeshow p=0.47). Phenobarbital responders demonstrated significantly lower median CIWA-Ar scores at 24 hours (8 vs. 18, p< 0.001) and 48 hours (7 vs. 9, p=0.031), while ICU length of stay was similar between groups (73 vs. 88 hours, p=0.64).


Conclusions: For patients experiencing severe RAW in the ICU, having a history of depression, receiving higher initial doses of phenobarbital, and having lower initial CIWA-Ar scores were independently associated with better treatment outcomes. Further studies are necessary to validate these findings and refine treatment strategies.