Comparison of Phenobarbital Monotherapy to a Benzodiazepine-based Regimen for Management of Alcohol Withdrawal Syndrome in Trauma Patients

Abstract

Abstract BACKGROUND Alcohol withdrawal syndrome (AWS) is associated with increased morbidity and mortality in the trauma population. Benzodiazepines (BZD) are standard of care for AWS; however, given the risk of delirium with BZDs and reports of BZD-refractory withdrawal, phenobarbital (PHB) has emerged as an alternative therapy for AWS. Safety and efficacy studies of PHB for AWS in trauma patients are lacking. Our aim is to compare a BZD versus PHB protocol in the management of AWS in trauma patients. METHODS We performed a retrospective cohort study at a level 1 trauma center of patients at risk for AWS managed with either a BZD or a low dose oral PHB regimen. Patients were excluded if they were taking benzodiazepines or barbiturates prior to admission, received propofol or dexmedetomidine prior to initiation of the study drug, presented with delirium tremens or seizures, or died or discharged within 24 hours of presentation. The primary outcome was complicated AWS (seizures or alcohol withdrawal delirium/delirium tremens). Secondary outcomes included uncomplicated AWS, therapy escalation, oversedation, delirium-, ICU-, and ventilator-free days, and length of stay (LOS). RESULTS 411 patients were identified; 118 received BZD, and 293 received PHB. The odds of developing complicated AWS with PHB versus BZD-based therapy were not statistically significant (OR 0.52; 95% CI, 0.21-1.39); however, patients receiving PHB were less likely to develop uncomplicated AWS (OR 0.08; 95% CI, 0.04-0.14) and less likely to require escalation of therapy (OR 0.45; 95% CI, 0.24-0.84). The PHB group had a LOS 3.1 days shorter than the BZD group (p = 0.002). There was no difference in ICU-, ventilator-, or delirium-free days. CONCLUSIONS A PHB-based protocol for the management of AWS is a safe and effective alternative to BZD-based regimens in trauma patients. Level of evidence Level IV, retrospective cohort