Background: Phenobarbital (PB) has important pharmacological advantages over benzodiazepines (BZDs) for treatment of alcohol withdrawal syndrome (AWS). The American Society of Addiction Medicine recommends PB monotherapy as an alternative to BZDs for prophylaxis or treatment of AWS by providers experienced with its use. Few studies have described the use of PB for AWS outside of academic centers and intensive care units (ICUs). We describe the implementation and evaluation of a protocolized approach to PB monotherapy for AWS at a 220-bed community hospital. The results of this quality improvement initiative are intended to provide practical information that may facilitate wider adoption of this promising strategy.
Methods: In March 2022, an EHR order-set for the use of PB monotherapy for AWS was adapted from a local partner institution. Focused education was made available to providers and nurses. The traditional BZD-based protocol with an option for adjunctive PB dosing remained an alternative to PB monotherapy. The novel PB monotherapy protocol was recommended for patients anticipated to experience moderate to severe AWS. Retrospective data analysis was performed to assess the impact of this quality improvement initiative. Patients with an alcohol-related discharge diagnosis were included from 9/24/2021 to 11/7/2022 (n= 788). We conducted two sets of comparisons: 6 months pre vs. 6 months post-intervention and PB protocol vs. BDZ-based protocol. We calculated the cumulative dose of BZDs and PB patients received. Outcomes included hospital length of stay, complications (seizure, delirium tremens, PB toxicity, intubation), treatment in the ICU, and maximum CIWA-Ar score by day of hospitalization. We performed bivariate analyses using either Welch t-tests or Pearson’s Chi Squared tests.
Results: Pre vs. post-intervention comparisons showed a >50% reduction in total BZD use and a 3.4-fold increase in total PB use without significant differences in evaluated outcomes. Notably, there were no differences in complications post-intervention. Patients treated with the PB monotherapy protocol were more likely to have a discharge diagnosis of alcohol withdrawal (96% vs 56%, p<.001) and had higher maximum CIWA-Ar scores in the first 24 hours after admission (mean 14.5 vs 10, p=.004). Patients treated with this specific PB monotherapy protocol had the same rate of ICU admission and intubation, similar CIWA-Ar after 24 hours of hospitalization, and shorter hospital length of stay (mean 5.7 days vs 8.6 days, p=.01) compared to patients treated with a BZD-based protocol.
Conclusions: We share the experience of a community hospital safely and effectively introducing a PB monotherapy approach for treating AWS in acute care and ICU settings. Using our EHR order-set, PB monotherapy was associated with equivalent to superior outcomes compared to a traditional symptom-triggered BZD-based approach. Patients treated with a PB protocol had evidence of more severe AWS than those treated with the BZD-based protocol, yet significantly shorter hospital length of stay. Our analyses are subject to methodological limitations including selection bias and confound; however, the practical nature of our order set implementation project makes our experience applicable to other hospitals considering the introduction of PB monotherapy for the treatment of AWS.