Rationale: Several institutions have implemented phenobarbital-based pathways for the treatment of alcohol withdrawal syndrome (AWS). However, little is known about the care processes, effectiveness, and safety of phenobarbital-based pathways for intensive care unit (ICU) patients. Objectives: To examine clinician acceptability and feasibility and patient outcomes after the implementation of a phenobarbital-based pathway for medical ICU (MICU) patients with severe AWS. Methods: We conducted a mixed-method study of a quality-improvement intervention designed to improve the workflow without deleterious effects on outcomes. We used semistructured, qualitative interviews and surveys of clinicians to assess the acceptability and feasibility of the phenobarbital-based pathway and a previous benzodiazepine-based pathway. We used a noninferiority interrupted-time-series analysis to compare mechanical ventilation rates before and after implementation among MICU patients within an urban safety-net hospital who were admitted with severe alcohol withdrawal. We explored several secondary outcomes, including physical restraint use and hospital length of stay. Results: Four themes related to clinician acceptability and feasibility of the phenobarbital-based pathway emerged: 1) designing a pathway that balanced standardization with clinical judgment promoted acceptability, 2) pathway simplicity promoted feasibility, 3) implementing pathway-driven care streamlined the workflow, and 4) ad hoc implementation strategies facilitated new pathway uptake. Two hundred thirty-three and 252 patients were initiated on the benzodiazepine- and phenobarbital-based pathways, respectively. The rate of mechanical ventilation decreased from 17.1% to 12.9% after implementation of the phenobarbital-based pathway, and an adjusted mean difference of -4.9% (95% upper confidence interval [CI]: 0.7%) corresponding to relative change in the 95% upper limit of 4%, which was below the a priori noninferiority margin, was shown. After implementation, use of physical restraints decreased from 51.6% to 32.4% (mean difference, -18.0%; 95% CI: -26.4% to -9.7%), and the hospital length of stay was shorter (8.6-6.8 d; mean difference, -1.8 d; 95% CI: -3.4 to -0.2 d). Conclusions: Clinicians believed that the phenobarbital-based pathway was more efficient and simpler to use, and patient mechanical ventilation rates were noninferior compared with the previous benzodiazepine-based pathway for the treatment of severe AWS.
Keywords: barbiturates; benzodiazepines; critical illness; epidemiologic methods.