Implementation of an Electronic Health Record Initiative to Reduce Deep Sedation in a Cardiac Intensive Care Unit

Background Leading cardiac and critical care society guidelines recommend limiting sedation among mechanically ventilated cardiac intensive care unit (CICU) patients, as deeper sedation is associated with worse patient outcomes. Yet, sedation use increased during the COVID-19 pandemic and remains high. This study sought to investigate if a low-cost, self-sustaining, clinical decision support tool decrease early deep sedation use in a CICU population? Methods This difference-in-difference analysis evaluates early deep sedation among adult patients on mechanical ventilation >24 hours, admitted to a single-center academic CICU or medical intensive care unit (MICU) 11/2020-3/2025. Patients receiving continuous anti-arrhythmics, neuromuscular blockade, and active cooling were excluded. An electronic health record-based clinical decision support tool, including physician and nursing behavioral nudges to utilize less sedation, was implemented 11/2023 in the CICU, but not MICU. After implementation, a difference-in-difference analysis was conducted to compare the pre-(11/1/2020-10/31/2023) and post-intervention (12/1/2023-3/31/2025) average hours of early deep sedation (Richmond Agitation and Sedation Scale ≤-3) in the first 48 hours after intubation between the CICU and non-intervention MICU cohort, adjusted for pre-specified covariates age, sex, body mass index, and acute physiology and chronic health evaluation score. Results and Conclusions The study included 1,610 patient encounters, 189 CICU and 1,421 MICU. The average hours of early deep sedation decreased in the CICU post-intervention, compared to pre-intervention (16.8 vs 12.4, p=0.04), but not the MICU during the same time periods (19.7 vs 21.2, p=0.06). In an adjusted analysis, receiving CICU care in the post-intervention period was associated with a 6-hour decrease in the number of hours of early deep sedation (95% CI: -11.1 to -1.5). Nurses reduced the continuous dose of fentanyl, propofol, midazolam, or dexmedetomidine 23% of the time the clinical decision support tool alerted. Future studies should evaluate the effectiveness of this low-cost tool in other medical and surgical populations.