Epic Deterioration Index: Clinical Use Case

St. Luke’s University Health Network (SLUHN) implemented Epic’s Deterioration Index (DI) to address preventable inpatient deterioration on adult medical-surgical units, where late recognition of decline contributed to in-hospital cardiac arrests and rapid response activations with unanticipated ICU transfer. Baseline FY2022 performance highlighted improvement opportunities, with survival to discharge following medical-surgical cardiac arrest at 25.8% and rapid response cases resulting in unanticipated ICU transfer at 32%. By FY2025, survival to discharge reached 38.0%, 16.5 percentage points above the 22.5% national benchmark, while codes per 1,000 discharges decreased from 3.45 to 2.56. The DI was selected to augment—not replace—clinical judgment by continuously analyzing real-time vital signs, laboratory results, and nursing documentation, and embedding predictive insights into existing nursing and rapid response workflows through Best Practice Advisories and automated escalation. A multidisciplinary governance model (clinical AI governance, rapid response/critical care committees, standards of practice, and IT change control) guides threshold setting, workflow alignment, and ongoing optimization to balance sensitivity and alert burden. Pre-implementation targets included improving cardiac arrest survival to greater than 28%, reducing unanticipated ICU transfers to below 30%, and sustaining reductions in codes per 1,000 discharges. The DI also supports health equity by providing objective, longitudinal risk detection that reduces variation related to communication barriers and inconsistent escalation practices.

Although SLUHN implemented Epic’s Deterioration Index, the underlying capability is not dependent on Epic. A comparable deterioration-detection model and escalation workflow could be developed internally or implemented in partnership with another organization or technology platform, provided it can continuously analyze relevant clinical data and integrate predictive insights and alerts into existing clinical workflows.