RWJBarnabas System Tied to 18% Drop in In-Hospital Deaths Using Epic Deterioration Index

A coordinated rollout that sent 15-minute risk scores as instant mobile alerts to Rapid Response Teams appears to have allowed earlier bedside evaluation and fewer deaths.

Overview

A multi-hospital study published in NEJM AI evaluated 23,132 high-risk patients across 11 RWJBarnabas hospitals and found in-hospital mortality for patients flagged as high risk fell from 23.1% to 18.6%.
The Epic Deterioration Index was set to recalculate every 15 minutes and to push 'Red' alerts for scores above 60 directly to on-duty Rapid Response Team mobile devices.
Rapid Response Team evaluations of high-risk patients rose from 25.3% to 37.5% after the rollout while transfers to intensive care units did not increase significantly.
The deployment included training, workflow changes and alert-suppression rules that blocked redundant notifications for patients already in ICU, on comfort care, or recently evaluated, and authors say the bundle—not the model alone—likely drove the benefit.
Researchers are moving to a next phase that will focus on detecting rapid rises in risk score ('velocity') to try to intervene earlier, and they position the operational model as one hospitals using Epic could replicate.