The basics:

• RWJBarnabas, Rutgers: Epic AI tool reduced in-hospital deaths among high-risk patients
• Study tracked 23,132 high-risk patients across 11 RWJBarnabas Health hospitals
• Mortality dropped from 23.1% to 18.6% — an 18% reduction in risk-adjusted odds of death
• Epic Deterioration Index alerted rapid response teams every 15 minutes using EHR data

The promise of artificial intelligence in healthcare looms large, and new research from RWJBarnabas Health and Rutgers Robert Wood Johnson Medical School shows how early-warning systems harnessing artificial intelligence can help lead to fewer deaths among high-risk patients.

The partners published the findings July 29 in NEJM AI, a journal from the New England Journal of Medicine Group. The study evaluated outcomes among 23,132 high-risk patients across 11 of RWJ’s hospitals. According to the results, mortality among the cohort dropped from 23.1% to 18.6% after implementing the AI-enabled early-warning system.

Researchers reported an 18% reduction in risk-adjusted odds of death.

The Epic Deterioration Index continuously analyzes information that exists in a patient’s electronic health record. These include vital signs, laboratory results, nursing assessments and age. The AI-enabled tool uses that data to ID patients at increased risk of serious clinical decline.

Beyond that, the EDI recalculates risk every 15 minutes. The system then alerts rapid response teams when patients reach the highest-risk category.

According to the lead author of the study, having timely critical care insights can help change outcomes.

“Our goal was to identify patients earlier, before they reached a point where intervention becomes much more difficult,” said Dr. Thomas Nahass, vice president of health informatics and intensive care physician at RWJBarnabas Health, as well as assistant professor of medicine at Rutgers Robert Wood Johnson Medical School. “The deterioration index gives us an earlier point in time.”

“Every minute matters when a patient’s condition begins to worsen,” said co-author Dr. Andy Anderson. “This study demonstrates how AI-enabled tools, when paired with experienced clinical teams can help us identify patients at risk sooner and deliver the right care at the right time.

“These findings highlight the potential for innovation to improve quality, safety and outcomes for the patients we serve,” said Anderson, also chief medical and quality officer for RWJBarnabas Health.

Before evaluation, the paper partners said they spent several years developing and implementing a systemwide approach to integrating EDI across RWJBarnabas Health.

Academic medical center Robert Wood Johnson University Hospital in New Brunswick first piloted the tool. That process helped refine how and when alerts arrive, established automatic notifications to rapid response teams, trained clinicians on its use and continuously monitored performance.

Next, researchers from Rutgers got involved to help evaluate the impact of the approach as well as rapidly roll out the platform to 10 other hospitals. The study included two large academic, six community teaching and three nonteaching community medical centers in total.

When patients reach the highest-risk threshold, EDI sent automated notifications directly to hospital rapid response teams. Following the rollout, activations for these groups among high-risk patients increased from 25.3% of hospitals stay to 37.5%, the partners found.

This study demonstrates how AI-enabled tools, when paired with experienced clinical teams can help us identify patients at risk sooner and deliver the right care at the right time.
—Dr. Andy Anderson, chief medical and quality officer, RWJBarnabas Health

And while rapid response increased, the number of transfers to intensive care units did not, according to the study. Meanwhile, mortality rates also declined substantially.

“This is what an integrated academic health system is for,” noted Dr. Stephen O’Mahony, senior author of the study. “We combined Rutgers methodological rigor with the operational reach of 11 RWJBarnabas hospitals. The mortality benefit was not produced by an algorithm but by the partnership around the algorithm,” the RWJBarnabas Health senior vice president and chief medical information officer added.

The researchers note a combination of factors likely led to the mortality benefit. They sited staff education, enhanced clinical awareness, EHR alerts and the automated rapid response notifications as coming together to create a coordinated, systemwide approach.

Looking ahead

Closing the paper, the authors advised, “Future research should focus on optimizing the effectiveness of EWS [early warning system] alerts for enhanced visibility and sustainability, as well as understanding the specific elements of RRTs [rapid response teams] that may lead to successful clinical rescue.”

Beyond New Jersey, the study could have implications nationwide, as well.

The EPI tool evaluated is already available from electronic health record software system Epic, offering a wide platform for adoption. According to Fierce Healthcare, Epic’s growing coverage area encompasses 43.7% of the acute care EHR market nationwide. It also boasts a 56.9% market share of hospital beds, according to cited KLAS Research data.

The next phase of the initiative for RWJ and Rutgers will focus research on identifying patients whose risk scores are rising rapidly to potentially enable even earlier intervention.