Sentara Health, Cleveland Clinic Making Significant Strides in Sepsis Care

Sentara Health has reduced sepsis mortality by 44%, and Cleveland Clinic is using an artificial intelligence tool to improve early detection of sepsis.

Sentara Health and Cleveland Clinic have been launching initiatives to improve the care of sepsis patients.

Sepsis is the body's extreme reaction to an infection that can result in tissue damage and organ failure. Annually in the United States, there are at least 1.7 million adult hospitalizations for sepsis and at least 350,000 deaths from the condition, according to the Centers for Disease Control and Prevention. Sepsis is a top concern for CMOs because the condition is a leading cause of death in the hospital setting.

Sentara Health has implemented a series of initiatives to improve sepsis care since 2021. The effort has achieved remarkable results, including a 44% reduction in sepsis mortality from 2021 to 2025.

"We identified that we were not performing at the level that we wanted to achieve with our sepsis patients," says Doug Browder, MD, emergency department medical director at Sentara Leigh Hospital. "We identified that we were not performing at the desired level through the Centers for Medicare & Medicaid Services (CMS) sepsis management bundle."

To begin the health system's sepsis care improvement initiatives, Sentara Health formed a clinical process improvement team, with an initial goal of improving performance on the CMS sepsis management bundle. The members of the clinical process improvement team included a physician leader, a process improvement engineer, a nursing leader, administrative leaders, and a CMO. Browder served as the physician leader.

With the clinical process improvement team in place, Sentara Health then created a sepsis care improvement and leadership structure and, with teams at each of the health system's hospitals. These teams included emergency department nursing leadership, ICU leadership, and physician leadership such as a hospitalist or emergency room physician.

"The hospital sepsis improvement teams looked at their performance on sepsis care and implemented solutions that could apply to their hospitals," Browder says. "Those teams meet on a monthly basis, and we take the lessons learned at the hospitals and share them across the health system."

Areas where Sentara Health made improvements included the appropriate administration of fluids based on the CMS sepsis management bundle, identification of organ dysfunction and shock, and appropriate treatment and documentation of hypotension.

The most important aspect of improving sepsis care is viewing sepsis as an emergency, according to Browder.

"Before we launched our sepsis care improvement initiatives, Sentara was similar to other health systems, where sepsis was not considered as emergent as a stroke or an ST-elevation myocardial infarction," Browder says. "When you start paying more attention to sepsis and view it as a time-based emergency, you can improve morbidity and mortality."

Doug Browder, MD, is emergency department medical director at Sentara Leigh Hospital. Photo courtesy of Sentara Health.

The main tool for improving sepsis care at Sentara Health has been the sepsis order set in the health system's electronic medical record (EMR), Browder explains.

"Our goal for the sepsis order set was to give our clinicians and nurses a better pathway to care for sepsis patients and ensure that care teams could effectively enter all the elements of the CMS sepsis management bundle," Browder says. "It was crucial to make the sepsis order set in the EMR work for the clinicians and the nursing staff."

An effective sepsis order set makes the job easier, reduces clicks in the EMR, decreases cognitive burden, and ensures patients get all of the appropriate care, according to Browder.

"We worked with focus groups of our physicians and nurses to create an order set that had all of the care bundle elements and appropriately added documentation into the EMR that spelled out why it was necessary to provide certain aspects of the care such as early identification of sepsis, fluid administration, and organ dysfunction," Browder says.

Early detection of sepsis and prompt treatment is crucial in achieving positive clinical outcomes. At Sentara Health, early detection of sepsis includes leveraging the health system's EMR to screen for the condition, Browder explains.

"Initially, we focused on any patient complaint that indicated a potential infection, particularly in our emergency departments," Browder says. "We have developed EMR tools that can detect hospital-acquired sepsis. We also have educated our nursing staff and clinicians about the importance of recognizing subtle changes in vital signs for early identification of sepsis."

Efforts to boost early detection of sepsis at Sentara Health have also included community education and outreach.

"We have improved awareness of sepsis in the community because many people in our community lacked an understanding of sepsis," Browder says.

Rohit Chandra, PhD, is executive vice president and chief digital officer at Cleveland Clinic. Photo courtesy of Cleveland Clinic.

Cleveland Clinic adopts AI tool for identifying sepsis

To improve the early detection of sepsis, Cleveland Clinic has adopted an artificial intelligence tool developed by Bayesian Health.

"Bayesian Health's AI-enabled sepsis tool integrates with our electronic health record," says Rohit Chandra, PhD, executive vice president and chief digital officer at Cleveland Clinic. "The AI tool makes sure that all of the data in our EHR that may be relevant for detecting sepsis is available. The AI tool monitors our patients constantly and looks for changes in vital signs."

The AI tool looks in the EHR for any early signs of deterioration that might indicate the onset of sepsis. If a patient's vital signs cross a certain threshold, the AI tool alerts frontline caregivers through the EHR. Then a nurse or physician conducts a bedside exam to see whether they agree with the indications of the algorithm. If they agree with the algorithm, they prescribe treatment such as fluids and antibiotics as quickly as possible.

"Any time the vital signs of a patient change in a way that indicates the onset of sepsis by the AI tool, it triggers an evaluation," Chandra says. "A physician may evaluate a patient for sepsis once a day, but the AI tool is running 24/7 and assessing data all the time. This AI tool is more likely to catch sepsis early than other approaches."

Data collected at Cleveland Clinic in pilot programs implemented at two of the health system's hospitals indicates that the AI tool is effective.

In one data set, Cleveland Clinic compared the health system's historical alerting system for sepsis with the AI tool. For the patients who were eventually determined to be septic, the patients who were identified by the AI tool were seven times more likely to have an alert in a timely fashion and trigger prompt intervention.

In another data set, Cleveland Clinic looked at false alarms. Compared to the health system's historical alerting system, the pilot program found that the AI tool generated 85% fewer false alarms.