Using AI and people power to combat sepsis

By Kat Jercich for Healthcare IT News

Sepsis, the body’s damaging over-response to an infection, affects at least 1.7 million adults in the United States every year.

Early treatment is key to staying ahead of the condition, which can be life-threatening. But sometimes the clinical signs can be hard to spot – making artificial intelligence tools particularly useful.

This past year, University of Missouri Health Care was recognized with a HIMSS Davies Award for its work pairing the National Early Warning Score algorithm with a rapid response team to contribute to a reduction in sepsis mortality.

Developed by the Royal College of Physicians, “the National Early Warning Score is a wonderful system that uses measures the nurses are already capturing to give you an idea that you might want to take a closer look at your patient, that they might be showing signs of early decline,” explained Dr. Thomas Selva, chief medical information officer at MU Healthcare and medical director for the Tiger Institute for Health Innovation.

“We like to say there is nothing ‘new’ about the NEWS score, but what we did was look hard at how we were not responding as well to early signs of clinical decline in patients,” said Selva in an interview with Healthcare IT News.

Selva, who will be presenting at HIMSS22 this year, explained that the team took the score, reviewed how nurses were already documenting care, and then combined input from the nursing staff and rapid response teams.

They also examined how to optimize processes to get scores in front of caregivers at key points in their workflow.

“This means the rapid response teams can come in earlier before patients have critical decline and either prevent that decline from occurring or get them to a higher level of care before decline occurs out on the floors,” Selva said.

The retooling required careful consideration, he explained, to ensure that no one would feel usurped. In a large academic medical center, physicians – especially those on the night shift – may have difficulties getting to patient bedsides as quickly due to other care obligations.

“With our implementation of the NEWS score, we were able to significantly increase the amount of rapid response team activations, significantly decrease the amount of code blue activations on the floor and, overall, reduce mortality over the timeline we were following – and that has continued,” said Selva.

Selva credited Ben Wax, a nursing informaticist at the Tiger Institute IT collaboration between MU Health and Cerner, as helping make the implementation a reality at the institution.

“What we have learned throughout this entire journey … is that, oftentimes, quality improvement teams will, in the end, want to make some change in our electronic medical record or health IT environment to ‘standardize the delivery of care,'” said Selva.

“Many times, when they come to that conclusion, either it’s a solution we don’t have, or can’t implement, or we already have one that’s very close to what they need; they just didn’t realize it was there because these health IT implementations are very complicated,” he continued.

So, rather than having the clinical informatics come in at the 11th hour, Selva says he’s learned to embed them in quality improvement initiatives early.

“They can either provide advice and consent, or at least guidance, as the teams are coming up with solutions, and then help them shape those solutions so that they’ll be much more effective,” he said.

Failing to do so, he said, will almost always lead to frustration.

“The role of a clinical informaticist is important because they are not just IT specialists,” he added. “These are people who really understand the workflow in their area of expertise.”

Selva also stressed the importance of health IT implementation governance.

“All too often in health IT implementations, we put an alert in the system and it stays there forever, even though it is not achieving the goal that you wanted it to achieve,” he said. “You have to put governance around that team, so you want input from not only physicians, but your nursing staff and your clinical informaticists.

“We follow carefully the five ‘rights’ of clinical decision support,” Selva explained: The right information, to the right person, in the right format, at the right time in their workflow with the right intervention.

“You want to make sure there is good evidence behind the alert and then to have good measures in reporting as well,” he said.