Developing a Learning Health System Approach to Sepsis

When she worked at the Wake Forest University School of Medicine, Stephanie Taylor, M.D., M.S., led development of the Sepsis Transition and Recovery (STAR) program, which involved a navigator-led, telehealth-based strategy designed to improve care for high-risk sepsis survivors. Now as chief of hospital medicine at the University of Michigan, Taylor has brought that learning health system approach to sepsis to Ann Arbor.

During a recent U-M Department of Learning Health Sciences meeting, Taylor described some of her work testing new interventions. 

Taylor began her talk by saying sepsis offers a big opportunity to improve outcomes, and it is a major focus of health systems, both because it involves high cost and high mortality, and because there are some policies around reimbursement and public reporting for sepsis outcomes — so health systems really care about this.

“When I got here, I tried to bring a lot of what I had built at Wake Forest here to build a sepsis learning health system. We call it acute care embedded research. Sepsis is the central part, but it also involves other things that have to do with inpatient conditions,” she said. 

As chief of hospital medicine, Taylor says she gets to see all of the pieces of the learning health system cycle directly. She takes care of patients clinically and is an operational leader. “I’m making decisions about what types of interventions to support, to give resources to, and then I'm accountable for all of those things as well,” she said. “My research focuses on this area, too.”

Her team first took EHR data and built out a deep, granular, feature-rich data set of sepsis patients so that they could learn from that. “From there, we listened to our patients,” Taylor explained. “We listened to health administrators about what interventions they were planning, what problems they wanted to solve, and depending on what stage they were on in the cycle, we either designed interventions to solve problems or we built trials to comparatively test different interventions or interventions vs. usual care,” she explained. 

They also focused on implementation science. “There are some evidence-based interventions for which there's not wide uptake, so implementation science is a really big deal,” Taylor noted. “There are a lot of unknowns in terms of treatment, so a lot of our work is designing interventions and testing new things to see what will work.”

Another interesting thing about sepsis is that historically the focus has been on inpatient treatment and outcomes, she said, “but what we've learned from patients over the last decade is that there are persistent long-term effects and bad outcomes that last a year or two, or even longer after hospitalization. So a new part of our research has been understanding how to support patients for full recovery — not this easy-to-measure inpatient mortality outcome, but the more tricky to measure but really important patient outcomes in terms of long-term recovery. We've been working on building interventions to improve that, as well as good ways to test the effect of those interventions in terms of outcomes.”

When asked about challenges to doing this type of work, Taylor mentioned the culture of the clinical teams as an area that needs transformation. “There's autonomy as a practicing clinician, but this openness to generating evidence and acting on that evidence in areas where there's equipoise has been challenging,” she said. “In our pragmatic trial work, we randomize people to do one thing or another where there's complete equipoise, obviously, or we wouldn’t be doing the trial. But we'll have clinicians who just say, ‘No, I'm not doing that because I think A is better.’ And we say, ‘Well, there's no evidence that A is better.’ And they say, ‘Well, I'm doing it because I think A is better.’”

Taylor said she thinks that's something you have to overcome as a culture. “Places like Vanderbilt, NYU and other places have made that happen relatively quickly, but I think there has to be a critical mass of people doing that type of work to get everyone on board with an understanding of why that's so important,” she said. 

On the flip side, she said, her team gets messages from the quality team saying they are rolling out an initiative. Her team asks if it works, and the response is '"We think it works." Taylor then asks, “Well, how do you know? Let's find out if it works.” And they say "No, because that wouldn't be fair. We want everyone to get it because we think it works."

So Taylor said there is some cultural transformation that still has to happen. “One of the things we really emphasize is the academic part of of the work,” she added, “making sure that we're directly helping local outcomes, but also creating generalizable knowledge that helps this condition across all populations.”