At Baptist Health, an Innovative Approach to Optimizing ED Traffic

For decades, hospital executives have been frustrated by processes around managing patient flow in emergency departments. There are of course numerous elements to the tangle of challenges involved, but so often, sub-optimal processes, combined with increased and/or fluctuating patient demand, lead to problematic ED boarding situations, which in turn lead to sub-optimal patient outcomes, not to mention staffing headaches for ED leaders.

That’s why senior leaders at the 225-bed Baptist Health Medical Center-North Little Rock in Arkansas three years ago launched Operation Raptor, a predictive analytics initiative designed to give staff a real-time view of demand, bed availability, and staffing. As a result, they have returned over 30,000 hours of unnecessary wait time to the ED patients in their area, increased admissions by 6 percent, and drastically decreased ambulance diversions.

Cody Walker, president and CEO of Baptist Health North Little Rock, has been leading the initiative, which began in his facility and has since then been implemented at all 11 Baptist Health facilities in the Little Rock metro area. In their work, Walker and his colleagues have been collaborating with professionals at LeanTaaS, utilizing their iQueue technology to help manage and use the data involved. Healthcare Innovation Editor-in-Chief Mark Hagland spoke recently with Walker about the initiative. Below are excerpts from that interview.

What was the origin of the initiative?

Baptist Health, including my hospital, but also overall, decided a couple of years ago to double down on predictive analytics with regard to inpatient flow and length-of-stay improvements. It was at a time, coming out of the pandemic, where capacity was at an all-time high. And ED boarding and length of stay were big issues. So we partnered with LeanTaaS, hoping we could leverage generative AI and predictive analytics, to better execute capacity management.

So we’ve been at it for some time, and it’s involved an overhaul of our culture, by leveraging this data. It’s led to significant improvements that have directly impacted patient care. Comparing where we are now to when we started in April 2022, we’ve seen a 34-percent reduction in length of stay, and 40-percent reduction in overall discharge processing time. It had typically been taking us two to two-and-a-half hours to get a patient out the door from the time of order to the time they departed. Now the average is around 80 minutes. So we’re seeing significant reductions because of the transparency of data. So it’s having that forward-facing view of what’s about to happen, versus it happening to us.

Tell me about how the process has unfolded?

Prior to LeanTaaS, we were meeting in a room, a small group of leaders at the hospital, pulling data manually—spreadsheets or even handwritten on paper. We pulled that from our EHR [electronic health record]—and we had already-outdated data, and sometimes even inaccurate data, by the time we pulled it. It was a very manual process and very human error-dependent. And oftentimes, there was an inability to execute. We realized we couldn’t do it within our current EHR; it was so labor-intensive to bring outdated data into the room, and speed us up to a consistent, high-level process. Instead of doing it with fifth-grade math, we can now predict outcomes.

We’re now not only looking at real-time better that’s accurate, and are actually looking at tomorrow’s data, and already know the names of the patients, their room numbers, and so on. Prior to doing this, we were looking at yesterday and what didn’t go well.

In other words, prior to applying innovative information technology, you and your colleagues had been engaging in a lot of guesswork, correct?

Yes, and it was really dependent upon who was pulling the information and how much time they had to look at the chart to get the data out.

Which stakeholders were involved in setting up the initiative at the start, and how did the leadership around the initiative evolve over time?

Early on, it was myself, our chief nursing officer, and a handful of nursing leaders. So really, it was five to eight people in a room, at a pretty high level, trying to analyze the overall facility situation. Today, we have about 30 people in a room, involving hospitalist attendings on staff; we have every nurse leader at a manager level in a room, given a count of their specific rooms and patients, including case management, utilization review, therapy, pharmacy, every ancillary leader is in the room. So the plan is already in the place as we walk into the room. We’re talking about anything that would keep us for executing the plan.

So we’re really talking about ways to execute on the plan, and any issues that might not be revealed in the data. So the room is much broader, and the issues are accurate to the patient we’re discussing; it’s no longer hypothetical. We’re talking about specific results we’re looking to execute on.

In other words, you can zero in on a specific patient, then? The proverbial Mrs. Smith?

Yes, and we have very accurate predictions of when Mrs. Smith will leave, and to what disposition—rehab, home, home health, SNIF. We’re not having to make those predictions anymore, the generative AI is driving that prediction, and it’s accurate to her condition and to where she’s going to go. So we’re proactively doing things in parallel.

What about the leadership elements in this?

Capacity management and patient flow are operational issues, but there has to be a level of executive urgency around this work, too, in order for an initiative like this to succeed. What we see on the horizon, with recent federal legislation, it’s even more of a level of urgency. And every one of your readers understands ED boarding issues and the capacity issues involved. It will only get more significant with uncompensated care, and the EDs are going to face huge capacity issues. So we need good leaders who are lifelong learners, and a sense of urgency to deal with these problems.

CIOs, CTOs, CMIOs, CMOs, CNOs, and other senior leaders, what should they be thinking about this subject right now?

I think for us, we’re in a situation that a lot of organizations are in today: do we build, buy, or partner, to solve this problem? And we attempted to build first, but we realized that it often took too long to get what we were looking for as a solution, and it was often inaccurate. So we were left with the only options of buying or partnering. But with LeanTaaS, it’s truly been a partnership. We often will have an idea this morning, will reach out to LeanTaaS, and the solution will be built by the afternoon. And that allows us to test hypotheses in real time without having to wait for the solution to come to fruition. That’s what I think has been transformational. We have solutions same day or next day, because we have a partner. It would be too expensive and time-consuming for us to do this to the same level.

How do you see next few years in terms of generative AI and its potential?

Everyone likes to think they’re unique; but generative AI has a way to debunk that. It helps us constantly learn about our specific hospital and the variables involved. And for us, leaning into the ways to relearn our own specific data about our own hospital and our own barriers to success, we’re leaning in, both on this specific topic and around everything else. And generative AI allows us to constantly learn. So we’ve become full-on believers, per our experience with generative AI. We used to have our case management staff put in potential discharge dates, and that process was only 40-percent accurate. Now with LeanTaaS’s help, we work based off the predicted discharge date, which is 90-plus-percent accurate, and that relieves our staff to be able to do other things. That gives us confidence and trust in the work, to help us use generative-produced data.

I’ll just add that when you think about how the technology can help, people think about generative AI as being fully agentic. But the work that Cody and his teams do every day is highly complex. Generative AI gives them an assistant all along the way, and stop the chart-diving, so the person using it can be the most effective. When we think about the labor shortages and the administrative burden on people to get their job done, that’s the real opportunity in the near term, to elevate everybody to their top of their game.