Epic to Pilot Decision Support Tools Using Real-World Data
In a recent wide-ranging interview with Healthcare Innovation, Johnston Thayer, M.B.A., R.N., director of clinical informatics and population health at Epic, described a new tool the company is getting ready to pilot called Best Care, which takes advantage of the EHR vendor’s huge Cosmos research database to provide better decision support at the point of care to clinicians by using real-world data about similar patients.
Thayer, who has experience as an emergency room nurse, said he is interested in the use of genomics, machine learning, and population health tools to transform care delivery. He said health systems are getting more sophisticated in their use of population health tools that can overlay the data to see, for instance, how many diabetic patients are overdue for a foot exam or for a regular screening. Previously, identifying those patients would have taken a long time combing through all of your chart. Using tools in Epic, he said, you're able to automatically identify all of those patients, automate that follow-up, and make sure that patients are getting in for those visits. “There's a whole set of tools for doing that type of work across all different types of populations,” he said.
Now Epic is starting to take advantage of its large data sets. Cosmos brings together data from more than 160 million patients representing billions of clinical data points and creates a platform where that de-identified data is available for research.
The new Best Care tool will offer the capability to see treatments and outcomes for patients similar to the one right in front of the clinician, Thayer explained. “Let's say I'm a physician and I've got a new hypertension patient named Charlie who has just been diagnosed and we're looking at prescribing him a new medication,” he said. “Even if I'm very studious and up to date on all of the current literature on hypertension and treatment options, it's unlikely that I've read something that deals specifically with patients like Charlie —looking at his age, his comorbidities, his previous treatment attempts, and other factors that make him unique. Best Care will allow me to see information from Cosmos about how patients that are similar to Charlie have been treated in the past and the outcomes of some of those treatments in terms of whether or not they were able to successfully lower the patient's blood pressure and to what extent, so it will really empower physicians as they're making decisions at the point of care to have that real-world data available to them.”
Thayer said Epic is rolling this out in a very safe and methodical fashion. “We're starting a pilot within the year where that information will initially be available within the Cosmos portal, but the goal is that the information will be directly embedded into the workflow. And we're going to start out with the hypertension use case as an initial pilot concept, but then we plan to work with the Epic organizations and some of the specialists in a variety of areas to identify use cases that we will continue to move into after that.”
Another new way that the company is taking advantage of Cosmos involves a project called “Lookalikes,” which Epic is rolling out in November. If a physician has a patient with a rare disease or a unique constellation of symptoms that the physician hasn’t worked with previously, the program will look out across all of the Cosmos patients and identify similar patients. In this case, the goal is to identify providers who have seen similar patients so that the clinician can connect with them and have a conversation, do a “curbside consult,” and ask them questions about how they treated that patient and what some of the outcomes were. “We really see it as an opportunity to improve and increase collaboration at a scale that's never been possible before,” Thayer said, “figuring out who has treated a patient like the one I'm seeing today. Before this, there just weren't any tools like that. We're really excited for that to come out and to see how that starts to change some of the physician collaboration opportunity.”
I asked Thayer whether Epic, other than at its annual users group meeting, has mechanisms that allow best practices or innovations to be shared and disseminated across health system customers.
“Absolutely,” he said. “Anytime we see an organization doing really exciting things that we think are measurable and reproducible at other sites, will write up what's called a ‘clinical program’ that outlines how they did it, what outcomes they saw, the tools in Epic that allowed it, the training, the change management that went into it.”
I mentioned that in looking at an agenda for the upcoming AMIA meeting, I saw a session about health systems that were creating a culture of EHR builder/developers amongst their clinicians. I asked him how Epic encourages this phenomenon.
“We have a program for helping to train physician builders and we have clinical builders as well, depending on the type of area that you work in and what your background is, but the goal of those is really so that you learn some of the more nuanced tools that we have available in Epic to build those out for your specialty.” He said there are specialty-specific cases where providers like to document things certain ways. Rather than putting in a request for the IT team to build out a new form, the clinicians actually learn how to do that themselves. “KLAS, which measures user satisfaction with the EHR, has actually been able to show that having physician builders embedded in the specialty makes a difference in terms of the satisfaction of users in that area.”
Among Thayer’s responsibilities is being involved in strategic planning for research and development, which he said is rewarding.
“We're working with our development teams to look at the things that we need to be working on next, and that are going to have the biggest impact for our community,” he said. Artificial intelligence and ambient voice are two examples. “Both are definitely big areas for us. We recently did work around creating some new tools for a number of our predictive models. I was specifically involved in working with our sepsis algorithm, and how to use tools like this. AI is still relatively new to healthcare and having a platform for actually utilizing that information and making decisions that are informed by some of that data that we're collecting is still a challenge for many groups. Being able to help put together a framework for doing that was a lot of fun. Ambient voice is definitely an area of interest for many groups. When you imagine the future of healthcare, many people think about the opportunity to just have a conversation with a patient and be able to automatically capture that conversation, turn it into something structured and meaningful and really reduce some of the manual input of information into a system.”
I asked Thayer if part of the decision-making process involves determining which things Epic wants to build in house vs. doing integrations with outside vendors.
“There are definitely areas where we would much prefer to partner with others who are doing the work. For instance, on the ambient side we are working with voice vendors who have developed platforms and programs for doing the actual natural language processing, rather than doing some of that ourselves,” he said. “There are also areas where we do actually work on some of it, but we also recognize we could never work on all of it. The AI space is a good example. We've got our own models and algorithms that we've developed in house at Epic, but there are so many possibilities in the space. We have a set of tools for embedding algorithms directly into Epic, the exact same way our tools get embedded. The goal of that is so you can use the best of what's out there and still have that integrated experience for your providers and clinicians.”
In consulting with health systems on implementations, Thayer gets to work with clinical informatics teams up close all the time. I asked him whether he has seen the roles and the relationships between the informatics executives and operations executives evolve.
“That's absolutely been evolving,” he said. “As you know, informatics is still relatively new in the grand scheme of things, and especially some of those executive-level positions. For a while, there was this artificial delineation between the operations side and the informatics or IT side under the CIO, and there was not a lot of alignment, especially from the strategy perspective. Within the groups, they often operated as independent organizations that were driving from their own plans and objectives. But recently, we've seen a lot more cases where those are becoming aligned as part of the overall organization. So now it's very common for a CNIO to either report directly to the CNO on the operation side or at least have a dotted line, so that they are embedded just as much on the business and operations side as they are on the informatics and IT side.”
I asked whether the Epic App Orchard is starting to allow third-party developers to create incremental improvements or add-ons rather than Epic having to build everything internally.
He said tools in the App Orchard could be for small changes or specialty-specific use cases all the way up to larger tools that are really intended to make an impact on the broader organization. “It’s just providing those APIs as the way to both send data out from Epic as well as write data back in when appropriate to create that integrated experience, so that physicians, nurses and others don't feel like they need to go out to some separate portal to find those details,” he said. “The workflow is so important and asking a clinician to go reference an outside portal and compare it to what's they're seeing in the system is really a difficult thing for them to do. So instead, pooling all of that together allows for a much more integrated and seamless experience. We see that both for groups that are creating offerings that they're intending to sell, but we also have it available for health systems when they've innovated and created some of their own custom development. They can make that available in the App Market, either to share freely or as a way to cover some of the costs of developing that tool.”
In addition to working with informatics execs at health systems that are live on Epic, Thayer also consults with groups that are in the process of going live as they're working through change management and setting up a governance structures for making decisions. I asked him how that process was impacted by the pandemic at organizations that wanting to make the transition to Epic, but were operating in an emergency mode.
“I don't know the exact numbers, but we actually had a lot of groups that still went live on time, even with everything that was happening, which required a lot of pivoting and a lot of flexibility,” he said. “But we have great teams and we work with great organizations, so in most cases we were still able to go live on time. There are certainly exceptions to that, but it involved a lot of figuring things out as we went along.”
