As physicians in practice become more and more burdened by administrative work, they’re finding that those burdens are impacting their sense of their effectiveness in patient care delivery. As a result, many medical group leaders are seizing the opportunity to implement artificial intelligence tools that their physicians can use. For example, AI can be leveraged to compile the most relevant patient data into a concise clinical summary before each patient visit. This enables doctors to review all of a patient's information in one place, saving 9 minutes per patient on chart reviews and helping to reduce burnout by 23 percent, according to an independent study by the American Academy of Family Physicians, and identify 500K conditions previously never documented.
One physician group that has taken the plunge is the Jefferson City Medical Group, a multispecialty medical group in Missouri’s state capital, with 70 physicians practicing across 30 medical specialties, and whose doctors see more than 80,000 patients per year. What’s more, the imperative to lift administrative burdens off its physicians is even more pronounced at Jefferson City Medical Group, as JCMG is one of nine organizations participating in the seven-state Stratum Med ACO, one of the largest accountable care organizations participating in the Medicare Shared Savings Program.
Beginning in 2023, Jefferson City Medical Group’s leaders partnered with the New York City-based Navina, an AI solutions provider. A study of Navina’s core AI Assistant solution published in January 2023 by the Leawood, Kan.-based American Academy of Family Physicians (AAFP) found “a significant reduction in burnout, including a 38-percent reduction in visit preparation time and a 23-percent reduction in burnout itself.
Benjamin Cook, D.O., a member of Jefferson City Medical Group’s board of directors and its family medicine chair and population health chair, and who is a family physician in practice for six years, spoke with Healthcare Innovation Editor-in-Chief Mark Hagland about the advances that he and his colleagues have been making, as they move forward in their leveraging of AI to improve both efficiency and physicians’ satisfaction with their clinical practice. Below are excerpts from that interview.
What was the origin of your and your colleagues’ decision to implement an AI solution in your group practice?
We’re always trying to do more agency coding, going from fee-for-service norms to more of a value-based way to code. We were struggling with trying to get doctors to code appropriately and get documentation from different hospitals and consultations. We were looking at different AI tools, and chose Navina, which was built by physicians for physicians, and is partnered with the AAFP Innovation Lab. It not only helps us to reduce pajama time [practicing physicians’ need to do documentation late into the evening in order to keep up]; it helps us by pulling all the data it can mine from the different hospitals, through the health information exchanges. It will read through the chart and give us a clinical summary for the patient, and will point out the last three things you saw the patient for. So it saves us time with pre-charting and post-charting. Before, it would take 20-30 minutes.
So you and your colleagues are actively pulling data from the EHR and the HIE [electronic health record and health information exchange]?
Yes. And the other aspect is that it helps our HCC [hierarchical condition category] coding by pulling codes from the cardiologist, for example. Let’s say that a patient had already been diagnosed with heart failure or had had a toe amputation, for example; it’s important to understand everything going on with a patient. Or, for example, a patient had a pulmonary nodule on an x-ray. It pulls that data together and presents it to the doctor in a fully organized way.
And when did you go live with the tool?
In September of last year.
Are all 70 physicians using the tool?
Right now our 20 primary care physicians and six nurse practitioners are using it.
Were there any challenges when you went live with the solution last September?
It doesn’t take long to train on it, but a couple of doctors were hesitant to start using it; they were concerned about accuracy. But we’re at a 97-percent user rate now among our primary care physicians. Every doc has shown they feel very comfortable with it. It pulls the data out and shows you where they got the data. And you can take that data and copy it forward into the chart.
Have you tried to measure any metrics yet?
Not yet. We just sent out a survey on this. But individual doctors are reporting that it’s decreased their administrative time. For me, it’s probably decreased over an hour of charting time every night. I would have been charting for three days. It has increased our HCC coding, and that’s good; and it’s made it more specific.
What have been the biggest lessons learned so far?
The biggest lesson is that we should have done it a lot sooner! Seriously, change takes time. And we continue to show docs that real-time data helps us all in our practices. And Navina has been there every step of the way for us, including with clinical quality measures and alerts to nurses, such as patients missing their mammograms, etc. Getting exams set up with scheduling.
Is the solution relatively easy to use?
Yes, it sits on top of our eClinicalWorks window. It really has made a big difference for us; we’re happy with it.
What should group practice leaders think about, per their implementation of solutions like this one?
This actually costs less than what ECW was charging for their population health module. And it’s more accurate. It’s decreased charting time and increased physician satisfaction with the EHR. And it’s improved our HCC coding by going through this.
And can you speak to the practicality of the approach?
When we were looking at AI, there are a lot of programs out there that were competitors to this, but they didn’t have the features we were looking for, or the ability to pay for itself. We could increase our HCC coding. And Navina has listened to us. It’s working on an AI scribe system and our clinical quality measures, and is already working on data-mining to find better cost approaches inside our ACO. It’s helping us to find those things.
