Q&A: Epic President Carl Dvorak
Epic President Carl Dvorak wrote much of the original code in the system and made some of the enduring architectural decisions empowering its growth. With the company since 1987, he is often referred to as the technology brains of the organization.
Healthcare IT News had an opportunity to speak with Dvorak about Epic's approach to population health. Here's what he had to say.
Epic's DNA is population health management-centric
"We are making significant investments in research and development around the broader issues relating to population health. Our history with population health goes way, way back to our very earliest electronic health records systems. In the early days of electronic health records, managed care organizations that jumped in first and because of that the core Epic system have just an absolute wealth of population health management features built in. Think of it as our healthcare operating system.
[See also: Epic defends interoperability bona fides]
"We have a rich history of clinical quality reporting, physician score-carding variation analysis to help understand the opportunities within the health systems to better care for patients. This empowers more consistency and less fragmentation, reduces variability and expenses, and enhances outcomes."
Patient engagement is a key to population health management
"We want the patient to be the person who cares about an issue or risk with their health. We want patients to be engaged and do the very best that they can do to stay healthy or avoid disease or a progression to a chronic condition.
"In about the last five years, lots of effort has gone into self-service capabilities, so doctors can engage with the patient more fully in their care. That's been a big driver for us in the population space.
"Some of the newer work we are doing includes bringing in the social determinants of care at scale for population health management: Understanding if the patient lives in a food desert. Do they have transportation issues? Is there concern with regards to the home situation? Is there depression involved? Is there loneliness? Surprisingly enough, there are a lot of healthcare expenses driven by loneliness."
Global footprint moves best practices across national boundaries
"Epic sites are quite sophisticated. And thankfully many of them have their mission to share. A lot of what our community does is shared freely with each other.
"We bring lessons learned from Europe to the U.S. and from the U.S. to Europe and Asia and other parts of the world with active implementations. In Finland, one of the big parts of our project there is integrating many elements of the social care fabric directly into our health record. So everything from child health and welfare up into school lunch programs -- all kinds of information -- will be factored into the patient's health care because in that country, it's socially acceptable. People feel it's a social responsibility to bring that information together in the best interest of funding healthcare for the population in a holistic manner."
Predictive analytics is the key to evidence-based medicine
"Over the last three or four years, we have made a tremendous amount of investment in predictive analytics. We can tell a patient is crashing in ICU before a person could notice it. We can tell a diabetic is likely to go out of control in a hospital around the holidays before they show up in a hospital. So the predictive analytics capabilities are another key investment area.
"When I think of population health, I think of algorithms. This is the next frontier. It is about the algorithm models in particular. I think that's the new currency. It's not an app. Apps have their utility. We ran the app experiment."
R is proposed as the universal predictive modeling interface
"As I project that into the next era in caring for populations, there are three or four things that I think are going to be extraordinarily relevant.
"At the foundational level, we use and expose R, a sort of the XML expression set that forms a programing language for statistical computing. That's the model that we use for the transportation of predictive models. That's the unit of delivery, if you will. When I talk about a predictive model, we're building them in such a way that it can be transported across Epic and non-Epic sites that adhere to that standard."
Epic holds that best-of-breed environments are not efficient
"In the '80s and '90s, people thought that best of breed was the thing, and it turned out it actually hurt people. Applications developed by different people, at different times, with different paradigms -- things didn't line up properly, and patients actually suffered harm because of it. What has really made a difference today in healthcare is integration. Building a comprehensive tool that clinicians of all sorts can use to collaborate on patient care has actually made a significant difference in outcomes as well as reducing expense."
Changing clinical behavior is essential to population management
"Every human being who comes in contact with the patient represents an opportunity to do the right thing. We want to make sure that the electronic health system -- we think of it as an operating system for health care -- is wired in such a way that every single opportunity can be maximized.
"When patients show up for a routine blood draw, and we find out they're overdue for a mammogram or a colonoscopy, and there's availability down the hall in 20 minutes, I want that front-desk person to take appropriate action."
"We need a support system for the algorithmic workers and the caseworkers to sift, sort, slice, dice, and understand their population, and for case managers and proactive outreach in order to have a work queue, to have a dashboard, to have a patient list to help do the outreach to help motivate people."
Closing gaps across the continuum of care is essential
"We also work to help create a population health management infrastructure that transcends any one site and vendors as well. So we have customers today who use our infrastructure and some of the new work we have done to bring in data from other practices -- eClinicalWorks, Athena Health, Greenway as examples -- and to bring in data from payers even for events that took place outside their health system and work with an Epic platform to manage healthcare and population health clinically integrated networks when some of the participants have heterogeneous technology. That is one of the new things we have done. And it is going really well. It's real successful. Places like ProHealth Care in Milwaukee and Cadence Health, now part of Northwestern Medicine, in Chicago are good examples that makes use of that."
Virtual care is core to population health management
"We also see as core to population health a recognition that we had to create vectors for more virtual care. E-visits, televisits -- all that is now built in the fabric of Epic. It is not an extra add-on, extra charge. We built that in recognizing that population care is going to drive people to do the right thing even if there are not service codes associated with it.
"Epic users are early adopters. Prior to Apple HealthKit integrations, we already saw integrations that brought Fitbit and cloud-to-cloud links using that data, and it is a little controversial. You know, I'm a doctor. What do you want me to do with all that data? We found there are important use cases. For example:
"If you have a new hip replacement and you put a Fitbit on a patient, you can actually tell if they are getting up and moving around the first week they are home or if they are moving around twice as much the next week they're home or three or four times the third week. Are they getting up and getting going? If yeah, then give them a congratulations email. If not, then send a nurse to the home; you may have a problem brewing. Even if you can argue [the devices'] accuracy can be valid, you know in some ways.
"The blood pressure cuffs and the scale -- all those things are actually in production at many of our sites.
"(We're) doing a lot of work with continuous glucose monitoring. Turns out diabetics do in fact study for the test. But if you put [in place] continuous monitoring, you can actually see that time and range is not satisfactory and is leading to the vascular damage or something."
Epic shares IP with its user health systems
"We are happy to share IP with our customers. This is part of the abstract clinical content algorithm approach. We have done that with a number of content pieces. We are working actively with Mayo Clinic to offer some of their clinical content as a paid-for subscription. Part of that money goes back to Epic to support having us manage that, having imports, updaters, all that kind of stuff. However, we leave the marketing to them. The app store will allow them to monetize solutions."