It’s all about data if you’re a health system seeking to get “closer to the dollar” by becoming a health plan too. And for New Mexico’s largest provider, that’s the good news. Albuquerque-based Presbyterian Healthcare Services is putting 30 years’ experience as an integrated health system to work helping other providers make the leap. Through its fully owned, for-profit subsidiary Fluent Health, it seeks to use its data muscle to partner with “like-minded providers across the nation” to help them integrate and thrive.
“Data on social determinants can help us focus interventions on the appropriate folks,” says Presbyterian’s Jason Mitchell, MD.
“We’re not looking just to provide third-party administration back-office services,” says Jason Mitchell, MD, Presbyterian’s senior VP and chief medical and “clinical transformation” officer. “We’ll actually help provide strategy, have skin in the game, and help providers become successful integrated systems.” Of course, Presbyterian’s “skin” means that if the game is victorious, it’ll be scooping up some of the winnings.
A 2017 Robert Wood Johnson Foundation report by research analyst Allan Baumgarten found that only four of the 37 provider-sponsored health plans launched since 2010 were profitable five years later. The plans studied were “only able to price competitively by paying their own providers below-market rates,” Baumgarten wrote. “That is not a strategy that can be sustained for long.”
Yet even though there are now plenty of other ways to share risk, starting a health plan remains a desirable goal for providers. As Geisinger Health Plan CFO and chief actuary Kurt W. Wrobel told the website revcycleintelligence.com a year ago, having the entire premium dollar to work with means “you can make investments. You can better coordinate the delivery of care with the administrative capabilities of the health system.”
That’s why Presbyterian invested $50 million over four years to “centralize and enhance” its data analytics system, creating “a large data warehouse.” The idea, says Mitchell, is to use data analytics to achieve a truly integrated view of the patient or member. To that end, clinical, claims, and pharmacy data are effectively merged. For some Presbyterian plan members, the mix also includes data on social determinants of health.
Data come into the Albuquerque warehouse not only from Presbyterian’s nine hospitals and roughly 1,000 employed physicians (users of an Epic EHR system), but also from reference labs all over New Mexico in “everything from direct interfaces to batch flows.” But don’t picture the warehouse as a single room; data specialists are physically scattered and often embedded in operations units, says Mitchell, because analytics needs to be “linked at the hip with the folks actually practicing medicine and running the business.”
Of course, traditionally staff-model plans such as Geisinger and Kaiser Permanente have been integrating care for years. But unlike their essentially closed systems, Presbyterian’s plan—a mix of Medicaid, Medicare, and the system’s comparatively low percentage of commercial business—isn’t conveniently coterminous with its provider organization. Some patients go to one of Presbyterian’s hospitals or employed physicians without belonging to its health plan, and some of the plan’s 620,000 members use Presbyterian’s coverage to patronize other providers. That’s one reason the new data warehouse must be “multi-tenant”—able to aggregate data across the nearly one million customer names on its “master patient index” while keeping firewalls between population segments for privacy and business purposes.
“We partition the data in very specific ways,” Mitchell explains, “to ensure that the delivery system never sees data on members who are not patients, and the health plan never sees data on patients who are not members.” Fluent, the for-profit subsidiary, stands ready to crunch data similarly for other providers across the country, helping them achieve the economies of scale necessary to take on risk for the full premium and still get to black ink.
Presbyterian can “resurface” claims data and bring them into Epic at the point of care, says Mitchell. For example, time and effort won’t be wasted reminding a patient with diabetes to get an eye exam if he or she has already received that service from another provider.
Mitchell takes special pride in an analytic tool called microsegmentation, which breaks down people in ways that are “not just disease-based or facility-based but based on how folks really use health care.” Clinical pharmacists can’t visit the homes of hundreds of thousands of plan members, for example, but such personal outreach may be useful once you’ve parsed pharmacy data to identify, say, the 700 people most notably “bouncing from medication to medication, and maybe having some ED visits as well,” he says.
As for social determinants of health, Mitchell says Presbyterian has loaded such data for some 100,000 members into its data warehouse. “Data on social determinants can help us focus interventions on the appropriate folks,” says Mitchell. “If you know members’ income, their general geolocation, whether they’re close to a pharmacy and whether they have transportation,” you can zero in on—and address—a very likely cause of their nonadherence to medicines.
Might the yellow brick road toward successful integration lead not to Silicon Valley or the populous Northeast, but to Albuquerque? Could be. “Sharing insurance data is a critical step, but it’s not the end-all, be-all,” says Andy Boyd, MD, associate professor of biomedical and health information sciences at the University of Illinois–Chicago. He hasn’t studied Presbyterian, but he believes partnership with a data powerhouse could be worth checking out. “If it’s structured right, it could be a win-win,” he says, adding that it’s only natural that a partner who shares the risk should also share the reward.
What just may be Presbyterian’s saving grace as an IT behemoth is something you don’t often get from data demons: a moment of humility.
“A lot of times our assumptions will be wrong and we’ll learn things from that,” says Mitchell. “So the goal is really to go into this with an open mind and let the data tell us what we need to do.”