Alejandro Reti, M.D., joined the Charlotte, N.C.-based Premier healthcare alliance early in 2013 as its senior director for population health. Since that time, he has been helping lead his colleagues at the health alliance forward to lay the foundation for a broader set of capabilities in integrated data analytics, as the healthcare industry moves forward on population health initiatives.
Last month, Premier announced that it was partnering with the University of North Carolina at Charlotte in a partnership to leverage UNC Charlotte’s academic expertise and community health technology, developed and led by Dr. James Studnicki, a UNC Charlotte College of Health and Human Services professor, to work with Premier, through its PremierConnect™ Enterprise business intelligence and enterprise-wide analytics platform, in collaborative initiative. Data connected to that initiative will be housed in PremierConnect Enterprise’s cloud-based data warehouse, which is vendor- and payer-agnostic, to enable timelier and more interactive data analysis and use.
With so much going on at Premier, Dr. Reti spoke with HCI Editor-in-Chief Mark Hagland shortly after the April 9 announcement to discuss Premier’s ongoing work in population health and analytics, and his perspectives on the future in those key areas. Below are excerpts from that interview.
Dr. Reti, can you give us a sense of the broader strategy behind the initiative that has been announced between Premier and UNC Charlotte?
Over the last 18-24 months, Premier has been quietly laying the groundwork for expanding our capabilities in integrated data analytics. One of our key initiatives is around population health management; but the other is to move the dial around that next tranche of cost and efficiency savings that healthcare organizations need to create to remain viable. So technically, architecturally, and from a process perspective, we’ve been looking at how to integrate data sets to make them available for data analytics. So we’ve been looking at meaningful community-based health analytics. So with UNC, we’re building on a foundation that Dr. Studnicki and his colleagues have been creating.
This is a solution, then?
It is. We’ve been helping our members perform community health needs assessments; that’s a consulting service we’ve been offering. But you end up building a report that ends up sitting on a shelf somewhere and doesn’t get updated. So by integrating key data sets from different sources, we’re helping providers develop a more integrated data stream around things like community well-being indices, crime statistics, etc., that are repeated regularly; and then tying that data with information on things like readmissions and mortality. And the point here is that these things all tie together; and if we can build a more complete picture, we can help them do a better job on what they should be focusing on, from a community intervention perspective.
Let me broaden the aperture a bit. One of the reasons we’ve decided to focus on this area is that our membership has been increasingly vocal that it’s very difficult to move the dial on your most difficult, expensive populations, if you don’t have a good community intervention strategy in place. If you look at the people—the “frequent flyers”—who cause a lot of readmissions and utilization, they typically are sick. But that is not necessarily the defining characteristic; there are mental illness and socioeconomic factors, and you need to address those. And this starts to bridge the gap between being a healthcare provider and a human services provider. So people are starting to ask, how do I address those gaps, and how do I use someone else’s money, to do those things?
So it becomes a public-private partnership, and a community thing. Let’s look at homelessness and lack of appropriate housing; Bon Secours for years has been working in Virginia with local housing authorities, to provide housing for people who need it most, and people with mental illness, and people who are in their ED 30, 50 times, a year, to help those people with those needs. And so our focus here is helping people identify where those priority issues are, and helping them start to make the right moves.
What are the biggest strategic opportunities and challenges that providers face as they try to reorient themselves away from being so inpatient-focused?
Because some of our largest members are non-profit systems, and many are Catholic health systems, they have always had a maybe above-average component of their mission devoted to caring for the population, so they have a little bit of involvement in caring for communities, and have some history in there, so it’s not a foreign idea to some of them. But historically, the reimbursement incentives have been elsewhere. I think organizations are best-positioned… Every organization has someone whose butt is on the line around global goals for care—connecting that person or those people to those who need to create community outreach—connecting those two. Because at the end of the day, it’s usually about a very specific subset of patients with a very specific subset of problems, and you have to improve those outcomes. I think that’s a bridge that’s often difficult to make. So we’re focused on that now.
What are the biggest challenges for those attempting to do analytics for population health right now?
You see charismatic leaders leading wonderful initiatives in isolated areas, and it’s kind of idiosyncratic what gets focused on. So the analytics has to be focused on what will give us the biggest impact, given limited dollars? And we really have a scarcity of good information on patient status—that means health assessments, well-being. And whether someone has good pain control; whether someone’s able to walk up and down a flight of stairs to go to the grocery store—whether someone’s able to engage in activities of daily life and have a good quality of life—we have great instruments out there to assess patient status in a variety of conditions, but we don’t have a good mechanism to collect that information cheaply and effectively to use it for analysis.
We’ve been working with the Dartmouth Institute, to do research around how, practically and cost-effectively, you can go around collecting these types of information. And one of the insights we got from some of our early pilots was that you don’t have a lot of face-time with patients to get that information; and a lot of them are not that responsive to long questioning—sometimes, some of these surveys involve 18 or more questions. So it gets hard to get them to engage. So the learning is around how to ask the right two or three questions to ask patients, and you need the right instruments.
Some patients are willing to have a smartphone that pings them and asks questions. But sometimes, you may need to call them on the phone, or call their daughter on the phone. So you need to get creative in that area. Also, ambulatory data integration is still an issue for some people. Their ambulatory EHR systems are on a number of platforms. And the third data integration challenge is around post-acute care. There’s a lot of activity there for these high-risk populations. It’s a bit of a no-man’s land right now.
And that includes behavioral health data, right?
Yes, and behavioral health is often its own thing, involving restrictions on getting data.
But per post-acute care data, generally, where are we?
The maturity in the post-acute area is still ten years behind where ambulatory data integration is, and that’s still behind. We’ve started to develop capabilities, but they’re generally at the c-plus level. Just barely a passing grade. It’s clumsy, generally expensive, not easily scalable, with unreliability of the data links. And with these CCD formats, you’re just not getting the data access you need. A lot of organizations are building systems based on the CCD format, but you’re generally not capturing everything you need.
What should CIOs and CMIOs know and do in all this?
I think number one a willingness to start exploring what they would be able to do with better access to the ambulatory, post-acute, and patient-reported data—because we’re going to start seeing more availability. And I think people at the CMIO and CIO levels need to have a plan for what they’re going to do with that data once they get it. And I think this general theme of incorporating data about your population that’s coming from outside your systems—this data about community well-being—some people are starting to bring in Experian data about their patients to help predict behavior for clinical reasons—it informs your clinical analysis around population health.
And if you fast-forward 10-15 years, you’re essentially competing for covered lives; it’s not service line by service line anymore, but for the loyalty of patients over a lifetime. So you have to figure out how to develop a relationship with them. And I think there are going to be blurred boundaries, right? I think people will be comfortable with you using that data for caregiving purposes, but not marketing purposes. So the lines are getting totally blurred. I think about patient privacy a lot; but I would hate to see us get restricted to access to critical information, because people would be afraid that data would be misused.
What do you see happening over the next few years?
I honestly see a period of retrenchment and learning. If you look at a lot of people’s experiences with population health, they’re often the results of chance. Some people do well, some not. And if people are being brutally honest with you, they’re recognizing that they know a lot less than they think they do. And I think we’ll see, especially people who have lost money or who haven’t performed as well as expected, will do a better job of looking at what has worked and not; so I think program evaluation, and taking this to the next level, will be a big focus.
