In the Trenches on Population Health: Medical Group Leaders Move Forward to Move Data (and Process) Mountains

Oct. 5, 2016
Pioneering medical group leaders are moving forward to pursue a wide variety of population health strategies, and are pushing ahead on all the hard, complex work of population health advancement

Amid all the broad, policy-oriented discussions, and the very broadly strategic discussions, about population health management, these days, what has it actually been like to execute on the population health promise, at the large multidisciplinary medical group level? In a word, challenging. Medical group leaders are clear on the fact that they’re trying to do something that essentially has never been attempted before in the U.S. healthcare system, and that is to transform the processes of care delivery.

What are the common denominators? In interviews with Healthcare Informatics, the leaders of pioneering medical groups, while pursuing a wide variety of strategies, are finding common challenges and opportunities in wading into the deep end of the pool in several key, overlapping areas. Among them:

  • Medical group leaders are beginning to effectively harness clinical information systems (some of them anchored in electronic health records, some of them systems being connected to EHRs) and data analytics, to perform health risk stratification across broad populations under defined contracts.
  • Medical group leaders are also moving ahead to put in place care management systems to support their chronically ill patients and enhance their health status—e.g., optimizing blood sugar control for diabetics, etc.
  • Closely allied to those first two elements, group leaders are moving forward to participate in what is often referred to as the “blessed cycle” of clinical transformation and performance improvement, meaning, a cycle of data collection, data analysis, data reporting, data sharing with clinicians and staff, data-facilitated performance improvement around care delivery processes, and then further cycles of data collection, analysis, reporting, and sharing, to support continuous performance improvement.
  • Embedded within all of these efforts are specific IT- and data-related efforts, including the creation of chronic disease patient registries, the IT facilitation of care management processes, the building out of data warehouses, the creation of dashboards for physicians and other clinicians, the facilitation of both clinicians and patient engagement via mobility and mobile devices, and the facilitation of data analytics combining claims-based and clinical (via the EHR) data, among other essential capabilities.
  • Leaders at pioneering medical groups are currently busy addressing the welter of strategic, operational, clinical care process, data analytics, IT-technological, and other issues involved, and are laying the foundations for successful initiatives that will be replicable across the U.S. healthcare system.

Among the challenges involved is a very fundamental one, says Bob Schwyn, a director at the Chicago-based Chartis Group consulting firm, and a former healthcare CIO. “Our experience across our client base,” he says, is that getting clarity on what population health is and what it means, is very important to understanding your market and where it’s going, and what populations you’re focused on, and all the considerations around the value proposition and focus. In many cases, when our clients seek us out for technology assistance, they haven’t yet created enough alignment around how the technology will support the business, and often, there’s also a lack of clarity around the broader strategic plan for the organization.”

Schwyn’s colleague at Chartis, Mark Werner, M.D., the firm’s director and national leader for clinical consulting, adds that “One of the things we’re learning is that it’s a phrase with a lot of meanings,” he says, speaking of population health. “Part of what I think is happening in the trenches is that people are gradually realizing that it’s not just an IT initiative or a primary care medical home initiative, or an isolated-contract initiative, but rather that it really does require an enterprise-level effort to link to your strategic plan. Part of the problem is that there remains confusion about population health at the public health or community health level, since we’re trying to achieve some public or community health goals via what is still an acute-care-based health system. So you have to begin to stratify populations and realize you’re already taking care of multiple populations.”

Below are three case studies that illustrate the challenges and opportunities involved, and the diverse approaches that physician group leaders are taking as they move forward to fully leverage IT and data analytics to facilitate population health management. Each illustrates different facets of the landscape.

A full-court press in New York state

At the Middletown, NY.-based Crystal Run Healthcare, a multispecialty group practice with 35 locations and 375 providers (300 of whom are physicians), all of the senior leaders are absolutely committed to a population health strategy as their organization’s core organizing strategy going forward.

Prefacing his comments modestly, Gregory Spencer, M.D., Crystal Run’s CMIO, says, “I don’t know that any of our learnings have been terribly profound.” Instead, he says, the basics are fairly clear: “Getting the data as right as you can at first is important; getting all the stakeholders in early to participate in the overall process, is really important. And having an iterative workflow so that people can see that there’s an end to the means, showing that you’re picking things that make a difference, so there are tangible real, a sense of purpose and gravity; those are the most important things.”

In fact, notes Scott Hines, M.D., Crystal Run’s chief quality officer, “When we first really started leveraging data to improve outcomes, our first steps were looking at creating registries to identify patients who had gaps in care, and then distributing those registries to every doctor and their nurse, to try to close those gaps in care. But we realized over time that primary care doctors in particular were becoming overwhelmed by tasks, and that that approach wasn’t the best use of their time. So we took a step back and what we could take off their plates.”

Scott Hines, M.D.

Three important initiatives have come out of that rethink, Hines reports. First, he and the other senior leaders in the medical group created a Care Optimization Team, led by a nurse and staffed by four non-clinician staffers. Each member of that team is assigned to one or more patient-centered medical homes, and it is those individuals who reach out to patients who it is discovered have gaps in their care for process measures such as immunizations, breast cancer screenings, colon cancer screenings, necessary labs, and so on. “We leverage data from the EHR and reports from payers, to help us to identify those patients and reach out to them as soon as possible,” he says.

The second initiative coming out of the rethinking process is Crystal Run’s Payer Quality Scorecard, developed in 2014. “That mechanism allows us to track internally what we’re doing in terms of quality performance for measures for each payer we have a risk-based contract wit. Prior to that, we had been using registries, but relying on the primary doctors and their staffs to reach out and close gaps in care, but since then, we’ve built this team.; so doctors can concentrate on performance measures like blood pressure control, that kind of thing. So that’s one way we’re using technology to improve the quality of care.”

And the third initiative has been the group’s Variation Reduction Program, which has led to reducing variation in medical practice across specialists working in the same specialty. For example, the Crystal Run physicians have examined their endocrinologists’ annual total charges around cancer care, including professional, lab, imaging, and procedure charges. As he and his colleagues, including Jonathan Nasser, M.D., Crystal Run’s chief clinical transformation officer, have found, variations in spending have zero correlation to clinical quality outcomes. Hines notes, “Jon and I meet quarterly with each division to perform on a guidelines adherence exercise. And ahead of time, we ask the division which diagnosis they want to tackle, and then we assign one or two physicians in that division to research what guidelines or evidence exist in the literature, and so they come to that meeting armed.”

Jonathan Nasser, M.D.

Specifically in the case of physician orders around cancer care, Hines says, “We’ll ask a question such as, how often do you do ultrasounds or tumor markers for patients with thyroid cancer? And we’ll always have a lot of variation in the frequency of what the physicians do. But it turns out that the American Thyroid Association has recommendations in that area. So in that case, we were able to incorporate their guideline into a guideline we’ve since developed here through consensus.” Based on that work, the Crystal Run physicians have reduced the number of visits per patient, based on appropriate adherence to consensus- and evidence-based guidelines, which then ends up allowing more patients to be seen, and helps balance out care delivery, with costs appropriately rising slightly among doctors who had been underutilizing and falling among those who had been overutilizing, but costs overall going down.

One of the challenges inherent in population health management work at the medical group level, Nasser notes, is that “A lot of utilization takes place outside our setting of course, such as through inpatient hospitalizations and ER visits, and care in nursing homes. We try to accomplish as much as possible for the patients who are in front of us. And we arm physicians and teams with data, and ask for suggestions for improvement and we try things out through PDCA cycles or meetings; and then when patients aren’t in front of us, we’re also involved in their care through care optimization, and we also utilize care managers who help to facilitate care for our sickest patients, and looking at telephonic outreach etc.” Events that take place outside the practice will continue to pose a challenge, Nasser says, “But leveraging data through scorecard reporting, team analysis and individual reports, helps us here internally, and we come together to determine things we need to improve and how we go about doing them. But that’s the structure of how we take a look at the population health work.

In fact, the very first Variation Reduction Program effort, Hines notes, was around diabetes.  Through a process of researching the literature, analyzing data, and convening meetings with the primary care physicians in the group, Crystal Run leaders were able to see clearly that there was a “three-to-four-fold variation in terms of cost,” among PCPs caring for diabetics. “And really, cost is just a surrogate for utilization,” Hines notes. In the event, it turned out that some PCPs were seeing their well-controlled diabetes every three months, some every four months, and some once a year, while they discovered that the American Diabetes Association had recommended an interval of every six months. Through data analysis, sharing, and discussion, a consensus was reached among the group’s primary care physicians to settle on a protocol of seeing their well-controlled diabetics once every six months. As a result, they realized a 9-percent reduction in charges per patient per year within six months after developing that initiative, compared to the level of charges per patient per year in the six months prior.

“Sometimes,” Hines notes, “you just get locked into your pattern of practice and you ask people why they do things a certain way, and it’s just how they were taught to do it, or how they did it. So this process forces you to look at the evidence and update what’s going on.”

Pennsylvanians focus on PCMH-building

For leaders at Lancaster General Health in Lancaster, Pa., which in August became part of the Philadelphia-based Penn Medicine system, two main areas of focus have been Lancaster General Health’s participation as an accountable care organization (ACO) in the Medicare Shared Savings Program (MSSP) for ACOs, and its universalization of the patient-centered medical home (PCMH) model across all of its medical clinic sites. Among those helping to lead the charge in Lancaster are Douglas Gohn, M.D., physician executive for population health at Lancaster General Health, and Michael Ripchinski, M.D., Lancaster General’s chief quality and medical information officer. With regard to ACO development, LGH is managing the care of 18,000 in the MSSP program and 70,000 in some sort of risk-based contract. LGH is also participating in the Bundled Payment Pilot Initiative out of the Centers for Medicare & Medicaid Services (CMS), doing cardiac stents, bypass surgery, pacemakers, hip and knee joint replacements, and some spine procedures as well, Gohn reports.

Douglas Gohn, M.D.

Overall, Gohn notes, “The starting block” for population health-based work has been “the patient-centered medical home. All 28 of our primary care practices are Level 3 PCMHs. We’re trying to upgrade to the 2014 standards for PCMHs. There are some changes, not real substantive ones, but that require tweaks. So the PCMH and team-based care are in my mind the foundation for population health. Then you need to connect the physician IT infrastructure to all of that. A lot of the challenge for us,” he adds, “has been a data challenge. The goal here is to aggregate claims and label data from disparate sources and put that into an analytical tool and derive appropriate risk. Claims, EHR, and some form of social determinant, all need to be added to that, and now we’re also beginning to look on patient-provided data, such as from wearables and implanted devices, though we haven’t even done that yet.” Importantly, he adds, “You need a care management platform that that sits on.”

As their colleagues at virtually every other medical organization are finding, Ripchinski reports that “It’s very difficult to manage both claims data and clinical data. There are companies trying to merge clinical and claims data to create a path forward, but we’re early on” in that journey, he notes. “And as part of setting up the MSSP in January 2014, we started to get claims data and began to do typical payer analysis—what’s the pharmacy spend, who are the high ED users, who are the chronic condition patients?”

Michael Ripchinski, M.D.

Importantly, Ripchinski notes, “Aligned with our claims analysis work, we’ve also done risk stratification of patient populations using the clinical data in the EHR. And we’ve timed these so one method of analytics can use the other.” For example, he notes, “In the EHR, we will find out a particular patient is in seven different disease registries, they’ve had claims for eight different conditions, they’ve been in the EHR five times.” So one of the key strategies involved, he reports, is marrying EHR-based and claims-based data “o look at managed lives by how many patients have had multiple hospitalizations, or high ED visits, in a two-year period.”

Texans confront the data challenges

At Texas Health Physicians Group (THPG), the physician organization integrated into the Arlington, Texas-based, 21-hospital Texas Health Resources, Shawn Parsley, D.O., president of THPG, and Barbara Adams, vice president, innovative technology solutions, for THPG, and Texas Health Resources. THPG is participating both in the MSSP program, in concert with the UT-Southwestern Medical School in Dallas, and in several commercial ACOs.

Shawn Parsley, D.O.

“Clearly,” says Parsley, “having analytics makes up a major portion of your ability to actively do anything with these contracts. There’s a period of time where we’re standing with a proverbial foot in the boat and foot in the dock, in terms of the fee-for-service world and the fee-for-value world.” What Parsley, Adams and their colleagues have tackled first is “a quality incentive program for the docs that was really payer-agnostic, and designed to represent the entire panel a physician had, with incentives. And we really want our physicians to think about how they take care of all their patients. So the first step was to develop a quality analytics dashboard that would have the capacity to look into the EMR data, without regard to which EMR is involved, and extract the information and compile it in a centralized database.” At least 50 percent of THPG physicians are actively and robustly using the dashboards that THPG is providing them, to improve their management of ACO patients with chronic illnesses, Parsley notes.

Asked what some of the main data analytics challenges have been, Adams says, “Guess how many places a doctor can document tobacco cessation? About five different places; but there is only one place in the EMR where it will give them credit, because [documenting patient tobacco cessation in that place] is mapped to the dashboard. And it’s not the doctors’ fault. You get into a groove with your EMR. And we can’t map five different places. But variation in EMR documentation leads to a complicated mapping processes,” she notes. Other challenges include the need to customize EMRs on a variety of different servers, and data validation, she adds.

Barbara Adams

Making it work for the physicians

It is very important in all this work, everyone agrees, to recognize how profoundly this important, innovative work affects physician workflow, productivity, and practice. Dean Field, M.D., vice president for informatics and operations at the nine-hospital, Tacoma, Wash.-based CHI Franciscan Health, says, “Among the biggest issues at the medical group level, in tackling population health, is around the added responsibility. There is a shrinking population, or at least stagnant, of primary care physicians, and yet population health management requires us to manage more in terms of what the patient came in with. So we’re asking primary care physicians to do more; so that’s one challenge. The second challenge is how we begin to capture the information critical for population health management, in a structured format. Most EHRs were implemented with adoption in mind, but allowing people to do free text.” But it will take considerable application of natural language processing in order to extract key data elements need for pop health, from electronic health records, he notes.

In the end, all those interviewed agree, making serious advances in population health at the medical group management level will inevitably involve years of foundational work in the coming many months. But these case studies demonstrate that the leaders of pioneering medical groups are indeed laying the foundations for successful pop health practices U.S. healthcare-system wide.