A carefully managed electronic data warehouse is a key component of improved patient care in any health-care practiceand is surprisingly easy to build on and improve, once you begin the implementation process. Those were the key points of “Use of Registry Tools to Improve Care,” presented by David Howes, MD, in San Diego, California, on October 7, 2013, at the annual conference of the Medical Group Management Association. Howes is president and CEO of Martin's Point® HealthCare (Portland, Maine), a nonprofit health-care organization. Howes explains that Martin’s Point’s data warehouse, initiated in 2000, is still a work in progress. He says, however, that it has already produced dramatic improvements in patient outreach and care. It has also made treatments and procedures more uniform, has allowed physicians to allocate their time more effectively, and has given Martin’s Point more leverage when negotiating with payors.

“Good management of quality of care leads to good management of cost of care. In 2000, before we had electronic health records (EHRs), we built a data-warehousing capability and began a data-management plan. We built forms for the major chronic diseases, collected granular data, and made them available to our data warehouse.”

—David Howes, MD, CEO, Martin’s Point HealthCare

Provider–Payor Perspective In addition to providing primary and specialty care in nine community health centers in southern Maine and New Hampshire, Martin’s Point insures military retirees, their family members, and active-duty family members (through a TRICARE plan) and Medicare patients through three Medicare Advantage plans called Generations Advantage. From the outset, the data were provided mainly so that clinicians could see the histories of patients with chronic diseases, including the progress of their treatment, their comorbidities and past conditions, and how their situations compare with those of other patients with the same disorders. Ideally, the data in a registry should address three questions, Howes says: What’s going on with this patient? What’s going on with other patients like this one? What’s going on with all the patients in my panel? “For data to be helpful, you must have the right data, at the right time, in the right hands,” Howes continues, “and that’s often not how data will arrive. It’s critical that the data be timely. If you use claims data as the core system for providing care data to your physicians, there will be a three-month lag—built into the claims process—and the data will be full of holes. People might not bill accurately; they might not bill for items that they think they won’t be paid for anyway. There’s nothing more frustrating than disorganized, three–month-old data. We had to produce our own, using our own EHRs and our own warehouse.” Population-health Nurses The ideal system, Howes says, uses transparent definitions and criteria. It needs to have a patient view of gaps and successes, a panel view of performance, and a trending view over time. The data should be timely, accurate, organized, actionable, comparative, and accessible—and they should make it possible to identify patients not meeting practice standards. “When we’re talking with health plans about improving quality of care and standards, we have to be able to show, organizationwide, how performance stacks up against their standards,” he notes. “[Our comprehensive registry] is now woven into the fabric of the organization, as we move from one health record to another. At the macro level, we use it to put together lists of particular gaps in care.” Howes notes that there are two reasons that patients come to a Martin’s Point facility: to take care of their health or because they’re suffering in some way—the dominant reason, by far. Recognizing that physicians are so busy that reviewing lists of patients who have gaps in care is a low priority (and hard to do), Martin’s Point has put in place population-health nurses, paid for by its health plan. The role of the population-health nurse is to identify patients with gaps in care (mapped by the day, week, and month); reach out to those patients to schedule visits and laboratory work; and tag them for later research. For example, Martin’s Point has, in its health plans, about 28,000 people over the age of 65; at least 10% of those have congestive heart failure and a gap in care. In addition to supporting the management of quality (and therefore costs) in its own insured population, Martin’s Point’s data warehouse has proved valuable in identifying patients with chronic illnesses and comorbidities for the purpose of developing expectations during negotiations with payors. One Datum at a Time Howes offers the example of a small practice that treated 186 patients with diabetes, 210 with hypertension, and 97 with coronary-artery disease. “We’d look at the overlap, which is striking,” he says. “We start with the overlaps and try to get those people in; the diabetic/hypertensive overlap is particularly lethal. These data give the clinician a clear picture of what the population looks like and who should be seen first—who’s at higher risk.” He adds, “Your leadership culture needs to support transparency in reporting. We found that once you have a little transparency, it’s easy to get a lot. You need a culture that’s data oriented, but my advice is this: You benefit most from adding one piece of data at a time and letting people figure out how to manage that addition. We sometimes [let the data get] ahead of the performance-improvement work, and people got frustrated, but we have a robust data-reading program, and we’re developing adaptability and agility. We try to improve one or two things at a time.” Treatment of hypertension shows the strongest improvement at Martin’s Point, Howes reports, but he adds that diabetes remains a challenge, since so many patients with diabetes have big treatment gaps. Remedying that situation, he says, is an organizational priority. At first, the organization was concerned about the expense of clinical improvements, since it wasn’t getting paid for them, but it soon became clear that patients were being treated more frequently and efficiently as a result of improved data management. “We’ve stopped marketing our practice,” Howes says, “and we focus on improving our registry numbers. We’re getting more visits from the same number of patients—getting more work done that needs to be done, in a way that patients appreciate.” He continues, “We’ve gotten much better at getting people in for preventive care. It’s not perfect, and there’s a long period of years where you don’t need to see people often, but a surprising number of those people have significant health risks. We see the people who come to us, as opposed to those who need to come to us. In Maine, there are nurse practitioners who can work independently; they do a lot of the chronic-disease care (and do it very well). This leaves the physician to deal with the sicker, more complex people who have more need for a lot of judgment-based care.” Sometimes, the nurse practitioners resist this system, insisting on being treated with more equality, “and that’s a difficult conversation,” Howes says. Just Do It In summary, Howes advises, “I don’t care where you start; just start somewhere. Keep working on process improvement—keep sharing it; don’t worry about getting it perfect the first time. If you look, for example, at best practices in hypertension, compared with what your group is doing, you’ll find your group is still flying by the seat of the pants. There’s a lot of inconsistency in practice. We begin to improve clinical practice by coming to agreement, within our group, on what the sequence of treatment is going to be.” When you have people naming the drug sequence that they will use in the group (instead of a little of this and a little of that), you’re making real progress, Howe says. “Then, go back and take a look: Are you following results correctly, doing things the same way, and dealing with exceptions as exceptions? If you look at the literature, you’ll see differentiated quality of care when the practice has decided what it’s going to be,” he notes. “Physicians have been unwilling to accept that processes need to be standardized. Standard work is a bad term, but getting clinicians to do standard work will unburden them from routine matters and let them work on stuff that demands their energy.” Martin’s Point is still using the same SQL database that it started with in 2000, updating to a new generation from time to time. New forms continue to be constructed, with new operational terms defined each time. Data Dividends “We’ve captured essentially all the critical clinical data, from immunizations for kids to preventive care for adults,” Howes says. “We hired someone to put the system together in 2000, and he has been the soul of continuity. He knows the routine. Our data warehouse is really consistent, over time. We started with a unique member-identifier number—thank goodness we did that—and patient 00001 is still a patient.” The data that Martin’s Point collects (on both its patient population and its insured population) have been pivotal in its rate negotiations with payors. “We own our own health plan, so we have people we insure ourselves; we have overlap of the EHR data and the health-plan data,” Howes says. “We have contacts with other risk providers in the state, and we can put our numbers on the table; we can show them if their numbers are wrong. If we have the data, the negotiation process changes. It makes them mad, but our conversations are very different from what they would be if we didn’t have this tool.”Joseph Dobrian is a contributing writer for HealthCXO.com.