Will accountable care organizations work? The University of Michigan Health System has a pretty good idea. Four years into its participation in Medicare's Physician Group Practice Pay-for-Performance demonstration project, staff shaved more than $15 million in Medicare costs while improving quality. Its faculty group practice is recognized as an ACO prototype and is one of only two physician groups in the demo to save money in each year of the project while surpassing 30 of the 32 quality measures tracked. Much of the health system's success can be traced to improved ambulatory care coordination for geriatric patients with complicated conditions, and investments to improve transitional care. Caroline Blaum, M.D., professor of internal medicine and geriatrics at the VA Ann Arbor Healthcare System, part of the U-M Health System, spoke with Bob Kehoe, H&HN contributing editor, about how some of these results were achieved.
WHAT ARE SOME HIGHLIGHTS FROM THE PROJECT'S FIRST FOUR YEARS?
The Medicare Physician Group Practice demonstration project covers 32 quality measures in ambulatory care, including the chronic management of diabetes, coronary artery disease, heart failure, hypertension and preventive services. The demo started in 2005, which was a while ago in the world of quality. We already were reporting some quality measures to other payers, so we had some mechanisms in place. For others, we put mechanisms in place.
Over the years of the demo, we improved our performance in all 32 quality measures—meaning we either hit our target or we hit 96 percent of the target.
WHAT KEY CHANGES WERE MADE ON THE CARE COORDINATION SIDE?
We invested in transitional care, which was seen as innovative in 2005 and now everyone is talking about it, and we invested in complex care management for patients who were going in and out of the hospital due to complicated conditions.
One component of our transitional care was a callback program. After patients were discharged from the hospital or emergency department, they received calls from a nurse, care coordinator or care coordinator assistant who would try to figure out if they were experiencing any problems during the transition. We called patients within 24 hours of discharge, except on weekends, in which case we called them on Monday.
The program was somewhat unique because the physician manager and care coordinators had almost a direct line to home care services, to a doctor or pharmacist if they needed help, and they could get clinic appointments rapidly.
HOW IMPORTANT WERE THESE ACTIONS IN RELATION TO THE $15 MILLION YOU SAVED?
We know that some of our success was because we reduced readmission rates. And even though it was small, we had decreased readmissions for some of our major high-cost patients, such as those with heart failure and frail elders.
WHAT TAKEAWAYS HAVE THERE BEEN FOR PHYSICIANS IN THE PROGRAM?
Primary care doctors learned that care coordination and teamwork are important. We've increased the number of referrals from physicians to care coordinators. We've increased communication between our hospitalists and our primary care doctors.
HOW HAVE NURSES, PHARMACISTS AND OTHERS IMPROVED QUALITY?
Nurses are the backbone of our care coordination program. We have a centralized approach to complex care coordination for patients who have the highest utilization and are the most difficult to treat. We also have expanded care coordination activities of our clinic nurses. Many nurses have not had phone management of patients as part of their traditional role. We're beginning to provide training so nurses can be comfortable interacting with patients between visits over the phone, or sometimes have a nurse visit to ensure chronic cases are managed better. We envision a very important role for nursing and we've tried to increase that role.
In our geriatrics clinic we have a consultant pharmacist who reviews medications of some discharged patients who come to the transitional care clinic. Some pharmacists also are involved in care management of patients with chronic diseases. For example, we're able to offer pharmacist consultation to self-managed diabetes patients who may be having trouble keeping their disease under control. We're not catching everyone, but our goal is to consult with patients who need it the most.
WHAT WERE THE CHALLENGES TO ATTAINING THE 32 MEDICARE QUALITY MEASURES?
We have two big challenges. One issue, as we expand the quality programs to other chronic diseases or conditions, is to develop the patient registry, develop feedback mechanisms and educate the physicians. With the quality programs that have been in place for many years, one problem could be complacency.
WAS THERE AN 'A-HA' MOMENT ABOUT WHAT COULD BE ACCOMPLISHED?
The "a-ha" moment for us was when we put our information systems together so that our primary care clinics would know which of their patients were in the hospital, so they could then get them back to the clinic or manage them as soon as they were discharged. That turned out to be hard to do, even in our integrated system. Our challenge is to have appropriate communication with the entire team—the patient, family and all the physicians involved—and put that in place every time. We still don't get it in place for every patient, but we're doing it for more and more patients, and when it works, it works really well.
WHY HAS THIS PROJECT BEEN MORE SUCCESSFUL THAN OTHER MEDICARE QUALITY IMPROVEMENT PROJECTS?
One big factor was that we had experience in managed care. Many of the processes used in this program were put in place for managed care. We already had started our chronic disease registries on the ambulatory side and were reporting diabetes clinical quality measures. We had some experience with complex care coordination through our managed care experience, and we already had this idea of population-based management. We had a group of patients for whom we were trying to manage their health.
Second, we had an electronic health record that exists in our outpatient clinics and hospitals and is available in our home care agency. We take care of 45 percent of our discharges in skilled nursing homes and we have put our EHRs in those nursing homes.
The other thing that helped was investing in quality improvement efforts to develop the registries, provide feedback to physicians, issue point-of-service reminders about what care has to be delivered and identify gaps of care.
Finally, our leadership invested in transitional care and complex care coordination infrastructure. Without that, there's no way to pay for a nurse care coordinator to follow up with discharged patients. There's no way to bill for that. I don't know what would have happened if we hadn't been successful with this program. The shared savings covered the cost of the program.
BASED ON YOUR GROUP'S EXPERIENCE, HOW DO YOU FEEL OTHERS WILL FARE AS THEY STRIVE TO BECOME ACOS?
A fully integrated system like that of the University of Michigan might be the most likely to achieve ACO status if it's big enough because all our doctors are employed, we all have the same electronic health record, we have a certain number of primary care doctors and subset specialties and our own hospitals. With that said, in the PGP demo, the Marshfield (Wis.) Clinic was the most successful and they're not structured the same way as U-M. That's a physician group that works with several different hospital systems. So there are different routes and models for success.
WHAT CAN OTHER PROVIDERS DO TO PREPARE FOR PAY FOR PERFORMANCE?
It's all about relationships, coordination and teams. Getting away from HIT, which is another discussion, we have to figure out how primary care physicians talk to specialists, what the referral mechanisms are, how they work and, most importantly, how primary care physicians are able to communicate with hospital-based physicians. Some of that is physician-to-physician, but a lot of it is a team concept. We have to make sure that nurses, pharmacists, social workers and other professionals are part of the team. Restructuring care delivery relationships is important. People want to do it, but it's a challenge.