As patient-centered medical homes have become more widespread, gaps have emerged between their popularity and research findings. While some studies tout positive results, others have found little to celebrate. Yet even when a pilot project doesn’t meet its goals, it can teach health plans a lot about how to structure medical home incentives and pay for primary care, says Mark W. Friedberg, MD, MPP, a leading researcher studying the field. He is a senior natural scientist at Rand, a professor at Pardee Rand Graduate School, and a clinical instructor of medicine at Brigham and Women’s Hospital and Harvard Medical School.
Friedberg led a study of the southeastern region of the Pennsylvania Chronic Care Initiative, one of the largest and longest-running early multipayer medical home pilots, and he participated in the study on the Rhode Island Chronic Care Sustainability Initiative Pilot Program. His teams found that these medical home interventions had limited effects on quality and no impact on health care costs or utilization. The results disappointed many, but there’s a lot to learn from them, he says.
People assume everything is the same if it is medical home, but that misses some very interesting and valuable variation in the design. What people mean by a medical home is whatever the practice and the payer agree a medical home means.
“As more pilots produce results, we can see whether the differences in their ingredients are producing the different effects that we want to see.”
Friedberg is a member of the Patient-Centered Medical Home Evaluators’ Collaborative, a work group convened by the Commonwealth Fund, and on a variety of expert panels for industry organizations. He has written more than two dozen journal articles and is a reviewer for publications that include the Journal of the American Medical Association, Annals of Internal Medicine, and Health Affairs.
Friedberg received his bachelor’s degree in economics and statistics from Swarthmore College, his medical degree from Harvard Medical School, and his master of public policy in health care policy from Harvard Kennedy School of Government. He completed an internship, a residency, and a fellowship in general medicine at Brigham and Women’s Hospital in Boston. He spoke recently with Managed Care Editor John Marcille.
MC: Are your findings on medical homes a cause for alarm for health plans?
Mark W. Friedberg, MD: Expectations are high. Advocates and policymakers want medical homes to improve the quality of care that patients receive, reduce unnecessary utilization, contain growth of the cost of care, and improve patients’ and providers’ experiences of care. But outside of a widely-cited study from the Group Health Cooperative, there has not yet been a published medical home intervention that has been proven to achieve all the goals that we might want a medical home intervention to achieve. In our Pennsylvania study, 32 pilot practices voluntarily joined the intervention. They had a financial incentive to become recognized as medical homes by the NCQA and received some technical assistance and coaching. The intervention did produce some quality improvements for diabetes, and that was the condition that they targeted the most. We just didn’t see any spillover effects on other measures of quality or on utilization or cost of care over a three-year period. The Rhode Island study was much smaller, only five practices. We weren’t able to detect any quality improvements. One of the utilization measures moved a little bit, but that study was much more resoundingly null.
MC: Have health plans proceeded in this direction too quickly without waiting for definitive research?
Friedberg: There’s a time lag on rigorously done peer-reviewed evaluation. The Pennsylvania Chronic Care Initiative started in 2008 and ended in 2011. Getting claims data, analyzing the data, and going through the peer review process takes a long time. Someone who is running a health plan doesn’t always have the luxury of waiting for that. I am very sympathetic to people who rolled this out quickly based on what they believed to be the best information available at the time.
MC: You’ve emphasized that we can learn a lot from the results of medical home studies by paying attention to the details.
Friedberg: There is a lot of heterogeneity in how these interventions are designed. Some people might assume everything is the same across interventions if they are called medical home, but this assumption would miss some very interesting and valuable variation in intervention design. Nobody is enforcing uniformity across these interventions, so in any given case the phrase medical home means whatever the practice and the payer agree it means. So one take-away from our papers is that if you have something that is called a medical home intervention, there is no guarantee it is going to do what you hope it will do. But if your intervention is different from the interventions we evaluated, it wouldn’t be unreasonable to expect a different result to emerge. If it looks the same though, the results might also be the same.
MC: Group Health is a special case, but what specific approaches to the medical home concept made it different?
Friedberg: The components of that intervention were very different from what most health plans are able to do, but they suggest some reasons why it might have worked a lot better than some of the more broadly rolled-out pilots. In that intervention, they abandoned fee-for-service completely. So they removed the incentive for physicians to do more office visits, which is remarkable. No other pilot is doing that. They also shrunk panel sizes. I am unaware of any other pilot that has smaller patient panels as a specific component of its intervention. The Group Health study was the grand slam. It hit all the marks. But it was one practice, so it’s hard to generalize from that. It would be reassuring to see the same intervention producing similar results somewhere else.
Group Health Cooperative abandoned fee-for-service completely. No other pilot is doing that.
MC: The diffuse nature of a medical home must make it challenging to study these models.
Friedberg: To avoid confusion, it’s a good idea to ask some key questions about each medical home study you encounter. For example, I think it’s helpful to ask whether a given study is about a model of care or an intervention and also whether the definition of medical home is being applied to patients or providers. Some studies of medical homes use a patient-based definition, which is whether patients report having a primary care physician or practice that they go to on a regular basis, that knows all of their medical information, and that might do some other things for them, like answer their questions at night and on weekends. But having this kind of relationship with a primary care physician can be as much a function of the patient as it is of the provider. Patients vary tremendously in whether they are going to seek out that kind of relationship with a primary care or other type of physician. In concept you could have two patients going to the same primary care provider, one who has a medical home and one who doesn’t, according to the patients’ viewpoint.
MC: That definition doesn’t seem to be of much use to an organized payer.
Friedberg: It still might be, even though these studies don’t necessarily tell us much about providers. If a payer wants to use these studies to make some kind of change, they could try to directly target patients and change how those patients interface with the health care system.
MC: For other efforts, though, they would want to look at studies with alternative definitions of the medical home?
Friedberg: Another study type looks at providers’ care models. You observe a bunch of primary care practices — maybe a national sample or a local sample — and you decide which ones are medical homes and which ones aren’t or give them a score on the degree to which they are medical homes. A common way of doing this would be to look at their structural capabilities: Do they have an electronic health record? Do they have a diabetes nurse educator? Do they have a system for tracking their sickest patients when they leave the hospital? And you combine all of these things into a score or maybe a collection of scores. The NCQA, URAC, and many other organizations have their own scales. And then, for patients who go to these practices that are more or less like medical homes, you say, what are the differences between them in terms of quality, utilization of care, and cost of care? All of these studies define medical homes differently, but that can be helpful. If one definition included enhanced after-hours access and another definition didn’t, and they have different findings, you can learn something about enhanced access.
MC: How do you define the medical home in your work?
Friedberg: We use a third definition, which is the medical home as an intervention. A medical home intervention is something that is done to primary care practices or to their environment or incentives to change them over time. It’s a bit like doing an intent-to-treat study of a new cholesterol pill. In those studies, you don’t just include the patients who take the pill every day faithfully. You also include those who, maybe because they had some kind of side effect, couldn’t take it every day. That is really important to understanding the overall effect of giving this pill to a population. People have written articles saying that the medical home is not a pill, and I completely agree. But you still have to evaluate medical home interventions in a scientifically rigorous manner if you want to have a strong evidence base for decision making.
MC: Health plans are the ones doing the intervening, so they must appreciate this type of study.
Friedberg: The intervention model is the only one directly available to a policymaker or a health plan. You can’t change patients’ care-seeking patterns or the structure of primary care practices without having some kind of intervention. All of the other study designs, like the study of a model adoption or studies of differences between patients, can suggest ways one might put together an intervention, but until you evaluate a pilot of that particular intervention, you don’t know if it’s going to work in reality.
MC: Is a practice becoming a medical home fulfilling for physicians?
Friedberg: We looked at staff satisfaction within medical homes in Colorado and Rhode Island. Our results suggested that if it’s possible to transform into a medical home without making the practice more chaotic or giving people an unreasonable amount of work to do, that gives you the best chance of improving professional satisfaction of physicians and other practice staff. We know from studies published by other researchers that medical home transformation is really difficult. So practice leaders would be well advised to carefully plan for medical home transformation and to make sure that they have enough organizational slack to do the new work that medical home transformation requires. If they are already stressed by competing requirements and pressures, it could be very challenging to have a meaningful medical home transformation without risking burnout.
MC: But once the transformation is successful, are primary care physicians happier than they used to be?
Friedberg: Group Health showed a reduction in burnout, but there haven’t been many other longitudinal studies on that particular question.
MC: Your research has shown that using decision aids can be another burden for practices.
Friedberg: We evaluated the first part of a shared-decision-making demonstration that was convened by the Informed Medical Decisions Foundation. They were trying to incorporate shared decision making into primary care practices, and we did a qualitative evaluation. We found that implementing shared decision making using decision aids — pamphlets and videos that patients can use to learn more about a decision they might be facing, like whether to get a certain kind of cancer screening — isn’t easy. Primary care practices are stretched already. There are so many different programs and initiatives out there that are focusing on primary care right now, including meaningful use and ICD-10, that adding one more thing is extraordinarily challenging, even when the practice truly believes in it. These were practices that volunteered for a demonstration, but they still had trouble getting decision aids into routine use beyond one or two champions who were leading the practice in some cases.
MC: But when they are used, that is good for patients?
Friedberg: The use of decision aids doesn’t necessarily mean that shared decision making is taking place. Critically, the provider has to loop back to the patient and have that discussion about what decision to make. And once the decision is made, care has to be given that is consistent with that decision. It is possible to drop the ball there when care is fragmented. You could make a decision to deliver care one way, and then it actually gets delivered a different way. At the time of our evaluation, none of the practices we studied had a reliable way of tracking any of this after the decision aid was used. Primary care practices in general don’t have a way of tracking a complex, longitudinal series of events for a patient. In a way, a practice might need to become a medical home to be able to really do this right, and more specifically, the practice might need to have some kind of longitudinal tracking capability that is integrated with its electronic health record.
MC: Why does that matter?
Friedberg: Shared decision making is a requirement for some Medicare ACOs. They have to do it. How that gets measured is really important. If it’s just distributing decision aids, that doesn’t mean that shared decision making is actually happening. We don’t yet have well-developed process measures for tracking whether the steps of shared decision making are occurring.
MC: Can a patient be permitted to make decisions that will increase payers’ cost down the line?
Friedberg: If a patient is making a decision that is consistent with medical science and also consistent with their values and preferences, it is certainly possible that decision might result in some increased costs of care. Let’s say that you inform a patient about some kind of elective surgery and you give them all the pros and cons, and most people would decline to have the surgery because it is risky. But for one patient, their values and preferences say to go ahead and get the surgery. If I were in charge of a health plan, I would permit that patient to have it because ethically, that’s the thing to do.
MC: As the patient-centered medical home becomes more common, do you see other approaches that should be tried and evaluated?
Friedberg: Primary care practices play a really important role in the care that their patients receive, but there are other participants in the health care system that also play important roles, like hospitals, subspecialists, and so forth. This is where you get ideas like the medical neighborhood and medical home recognition for providers who aren’t primary care providers. Those kinds of interventions and models are at an earlier stage of development and probably have a major role to play, but we are at an earlier stage of evaluating them as well. Hopefully, they’ll be subject to rigorous evaluation so we can learn as much as possible from them.
MC: Is the ACO one of those?
Friedberg: Absolutely. The difference between a medical neighborhood and an ACO might not be all that much. It might be a labeling issue and probably depends on the details of the contractual relationships.
You have to evaluate medical home interventions in a scientifically rigorous manner.
MC: Just like medical homes?
Friedberg: With all of these interventions called medical homes going on, if we stop paying attention to whether they work as intended, or we paper over the differences between them and just call them all medical homes without looking carefully at the details, we stand to miss out on a lot of scientific knowledge that could make these interventions more effective. Hopefully over the next couple of years, we will see published evaluations of medical home interventions that show different kinds of results than what we showed in southeast Pennsylvania. The key isn’t then to say, “Oh, that first study was wrong and these other studies are right.” The better response would be to ask: Why they are different and what can we learn from that difference that would then help inform the evolution of these interventions to make them more effective?
MC: Thank you.