Leemore Dafny, PhD, interviews Michael Chernew, PhD, the Leonard D. Schaeffer Professor of Health Care Policy and Director of Healthcare Markets and Regulation Lab at Harvard Medical School.
Leemore Dafny: This is Leemore Dafny for NEJM Catalyst. I am speaking today with Michael Chernew. Professor Chernew is the Leonard D. Schaeffer Professor of Health Care Policy and the Director of the HealthCare Markets and Regulation (HMR) Lab in the Department of Health Care Policy at Harvard Medical School. His research examines several areas related to controlling health care spending growth while maintaining or improving quality of care, and he studied a variety of insurance-driven payments reforms, including value-based insurance design and the alternative quality contract in Massachusetts. Today we’ll be focusing on Professor Chernew’s work and insights regarding alternative payment models. Welcome.
Michael Chernew: Thanks for having me. It’s wonderful to chat with you.
Dafny: My first question for you is that we’ve known for quite some time that fee-for-service is broken. Why, in particular, is the government doing something about that now?
Chernew: We’ve been at it for a long time. In the 1980s, they started with the DRG system for hospital payments. By the 1990s, they were on to physician payments with the RBRVS system, which has given us relative value units. Then in the late ’90s — but it really got started in the early 2000s — there was a sustainable growth rate system for physician payments, which was many things, but not sustainable.
So there’s been this evolution of activities regarding payment reform, primarily, frankly, to control health care spending growth. And we hadn’t been that successful with these other strategies for a variety of reasons, and so that’s why I think there are these new ideas. I think in addition to this desire to have a payment model designed to help control health care spending growth, we wanted a payment model that allows providers that practice care efficiently to benefit economically from that behavior, and by doing that we can create an incentive for efficient practice of medicine.
Dafny: The government (CMS) and payers have been serious for quite some time. They’ve implemented other approaches that basically have failed in terms of controlling health care costs. What is new and how is it working?
Chernew: Well, there’s a number of things that they’re trying to do. Let me start with what I would call, broadly speaking, the changes in the way they’re trying to bundle payments. They’re trying to take the payment model and give a payment that spans providers and spans time. So that would include a population-based payment model like global budgets, for example, and also episode-based payment models.
The population-based payment models set a fixed target of spending for an entire person over typically a year, and the episode payment model looks at a particular type of client care — say, a hip replacement and knee surgery, something like that — and sets a budget for that that spans providers and spans times. You see that, for example, in the Bundled Payments for Care Improvement initiative. Personally, I have some concerns about how broadly applicable episode-based payment models can be, because so much of our spending is among patients with multiple chronic conditions. So it’s very hard to divide up that spending to specific episodes and figure out exactly which provider should be accountable. But there’s a lot of innovation going on in that space, and we’ll see where they actually go. And those types of episode models can actually be well suited to target specific providers. The population models are much broader, but they require large organizations to manage them, and they require those organizations have mechanisms to allocate this broad population-based payment down to the different providers that care for their patients.
Dafny: At some level, as you described it, population-based health has really been what insurers have been doing, which is taking a set amount of funds and managing the care to some degree of enrollees and dispersing those funds. So perhaps you can explain to me why policymakers think that providers will do a different job, and hopefully a better job, of managing population health with the funds.
Chernew: I think, fundamentally, the success of the health care system requires that the people who deliver care, and the organizations that they’re affiliated with, practice care efficiently. And the insurers, in accepting this population payment — which we call a premium — when they allocate it to the providers, the incentives that the insurers faced didn’t get translated very well to the organizations that were actually delivering care. And the essential idea behind these population-based models is to move that incentive to manage spending for a population away from the insurers that had very weak levers and very weak control over the actual delivery of care down to the organizations that actually are on the ground, interacting with patients. And that, essentially, had been the theory, and of course, as these population-based payment models move forward, many of them currently — and there’s increasing aspirations to have more of them — require the providers’ systems to bear risks.
Dafny: You’ve done extensive research on alternative payment models. Can you give me a summary of your understanding of ACOs and how they’re working?
Chernew: The official ACOs are in the Medicare program, and our sense is — it’s very early on in the game — they’re saving a few percent. That’s not nearly as much as advocates would have hoped for, but on the other hand, if you can save a few percent, that’s better than not saving a few percent, and they’re expanding. Again, somewhat more slowly than advocates would have liked, but more organizations are accepting these types of models. Here in Massachusetts, there was an early ACO-like program — it was called the Alternative Quality Contract implemented by Blue Cross and Shield of Massachusetts — and our estimates were that by four years that program saves roughly 10 percent relative to what otherwise would have been spent. It’s really important to understand that spending is not going down. It’s just going up more slowly. A lot of those savings in the Alternative Quality Contract was because in the private sector there’s very wide variation in prices, and when the providers had the incentive to direct patients to the lower-priced settings and organizations, they did that. And about half of the savings were due to those types of activities.
Dafny: What I’m hearing you say is that the way in which a commercial ACO reduces health care spending, and presumably while achieving or exceeding quality targets, is by redirecting patients to lower-cost sites of care or lower-cost providers. That — at least the lower-priced provider angle of it — is not going to work for Medicare ACOs.
Chernew: Yeah. So that’s about half of the savings in the commercial programs we’ve studied. And it still can work in Medicare to some extent, because there’s such different prices for similar services delivered in an office-based or facility-based setting. So there’s some amount of price variation, but yes — you’re correct. There’s a lot less room for that in Medicare ACOs.
So we would expect, long run, the saving opportunities are less. But remember, we’re not trying to lower spending; we’re trying to change the rate of growth in spending, and historically the rate of growth in spending has been driven largely by volume increases. So in a population-based payment model for Medicare, if you can control the benchmark, which is really the central policy parameter in my view, you can control the rate of growth and spending and essentially hold the providers, as opposed to the insurers, accountable for a rate of spending growth that’s more in line with the rate of income growth.
Dafny: Right. So a sustainable rate of spending growth certainly sounds like a worthy target, and actually reducing spending is not something that, I think, is at all likely to happen. So if I’m a provider organization, what kinds of advice would you offer to me as I think about how to succeed in this new set-up?
Chernew: What is essentially happening with all these new models is that they’re changing the business models that provider organizations face. And the organizations, to be successful, have to be more sophisticated and think more broadly about populations, because in general, they’re accountable for care outside of their walls in a whole variety of ways over time with other providers. So most of these models, the population ones, are centered around primary care physicians. The episode models, less so. They’re developed, in some ways, in ways that can work more for specialists. But all of the organizations have to figure out how to find where the value lies and succeed by reducing the amount of care — ideally, reducing the amount of wasteful care in the system, as opposed to increasing the amount of care. You succeed, essentially, by translating reduced waste into profit, which can only be done in these new models.
Dafny: Can you speak to the questions surrounding how providers are organized to achieve this? And here, specifically, I’m referring to whether physicians need to be jointly owned with hospitals. Do they need to jointly own all of the different sites along the continuum of care in order to achieve this? What does your research tell you about whether that’s likely to be the route?
Chernew: In our first analysis, we found that you could succeed under either organizational structure. Essentially, if you’re a hospital-based organization, very integrated between hospitals, physicians, and perhaps other providers, post-acute and others, you really do have the ability to coordinate care across the spectrum. The challenge, of course, is as you try and reduce wasteful care, many of the providers of that care are part of the organization, and it really becomes a challenge. So if your strategy is to reduce hospital care, that might be harder to do if the organization is dominated by the hospital. On the other hand, organizations that don’t include those other facilities inherently are going to need to refer their patients outside of their control, and they’re going to need to manage that better through contracting, referral patterns, and things of that nature. And we’re in a period in which we’re exploring which of those organizational forms work. Frankly, I don’t think there’s going to be one answer. My general sense is organizations of both types can be successful. It’s just that they have to execute on their strategy well, and understand that in order to succeed, there will be some organization that’s going to have to receive less money than it otherwise would have received, but still probably more money than it had in the past.
Dafny: Right, and operate under an entirely different incentive structure than it currently does, which brings me to the next question. We’ve actually heard and read a lot about ACOs, since even before the Affordable Care Act was passed in March 2010, and it still seems like they are just taking root, and that fee-for-service is a better descriptor of our health care payment system today. What do you think is slowing down the transition, and are there any steps that you think either the government, or state legislators, or providers, or payers, could take to accelerate the pace?
Chernew: Yeah. I think, in fact, these types of models are growing, both in public and the private sector. They might not be growing as fast as one would expect, but the trajectory still seems to be moving upward. I think the fundamental issue in terms of accelerating the pace is the extent to which you maintain pressure on the fee-for-service system, which is currently scheduled to happen — if you look at the new fee schedule that was put in place following the repeal of the sustainable growth rates through a legislation that was called MACRA in the system, [and] related to this MIPS, or the new quality payment model. In that system, the updates for physician payments are 0.5 percent through 2019, and then they drop down to zero. There are quality bonuses that go on top of that.
Ignoring whether we think the quality measures are good or bad, that system is designed to have winners or losers. So the net amount of money going into the physician system, and to some extent the hospital systems with the productivity adjustments in the Affordable Care Act — all of the fees in the fee-for-service system are scheduled to rise very slowly. Much more slowly than we would anticipate inflation being. In order for providers to succeed in that model, they need to be able to convert efficiencies into income, and these new payment models allow them to do that. If we relax these payment models — obviously, put more money into the system one way or another — we will discourage the transition to the alternative payment models. The other thing that’s really important is that the rules around the alternative payment models, particularly the federal ones, have been changing periodically. We had the Pioneer program, now we have the Next Generation program. At some point, we’ll probably have the next, next generation model, but in any case there are a lot of subtle rules that sometimes make it hard for organizations to succeed — the way the benchmarks are set, the size of the shared savings, a number of other governance rules and a whole bunch of rules. And we’re still in a period of experimentation, both in the public and private sector, to figure out exactly what works. I could give examples of private-sector versions of this where these are evolving and moving forward. So it’s a combination of push and pull.
This interview originally appeared in NEJM Catalyst on February 12, 2016. Listen to or read Part 2 of the interview.