Health systems continue to face strong headwinds as they seek to shift to value-based care models. They're not able to advance the redesign as rapidly as they'd like, given the small total share of their revenues these value-based payment models represent, explains RAND Corp.’s Cheryl Damberg, Ph.D., M.P.H.
Damberg, director of the RAND Center of Excellence on Health System Performance, distinguished chair in Health Care Payment Policy, and a principal senior economist at RAND, was speaking at the March 7 meeting of the federal Physician-Focused Payment Model Technical Advisory Committee (PTAC). She and Michael Chernew, Ph.D., a professor of healthcare policy at Harvard Medical School and the chair of Medicare Payment Advisory Commission (MedPAC), described some challenges with the current situation and offered some suggestions for how CMS might proceed.
Damberg’s organization has studied accountable care organizations and other entities that are being held accountable. She noted that they tend to have incentives to lower spending on care that they actually don't provide. “What we see here in California is these large physician organizations — there are about 180 of them in the state — being held accountable for total cost of care, but where they have looked to reduce spending has really been on the inpatient side, so it hasn't really affected their personal bottom line, if you will.”
The other thing that her research has found is that the uptake of these models has varied. “I have to say, looking at the list of entities that signed up for the Direct Contracting model, the most recent CMMI demonstration, I was actually kind of surprised at who I did not see at the table,” Damberg said. She added that regulators should be talking to those groups about why they're not coming to the table. “A lot of the risk-bearing entities that have a lot of experience in this space were not at the table for that demonstration. The question is, why are they sitting it out.”
She also noted that a lot of these alternative payment models are built on a fee-for-service chassis. “So there's lots of bleed-out and challenges that providers face in managing the total cost of care for the beneficiaries or patients assigned to them.”
The work Damberg’s team has been doing over the past five years has involved studying U.S. health systems and cataloguing what they are doing to try to drive performance improvements ranging from total cost of care to clinical quality to reducing low-value care. “We looked at large health systems — these are physician/hospital health systems that have 50 or more physicians of which at least 10 are primary care physicians. When we looked at those who were participating in the Medicare ACOs, whether the one-sided or two-sided risk contracts, we find that a fairly small fraction of their beneficiaries are actually enrolled in these ACO arrangements,” she said.
Fifty percent of the entities had 18 percent or fewer of their beneficiaries in ACO contracts. “When we look at the beneficiaries who are in ACOs in these health systems, we are finding higher performance on clinical quality measures and better performance on lower value care,” she said. Trying to push those numbers upwards can help drive the value part of the equation. But it's very hard to redirect resources to population-based care delivery if only a small fraction of your population is under these models, Damberg added.
“We face some pretty strong headwinds,” she stressed, “because these health systems report that they're not able to advance the redesign as rapidly as they'd like, given the small total share of their book of business that these value-based payment models represent.”
Generally, they would report that 5 percent or less of revenue is tied to value-based payment arrangements. Oftentimes, they were in markets where, at least on the commercial side, the commercial payers were not facing a lot of pressure from employers to shift toward those models, she explained. They are still operating in this fee-for-service space. “Because they have so many different payment models that they face from Medicaid, Medicare and the commercial side, they're trying to figure out how to balance all these different incentives,” she said. “What they do is play to the middle, and right now that middle is skewed heavily to the fee-for-service delivery.”
Front-line physicians in these large health systems are largely living in a fee-for-service world, too. The incentives are very small for anything around total cost of care; in most cases, that was missing from what they were held accountable for, she said. The incentives tended to focus mostly on things like clinical quality, patient experience, and increasing the volume of patients moving through the system.
“When I talked to physician organizations and health systems, they're staring down more than 200 quality measures,” Damberg said. “They don't feel like they could ask their front-line physicians to focus on any more than a handful. They're really struggling with thinking through what share of their revenues is coming from which sources and which incentive to pay attention to.”
She said we need to think about the incentives to reduce spending, and whether they're high enough not only to induce participation, but also to cover the costs of participating, and the types of investments that providers have to make. “I would encourage CMMI as well as private payers to emphasize testing models that really start to shift toward true population-based payment. I think we've seen very few of those models. Direct Contracting is one such model. I hope that there will be other such wholesale type models tested in the future.”
Harmonizing alternative payment models
Chernew stressed that in his comments to PTAC, he was speaking in his role as a professor of healthcare policy and not as chair of Medicare Payment Advisory Commission (MedPAC). But he did discuss the idea of harmonizing APMs, which is something MedPAC has proposed to CMMI.
He spoke about why it is important to have alternative payment models work in harmony. “When this current journey was launched into payment reform, we launched it under what I used to consider a “let many flowers bloom” test-and-diffuse paradigm. The basic idea was we were going to try a bunch of payment models. The ones that work, we were going to diffuse and the ones that didn't, we weren't.”
Chernew said that was complicated, because it was difficult to know what the control group was and the environment was always changing. Also, you never knew if you participated in a model whether that model was going to continue. “Every model had an uncertain future, which discouraged participation, and it tended to disincentivize savings, because you don't want to make a big investment to succeed in a model that may be ended,” he said, “and if you have a lot of models occurring at the same time, the savings might get siphoned away.”
He explained this by noting that there's a certain amount of waste in the system. When we set up a portfolio of payment models, we are essentially assigning the waste to a delivery organization. If they can eliminate that waste, they get to share in some of the savings. But if we have broad, population-based payment models, where the waste is assigned, for example, to an organization that employs the primary care physician, and then we take a portion of that waste and assign it to another model, say an episode model, that removes the potential savings that the ACO could have had from trying to eliminate the same waste.
“Now they might not have done it, but you're reassigning where the waste goes, and that discourages participation and disincentivizes saving,” Chernew explained. “For example, if you had a model that was assigned to physicians to manage congestive heart failure, but you gave savings associated with anything post-hospitalization to a hospital as opposed to the cardiologist, you discourage the cardiologist from participating because you've taken a certain portion of waste in that stream and assigned it to some other entity.”
MedPAC’s recommendation, Chernew noted, was that the secretary of Health & Human Services should implement a more harmonized portfolio of fewer alternative payment models that are designed to work together to support the strategic objectives of reducing spending and improving quality. “What that essentially means is that instead of just launching models as they come across the transom, one should think strategically about the portfolio of models that are launched, and make sure they're harmonized. You might not want three lower-extremity joint episodes, for example. You might be careful if you started launching episode-based payment on top of ACOs, because every time you do, you siphon some of the savings away from the ACO towards the episode. There should be fewer types of models, and they should be designed in recognition that the others exist.”
Chernew also said these models should be designed in a way to avoid the “ratchet” in the benchmark. “It becomes problematic if when you save during one performance period, that lowers your benchmark and in future performance periods, you basically just have lag penalization, so you're always competing against yourself. Eventually that model will fail,” he said. “You need to design the payment models in a way to avoid that ratchet.”
Chernew closed by saying that “MedPAC is not CMMI, so Liz Fowler and her colleagues at CMS have a much more difficult job than I do. We say conceptually, here's how you might think about things. But the rubber hits the road where you actually have to make all these principles work in practice.”