How value-based care does (and doesn’t) improve health equity

I think we've learned a couple of other elements. We've learned that there's most likely three core types of payment programs that we're likely to pursue in the future. There's an orientation around supporting primary care physicians and primary care payments. That is likely to live on. In what form, I think we're still figuring out, because the evidence on primary care payment itself has not been the most promising.

A second area where we've seen successful testing has been in the space of population health-based payments, population-based payments. So, these are models, like direct contracting or Accountable Care Organizations, or ACOs. I think we've seen promising, not definitive, but promising, evidence in this space. And I think we're likely to see that chassis become the payment chassis of the future.

And then the third area is specialty in hospital-driven payments, in an episodic fashion or a bundle payment fashion. And there's similarly promising evidence, some home-run-type evidence that we've seen in that space, certainly figures to be an important lever that will be a part of the portfolio going forward. But I think we don't yet know how we're going to stitch all of these pieces together.

Melissa Clarke: That's an excellent summary. I want to pick up on one of the things that you mentioned about some of the successes in bundled payments around specialties. They've seemed to be more successful for surgical procedures, like joint replacements, but not so much for non-surgical ones, like heart failure or cancer. Why do you think that is?

Amol Navathe: I think that's a terrific question. What we have learned over the past several years, and much of my research group's work has been dedicated to building the evidence base specifically around episode and bundle payments, we've learned that the types of care redesign and practice transformation that are needed for surgical, versus, say, medical condition types of bundle payments, are considerably different. And, so, if we take each in turn, I think we can elucidate a little bit about what I mean here.

So, take a hip and knee replacement or hip or knee replacement surgery as kind of the canonical surgical bundle payment. Right? And, so, this is grandma, who needs a new knee, she is probably reasonably active. She's active and healthy enough that her primary care doctor and the orthopedist or the sports medicine specialist says, "Hey, you know what? We want you to keep yourself moving and, therefore, it's a good time to get a new knee."

It also means that the anesthesiologist and the primary care doctor say, "You're healthy enough to go through an elective surgery." And then when she goes into the hospital, the surgery is the surgery. And then, as we're thinking about what happens next, there's now a greater rationalization under bundle payments of, what level of post-acute care does grandma need? Does she really need to go to an inpatient rehab facility and stay in the hospital? Could she go to a skilled nursing facility, where there's a lot of onsite physical therapy, or maybe she's actually healthy enough, and she has some support at home such that she could go home and get physical therapy, either in the home, or she could travel once a day to physical therapy.

And what we found is that, historically, when the system didn't really have incentives, really, frankly, to care, if you will, about what level of intensity of post-acute care, we sent a lot of people to inpatient rehab and to SNFs, to skilled nursing facilities. And we probably didn't need to do that. And what bundle payments in the surgical setting have shown us is that grandma can do quite well if we send her home, and she gets either physical therapy in the home or goes to visit a physical therapist.

And, so, I tell that narrative here to highlight that the number one mechanism for savings under surgical bundle payments have focused on shifting the intensity of post-acute care from the facility to the home. And that is a relatively discrete decision, that can be coached into patients, as they go through the pre-op process and their surgical process, by well-informed surgeons and anesthesiologist and nurses who work in the hospital.

And, yet again, the point to make here is, that it's a relatively discrete decision. A lot of it is around expectation-setting and appropriate risk stratification.

Now, let me contrast that with the other part that you were mentioning, which is, what about a patient who's in the hospital for congestive heart failure or sepsis? Right? These individuals tend to have many chronic conditions. If you have congestive heart failure, you most likely have something else too. You probably have coronary artery disease. You might have a touch of renal failure. Right?

So, these individuals are sicker, on average, and there's a lot more complexity to what's happening before the hospital, in the hospital and, most importantly, here, after the hospital.

And, so, what we've found there is the types of care redesign efforts that do yield benefit, they tend to be a stronger network integration between hospitals and the post-acute care facilities, again, thinking about the inpatient rehabs and the skilled nursing facilities.

But here, we're really talking less about not discharging grandma to the skilled nursing facility, but more thinking here about, how can we get grandma in and out of the skilled nursing facility in an efficient fashion? That's where we have seen the most amount of savings. Now, candidly, they're smaller in magnitude than we see on the surgical side, but that's just a much more complicated situation.

And I think that's one of the reasons that we saw savings emerge for medical conditions, like congestive heart failure or sepsis, et cetera, but they took a longer time to emerge. They took years, three years, four years, in the programs, before we started to see savings.

And, while quality didn't seem to be harmed, I would say that this magnitude of the savings was not huge. And, so, I think that, to some extent, sets the backdrop of the evidence that, Dr. Clarke, you're referencing, which is, do bundle payments work better for surgical conditions?

Melissa Clarke: That makes a lot of sense, in terms of that discrete decision point for surgeons and surgical patients, versus the complexity of factors that are happening with medical patients and, therefore, the longer time frame.

And we're talking a lot about behavior change within the system. How do you think that CMS can really, ultimately, via the CMMI or other levers, mobilize late adopters of payment reform models, sort of away from that fee-for-service model, and leapfrog them more towards advanced payment models?

Amol Navathe: That a fantastic question. I can tell you that it is very front of mind for policy makers currently, both on the legislative side, as well as, as you're pointing out, at Medicare.

Conventionally, we tend to think about, and by we, here, I'm saying more of the research community and the policy community at large, we tend to think about the early adopters as the ones who are most likely to succeed, and the late adopters to be ones perhaps that are a little bit less willing to dip their toe in the water because they might have less capabilities and capacity for change.

Interestingly, I actually don't think that the evidence supports that, from what we know at present time. So, to be very clear here, I think there is some evidence that that exists. But at the same time, when we have looked at organizations that are mandated, for example, into a hip and knee replacement bundle payment, versus those that chose to volunteer earlier, we found that actually they have pretty similar savings. They have pretty similar care redesign success.

And, so, this notion of, do we need to mandate, versus can we have people opt in, have organizations opt in, as voluntary participants, and what is that variation in timing, I think we're still piecing some of these things together.

Now, what we have learned, and I think this is really a fundamentally important point, is that if we can create the right incentives for participation and adoption, we can still be quite successful. And, in other words, the organizations that choose to adopt later can actually be quite successful.

And I think we have seen that in the context of the Accountable Care Organizations Program, where, year after year, organizations that have joined, they tend to improve over time. And what is more important is how long they've stayed in the program, and what is less important, is which year they decided to join. Did they join in the first year of the program or the fourth year of the program?

If we have longer durations of committed participation, if the incentives are rolled out in a way that longer participation actually is more beneficial, rather than static or less beneficial, then I think we can start to change the calculus. We can start to change the math here a little bit, to where those late adopters, while they may adopt later, that doesn't mean that they won't necessarily adopt in a sustained fashion. And I think that's where we can start to really make gains.

And, so, I would start to shift our paradigm a little bit, away from, "Do we care about early versus late?" to more, "Can we really drive sustained adoption or sustained participation?"

Melissa Clarke: When talking about participation, it makes me really think about health equity because we know that providers and communities that serve low-income populations and communities of color were less likely to participate in both the comprehensive primary care programs, and as well as the Accountable Care Medicare Shared Savings Programs.

And as you talk about operational investments, it makes me think that perhaps this is one of the reasons why they didn't. Do you think that that might be the case, and are there others?

Amol Navathe: Dr. Clarke, this is a fantastic point. And I think, if I can emphasize one thing, you asked me kind of, "What did we learn in the first decade post Affordable Care Act?"

And I think one thing that we also learned here, which is just an incredibly important lesson, is that chasing value, value-based payment models, is not the same. It is not equivalent to equity.

Once again, if I restate that again, I would say, value does not necessarily translate into equity, unless we reframe what we mean by value. But I think, conventionally, when we've thought about value as how much payment do we put, how much cost do we have in the system, for how much health do we get out, in some formulaic way?

Notice, there, equity is not a part of that formula. And I think that has been the governing formula of how we've thought about value.

And, so, your points are really, really fundamentally important, which is, if we continue down the path that we've been going for the past decade, we're not very likely to make great gains in equity.

Well, we just talked about how most of these programs have had benefits, in terms of there's been quality gains across the ACO program, for example. There's certainly been cost savings that are generated. Some of those cost savings end up in the pocket books of our patients, of our beneficiaries, through cost-sharing savings.

And, so, it should feel very inequitable if some portion of our population is more likely to get these innovations that either yield them quality benefits, or put more money in their pocket books, or both. Right? This is a fundamental problem of equity in our system, which, as it is, has some deep-seeded structural inequities and disparities that we've seen over decades.

So, number one point here is, I think your question is great because we really need to care about that question, around participation inequity, participation inequities, I apologize.

And I think we also have to differentiate two other pieces here, alongside that. I think sometimes we look at this view and we say, we look at the model evaluations, and we look at the research around the programs, and we see, while it looks like populations who are marginalized or traditionally faced disadvantages, aren't harmed by or are included, to some extent, in the care redesign of the model. And I think that's important to differentiate that from, what is the impact of the value-based program on those communities and on those patients as a whole?

And I'll kind of end this point by highlighting one other important piece, which is, there are reasons to worry about value-based payments, as they're currently and historically have been constructed, in terms of their impact on equity for marginalized populations.

And the reason for that, the principal reason for that, is that, when we design and implement value-based payment models, we are intrinsically putting the clinicians and the delivery organizations in a position of accountability for cost and quality. And, so, populations who, on average, are more difficult to change cost and quality for, are going to be less advantageous, if you will, to have in the model.

And I think that's, in part, what you're highlighting, in terms of the participation incentives, that we've seen less participation from organizations that are situated in markets, where there's a greater proportion of communities of color, of communities of low socioeconomic status. And I think that is something that we need to deliberately try to address, going forward.

Melissa Clarke: No, I think you hit on some excellent points. The only thing I might add in is that, when we think about value, in terms of what we're putting in, the health that we're getting out based upon the costs that we're putting in, if we think about the fact that health care disparities, per some research that originally came out of Johns Hopkins, showed that not closing health disparity gaps costs society $230 billion a year. So, somewhere in our value calculation, that really needs to enter into the picture.

And I think that probably goes to the point that you made earlier about participation in the model being more difficult for those providers who sort of had a heavier lift due to adverse social risk factors. Is that sort of one of the things that you were getting at?

Amol Navathe: Yes, that, indeed, was one of the things I was getting at, in that implementing care redesign and affecting the outcomes for populations who have traditionally faced disadvantages or are marginalized, is harder, and it requires a more multi-dimensional, if you will, approach.

And if we take the example that I was using earlier, around grandma getting a new knee replacement, if, unfortunately, grandma comes from a situation where she doesn't have caregiver support at home and somebody to get her from home to a physical therapist or somewhere else for her doctor's appointment, then you can imagine that actually sending grandma home is less attractive and probably not the right thing for grandma.

And, so, if you're an organization, whether you're a primary care practice or a ACO or a bundle payment hospital, or what have you, you might not feel comfortable doing that. And that, unfortunately, in the value-based construct, makes grandma a less attractive financially, financially less attractive patient.

And, so, I think you're absolutely right that the complexity of addressing the multifactorial nature of challenges for some of these populations, where we can't just put blinders on and focus on the medical aspect, that we, intrinsically, to take good care of these populations and to change the way we care for these populations, means that we have to address the social and the behavioral elements as well, makes it a more daunting challenge. And I think, in part, what you are alluding to, is it makes the investment required greater, greater in terms of financial resources and just greater in terms of effort, the number of hours that men and women are spending, trying to address this, these issues.

Melissa Clarke: Are there any remedies or levers that CMS has, going forward, to really try to address some of these issues around participation of providers that serve marginalized communities or the differences in shared savings received, to make it more feasible for providers to participate?

Amol Navathe: That's a terrific question. I think it's a work in progress, quite candidly. That being said, there are examples where this is working. So, there is a CMMI model called the ACO Investment Model, where I think there was a recognition exactly, Dr. Clarke, of what you were saying, in terms of the Johns Hopkins study, which is, if we don't actually engage these communities, if we don't engage the health care organizations and clinicians that serve them, then, in fact, down, in the long run, we're going to have a less efficient system, and we're going to have a lot of wasted health spending.

And number one, it was more successful. The ACO Investment Model was successful in enrolling ACOs in those areas. And number two, they actually were successful also in generating quality gains and saving dollars.

So, to some extent, I may not say that we have a perfect template quite yet, but I think we at least have a great use case that showed that this is feasible, in the context of value-based payments for value-based care in underserved and rural communities.

And I think if we can take that as a really important case study, we can apply them in other areas. I think we have seen this administration start to pivot in that direction in a general sense. And, by that, what I mean is, I think to truly make gains on equity, we can't make it an afterthought of how we design policy. In other words, what I think the typical approach has been is to design a model, change a policy, and then to evaluate and monitor and say, "Well, how is this model affecting populations that are more underserved? What's happening to inequities and gaps and disparities?"

And if we do that, we're always going to be chasing our tail and unlikely to really proactively make progress against these social challenges that we have. Instead, if we directly say, "Part of the goal of these models is to improve equity," then I think we're playing in a different playing field.

And we have seen the administration do that, in the context of two programs that I'll mention. One is the ACO REACH Program, which is the new incarnation of direct contracting. And in that program, there are direct incentives around equity. And there is a new precedent, in terms of incorporating area-level social factors at least, in part of the risk adjustment and part of the financial calculations, the benchmark policy. I think we're still learning more about the details, but I think the intent is one that is capturing the essence of what you are headed toward, which is, are there policy levers to address this? I think the administration currently is trying to do it through that program.

There's another example in specialty care, which is also worth mentioning, which is, there's a program called the End-Stage Renal Disease, or ESRD Treatment Choices Model. And traditionally, for beneficiaries with end-stage renal disease who are on dialysis or will require dialysis, we have found that there are pretty startling disparities along racial lines or on ethnic lines and around socioeconomic status lines, around getting even wait-listed for a transplant, for a kidney transplant, much less getting a transplant itself, as well as use of more innovative modalities, like in-home dialysis, relative to in-facility or in-center dialysis.

And this program, in fact, has a direct financial incentive for nephrologists and dialysis centers, if they increase the wait-listing for transplants and increase the use of home-based dialysis for populations who are of low socioeconomic status.

Melissa Clarke: Great. That an excellent summary, Dr. Navathe. You mentioned that in one of the models, the ACO REACH Program, that area-level social factors are part of one of the levers that are used. Do you ever think that individual-level social factors might also come into play in, for example, payment adjustment, for those providers who care for larger numbers of patients with increased social need?

Amol Navathe: Conceptually, I think it is very appealing. The idea that we can adjust the payments that we are making, say, to accountable care organizations, even potentially Medicare Advantage plans, and really the intent here being to increase the amount that we're paying for their care, to allow for organizations to then try to address some of those social determinants of health. I think, conceptually, it's very appealing.

I think, however, we have to be very, very careful and very thoughtful about how it's implemented. I think there's a very intuitive reason for why that is. The reason is that, the very populations that we're seeking to increase payments for, are frequently the same populations, beneficiaries, for example, in the Medicare program who are of Black race, of Hispanic ethnicity, those who are in the dual eligible program, who are simultaneously eligible for Medicare and Medicaid because of their income and wealth status, those are the same populations that tend to struggle with access to care.

And, so, it is possible that if we just did the math, so to speak, that adjusting for a social risk factor, could, in fact, decrease the payment, or decrease the benchmark, depending on the program.

And, so, I think before we roll these programs out and say, "Social risk adjustment is the answer," I think we need to be careful, and we need to test, and we need to validate, and we need to kick the tires on twice, three times, to make sure that the policy intent is matching the policy implementation. And I don't think, as I mentioned, that just doing the math is necessarily going to be a one-to-one relationship there. I think we have to be thoughtful about it.

The second point that I would highlight that is important to keep in mind, as we think about future policy here, is over-medicalizing the social system, Social Services system. So, what do I mean by that?

Imagine that we are trying to provide housing services for an individual who is homeless or who has an unstable housing situation. Well, conventionally, we would do that through the Social Services system. If we did that, for example, through an insurance company or through a hospital, because we'd want the medical system to be more accountable, if you will, for social circumstances of patients, then the way we're designing that is we're creating an incentive for the insurance company or the hospital, which isn't an ACO, to provide that service because it is likely to reduce health expenditures in the medical system. Yet, there might be reasons that society wants to provide a more stable housing situation for a community that isn't related to health expenditures itself.