The Future of Medicare Advantage: Top Policy Issues for the Nation

Welcome back. I'm Susan Dentzer, and I'm happy to be here with Michael Chernew. Mike is the chair of the Medicare Payment Advisory Commission. He's also a professor at Harvard Medical School. He's the Leonard D. Schaeffer Chair of Health Care Policy there at the Harvard Medical School. Mike, welcome. Great to have you today. Susan, it is great to be here and great to talk about Medicare Advantage, so thank you for having me. Great. Well, Mike, we look at the two programs today, Medicare Advantage and the traditional Medicare program, and sometimes I like to think about sort of driving through the countryside of New England and coming across one of these connected farms where you have a house next to a barn next to an annex next to something else. And if we think about the Medicare program, once upon a time, there was this little annex to the Medicare program called Private Plans. But now that annex is getting to be almost the main building, or at least a co-equal part of the building, now that enrollment in Medicare Advantage is almost half of the program, at least in terms of all the beneficiaries who are enrolled in Parts A and B. So as we think about this very different scenario now where you've got this very, very large program that is probably going to continue to grow in magnitude, at least to all indications, it's raising so many challenges about the comparison and contrast between these two programs from a very important spending standpoint, as well as much else. So let's start there. From the perspective of MedPAC, as we think about the top issues that are rising up now in terms of policy, how do you rate the most important issues that arise from this difference between the two programs? Because they essentially are evolving into very different programs that, frankly, are serving different purposes. Yes. So let me talk about your question in two different threads. The first one is a very short-term issue, which I think is the biggest one, which is the Medicare Advantage payment rates, often called the benchmarks. And for a range of reasons I imagine we'll get into, Medicare Advantage plans are paid more than the Medicare Advantage enrollees would have spent had they been in fee-for-service. And bringing that misalignment into some type of alignment I think is the most important topic. And there's a bunch of subsets, how coding works, for example, that all fit into how benchmarks are set, that all fit into that thread of discussion. Then there's the sort of more fundamental long-term issue, which is, as you pointed out, and by the way, people should come to New England, particularly in October. It's a wonderful countryside to drive through. But as you pointed, they can think about Medicare Advantage now instead of leaves. In any case, the program was built as if it was going to be small and traditional Medicare was going to be big. And they set the payment rates for the Medicare Advantage program based on the traditional Medicare program, the spending for beneficiaries there. If the traditional Medicare program were to shrink or, in the extreme case, go away, just mathematically you would not be able to have a Medicare Advantage program because you wouldn't be able to create the benchmarks necessary. Already for 15-ish percent, a little bit less maybe, of counties, CMS has to do what's called a credibility adjustment where they have to adjust the way of setting the benchmarks because there aren't enough fee-for-service people there. And there's a bunch of other issues that the credibility adjustment doesn't really account for. So I think there's a long-term question about how would you prepare for a much bigger Medicare Advantage program, which is separate in some ways as a longer-term set of strategic thinking, than right now how do we try and align the payment a little bit better. Great. Okay, so let's go to that essential issue there, the benchmarks. And, of course, the benchmarks are the targets, in effect, against which the MA plans bid. And those, as you said, are set on the basis of fee-for-service spending, with the exception of the places where there aren't enough fee-for-service beneficiaries. So let's just talk about that issue. You and some of your colleagues at Harvard authored a paper recently in Health Affairs where you looked at what the implications would be of lowering the benchmarks. Why would we do that from a policy standpoint, and what would be the effect of lowering those benchmarks? So, first, I should say that that paper was written as Michael the professor, not Michael the MedPAC chair. But I think what you would see from the MedPAC analysis, which is very clear, is that the supplemental benefits that the Medicare Advantage plans offer to Medicare beneficiaries are basically at an all-time high. And that's largely arisen because the payment to the plans is significantly greater than the payment in the traditional Medicare program. So you can get a really good deal from the point of view of a beneficiary in terms of filling in the holes of the Medicare program. So the concern is that when you pay Medicare Advantage plans more than you would pay for the same people had they been in fee-for-service, the first problem that arises, of course, is you're just spending more money than the federal government is spending more money than they would otherwise spend. So there's a core fiscal concern. Then there's a concern about the imbalance that arises as you pull people in to the Medicare Advantage program and create the growth in that program. The imbalance in enrollment makes setting the benchmarks hard, for example, and doing a whole bunch of other things. So the sort of argument, if you will, for cutting benefits basically flows from a narrative that I would argue was largely the original motivation for Medicare Advantage, which is essentially plans can deliver care, A and B services, probably at a lower cost. And I would argue possibly a higher quality, we could debate that, the evidence is mixed, than the traditional Medicare program would do it. Those savings that the plans generate can be a portion of them, can be funneled to the beneficiaries in terms of better benefits, and a portion could be shared by the Medicare Advantage program. So in that sort of original conceptualization, the taxpayer wins and the beneficiary wins. What has happened for a range of reasons is now the taxpayer is not winning and the beneficiaries, particularly the ones that choose Medicare Advantage, so it's not necessarily equal, it's the 50% of the beneficiaries that are in Medicare Advantage, they actually do quite well by the payment. And so the idea of reducing the Medicare Advantage payment would be to try and get closer to that original conceptualization of the program The beneficiaries would still be doing very well. In fact, the paper we published again in my professor role shows that it is true that some benefits would go away, but not a ton. The plans would adjust in a bunch of ways. So just as if benefits became more generous because we paid a lot, if we paid a little bit less, the benefits would be less generous, but you wouldn't necessarily lose all that much and you would stabilize, if you will, this enrollment balance between fee-for-service and MA to hopefully now allow a better mix, a better balance across the population. You wouldn't be tilting the playing field, if you will, and the government would save some money. And let's go back to discuss how those benchmarks got so high to begin with. That was purely a congressional decision in the earlier days of the MA program to basically make sure that the plans had an incentive to spread across the country, particularly into rural areas, right? So we once upon a time had a policy of setting these benchmarks high. You're saying that now we've overcorrected for that and it's time to readdress that policy. As most things, the we is complicated because there's different people with different orientations here. So let me separate out two things. Actually, let me start with a fundamental point. The Medicare program, the core Medicare benefit package, is not a particularly generous Medicare benefit package. There's a lot of cost sharing required. There's a lot of services, vision, dental, hearing that aren't included. So the Medicare Advantage program, ideally, through the efficiencies it generates, could help fill some of those gaps. In places that were lower spending, it is harder to fill those gaps. And so policymakers made an explicit choice to pay more than fee for service. For example, in some counties, they pay 115% of fee for service. Those tend to be the low spending counties. So they're getting more benefits. Essentially, to equalize the benefit that beneficiaries in those counties would get relative to beneficiaries in, say, much more expensive counties where the MA plans can save a lot more money. So that's one issue which is important. But the issue that's also important about how we got here is it's important to the Medicare Advantage program to adjust for people's health status. And there's an elaborate system for doing so. So while some of the higher payments and higher benchmarks is an explicit policy decision, a lot of it is because of aspects of the risk adjustment model that has enabled the Medicare Advantage plans to get paid more than they otherwise would, essentially because they will record more diseases for a given person than might be recorded in fee for service. And what is the essential mechanism then that is floating those risk adjustment payments back into the benchmarks? Well, I think the plans get paid their benchmarks adjusted for the risk adjustment. So if they can code somebody at a 1.2, they would get paid 120% of sort of where the benchmark is. So the benchmarks, and just to be clear, in some areas the benchmark is 95% of fee for service. But in other areas it's 115% of fee for service. Whatever that ratio is, then gets multiplied by the coding. So if you can capture more disease, you can multiply your benchmark by a higher number. The same, by the way, is true of the quality payments. The quality payments also are above the benchmark. So the plan, you can get paid for performing well on the quality measures that CMS has. It's called the STARS program. Now, you could debate whether those really measure quality. That's a separate topic. But the point is it adds more money to the system and contributes to what plans are getting. So it's the combined nature of all of this, then, that is driving this payment differential, which now MedPAC has calculated as 6% on a per beneficiary basis, that overall the payments are 6% higher. Yes. And that 6% doesn't fully capture something that Medicare calls selection, which is a separate topic, which is for a person with a given set of coded disease, so now two people with the same set of disease codes, there's some evidence that the Medicare beneficiaries are actually health, Medicare Advantage beneficiaries are actually healthier than the comparably coded Medicare fee for service beneficiaries. Again, there's a lot of analytics that goes underneath this that is probably more than we want to get into, but I'm happy to. But I think the bottom line from all of this is if you combine the coding effects, the selection effects, the benchmark effects, and the quality payment effects, you end up with payments in Medicare Advantage plans that are higher than they would be if the beneficiaries were in fee for service, and that extra payment is used to provide more benefits to Medicare Advantage beneficiaries, which is a great deal for the beneficiaries. They get lower cost sharing. They get reductions in their premiums, their Part D premiums, for example. They get extra services, vision, dental, hearing, and a bunch of other things. All of that is how the money is being used. The point of the academic paper that you mentioned is if you reduce that overpayment, some of those things that beneficiaries get will go away, but the plans will adjust in a bunch of other ways so that it's not sort of dollar for dollar. You take a dollar away from the plan, the beneficiaries don't lose a dollar, and that's, I think, the situation we've found ourselves in. Okay. Now let's move to what we have just been through, which is an effort to address one of those factors that you discussed, which is the coding, and the final rate notice that was published. Obviously now we're going to see some major changes in coding over the next couple of years in an attempt to tamp down that issue, at least partially. Let's play that out because it does seem as if we've had independent of, say, fraud and abuse in coding, which obviously is a real phenomenon, there is a separate phenomenon of coding creep over time, I guess, driven by just the reality that if you introduce a coding system, people get better at it over time and get more precise in diagnoses and maybe obviously are incentivized in the MA program to find diagnoses because that's how you try to keep people healthy and out of hospitals and save money from those efficiency perspectives. But are we going to be stuck in this cycle of coding creep and these coding adjustments for all time, or is there another way to address that set of issues? First, I hope I am not talking about coding five years from now. I fear I might be. Let me say a few things about this in general. As you pointed out, there is some aspect of this which is just a straight program integrity issue. But a lot of what's going on might not be coding integrity. It might be that the fee-for-service system just doesn't code people very well and the plans can code them better. Just to be clear, you can code better without getting paid for those codes. So the argument that we need to identify these people with disease, and that's why it has to stay in the risk adjustment model, isn't an argument that's very convincing to me. You can code for clinical care, which is separate than how Medicare pays. That being said, I think the core challenge here is that for a particular condition, for many conditions, it is not cut and dry whether the person does or doesn't have the condition. You might look for it more. Some of the codes might be a little squishy in terms of whether the person really meets the complex criteria for qualifying to have that code. And the coding seminars that you might go to are stunningly detailed and very regulatory in nature. So what CMS tried to do is find some of the codes that they considered. I'm going to use the word squishy. That's not a technical term. You could call them discretionary. There's a bunch of issues. The language is difficult. I apologize for not getting the language right. But codes where they thought there was a lot of discretion in how they were applied and a lot of variation across different plans and between the MA program and the fee-for-service system, and they tried to remove those codes. In part, that was happening because they had to make an adjustment to seeing ICD-9 and ICD-10. So some of it was sort of a technical adjustment of what was going on. So they were really doing two things. They were trying to move from ICD-9 to ICD-10, and they were trying to get rid of some of the codes that they thought were not performing well in terms of consistency between the two programs. It's not a question of accuracy as much as consistency. And so that's what they did, and I think as long as we have a system where plans are reporting diseases and CMS and Medicare is paying people for those diseases, we're going to struggle with this sort of coding issue. The other thing that you didn't mention, which I think is actually increasingly important, is the ability of different organizations to code completely is the word I guess I'll use, varies. So some organizations are quite good at capturing them, and there are certain mechanisms, the health risk assessments, for example, where you can pick up codes, and some people can do that better than others or more completely than others is probably a better word than better. And that creates an imbalance across the plans, and so it actually I don't think behooves anybody to have an inconsistent coding system. I'm sure it behooves the ones that can exploit it the best, but for the Medicare program and for many plans, it doesn't help if we end up spending all of our time trying to win a coding race as opposed to providing better care, and that's what CMS is trying to address. So let's go back to the point that you made, though, just a couple of moments ago, which is that we don't have those same incentives to code in the traditional program, and therefore we don't code at least anywhere near as much in that part of the program, and therefore we don't really know as much about diagnoses among those in the traditional part of the program. How big a problem is that, that fundamentally we just don't know? We know how they're treated, right, because we can see claims, but we don't really know quite as much about the underlying health status of beneficiaries in the traditional program. Yes. I mean, obviously having better information, if everybody could be coded perfectly in non-discretionary, you know, if it was just that would actually be better. Knowing more about health status I think would be better. I don't think it's a huge policy problem per se. It's really a calibration issue in my opinion. So there are certain things you could do, add two years of claims versus one year of claims, for example. That would help. There's other things that help with consistency. They might not help with accuracy, the health risk assessment, for example. So, for example, if fee-for-service isn't doing them and the MA plans are, you get an imbalance, even if the health risk assessments are accurate in some way. I don't think the solution is to do health risk assessment for all the fee-for-service beneficiaries because there's a huge cost of doing that. That's hard to do. I think the solution is to try and find ways to address the imbalance that arises because of the asymmetry of how disease is being captured. And just to be clear, the plans could still do health risk assessments to figure out people's disease and treat them. This is just a question of the imbalance in payment. So what is the answer then? Well, again, I think the answer is doing some of the things that I think CMS will continue to do, which is to refine the set of codes and the underlying models in the risk adjustment model. I think thinking through how we deal with, for example, two years of data is a MedPAC recommendation, addressing the health risk assessments, that that's another MedPAC recommendation, is to not allow those codes to be used, and the GAO has done some analysis there. I think longer run, there's other things one might do about bringing in other types of data or capping the differences to try and get the coding analysis more consistent across the organizations. I don't want to say more accurate, more consistent across the organizations. And then I think for the big picture issues, which is the overpayment, you can adjust payment overall. You can either lower the benchmark sum. In the Medicare program, there's something called the coding intensity factor, which is the SIF, which basically is supposed to adjust for this. The MedPAC analysis that I would agree suggests the SIF is too low, so you can raise that sum. The MedPAC recommendation is that we should lower benchmarks at least 2 percent, which is much below what people think the gap is, to begin to see what would happen. Because understanding getting this right will achieve the laudable goal, in my view, of aligning the payments better. But you will see, again, in my view, modest, but you will see some changes as everybody adjusts. And so I think it behooves us in the MedPAC recommendation to understand that before jumping to a 9 percent payment change or something that I think the analysis might suggest. And that would require a statutory change, correct, by Congress to undertake that kind of a benchmark? The coding intensity factor adjustment could be done by CMS. Broader changes to the risk adjustment model, I think that would require, depending on how broad you want to get that, CMS can change the risk adjustment model how they want. They don't need an act of Congress to change the risk. The advance notice you spoke of didn't need congressional approval, for example. So they can make risk adjustment model changes. If you were going to do other bigger picture changes, for example, lower the benchmarks just directly from the system we have now, that would require congressional action. Let's talk about coming at this from a different direction because there have, of course, been calls to equalize the benefits more between the two programs. Essentially create within traditional Medicare some of those, in effect, supplemental benefits that are in MA, dental, vision, et cetera, but also given our increasing awareness of the need to address individuals' health-related social needs to really get the maximum benefit of the health care that we provide them. Not to put too fine a point on it, but if you're going to have people go to doctors, make sure they can get there through transportation, et cetera, what is the argument in your view for trying to equalize the benefits from that perspective versus trying to tamp down on their availability in the Medicare Advantage program? Well, so just to be clear, if money were no object, a more complete Medicare program would be better. And so that puts us into a discussion, which I'm going to avoid, around taxation and a bunch of revenue kind of things. We have a much bigger problem in this country, problem might not be the word, but there's a much bigger issue to address, which is the demographics of the country are changing in a way that makes financing Medicare more challenging as the baby boomers move on to the Medicare program. That is a bigger, bigger public policy taxation-type issue. But if you weren't worried about that, it would be much better if everybody could have more benefits. So the argument for limitation of benefits in fee-for-service or for Medicare Advantage is really one of a combination of we just don't want to spend the money for these things. And in some cases, we worry that the benefits actually would be overused or abused. So if you look at a benefit, which MedPAC spent a lot of time on telehealth. So telehealth is obviously unambiguously, in my opinion, a set of services that offer tremendous potential. And we saw that during the pandemic. But even post-pandemic, there's certain areas, behavioral health, where new telehealth approaches and digital health more broadly offer unbelievable potential. But there's a lot of concern about the ability to deliver those services efficiently and maintain program integrity. So you have this balance between making sure people have access to what they need and the integrity of the program in financing. Which would be the argument for, and in fact, this is the reality, right? If you're delivering services in a value-based context within a budget, you can make those decisions about how much you want to deliver care in what modality, as opposed to just in a fee-for-service context where you have this volume-inducing phenomenon going on. Absolutely. In fact, I think telehealth is a perfect example. Telehealth is incredibly poorly suited for fee-for-service. Defining the services are hard. The regulatory efforts to try and figure out when someone qualifies for an e-visit versus whether the portal message was related to a past visit, understanding what the units of service should be on remote patient monitoring and how you should pay for the costs in different ways, it is really, really going to be increasingly challenging to fit some of these services into a fee-for-service paradigm. Others, you could do some vision dental hearing services are much more amenable to that, although it is true that one advantage of the Medicare Advantage program is for these services where it's hard to manage excess use or use with low value, the Medicare Advantage plans are much better suited to make the changes necessary to ensure that the services, which are valuable, are used in the situations where they should be. Well, depending on where all of this discussion comes out, then, you could continue to, if you will, unequalize the benefits, right? If MA plans are able to much more freely adopt telehealth because we decide, for whatever reason, that we're not going to extend payment for telehealth beyond the extension, beyond the public health emergency, we could make that situation even worse in terms of unequal benefits, correct? You have to be careful when you add any benefit to the fee-for-service program that you can make sure that the service is used appropriately. That is true. In a world of Medicare Advantage, you may worry about that less. But a lot of these services, there's a separate issue between whether the services is paid for versus whether the service could be provided. So you could use telehealth in MA and fee-for-service. This is really a question of how it's paid for. So, for example, imagine you thought telehealth reduced spending. The Medicare Advantage plan could use telehealth services, could opt to promote them in a range of ways without receiving any extra money for the telehealth service, right? Because that would just be a more efficient way to deliver care. The entire purpose of Medicare Advantage is to – I shouldn't say the entire purpose. One of the purposes of Medicare Advantage is to allow plans the opportunity to find efficiencies in care delivery and pass some of those efficiencies on to beneficiaries. That can be done much more effectively in an organized system like Medicare Advantage or the groups – honestly, the groups that they contract with can do that much more effectively than in a very fragmented, unmanaged fee-for-service system. So that's exactly right. You don't need to have the exact same benefit package. You just have to decide when you put something into fee-for-service how the cost for that in fee-for-service would translate to the Medicare Advantage benchmark and what would happen there. Great. Okay, so to sum up and basically tell those listening to this conversation today what is likely to be ahead in terms of policy around Medicare Advantage, MedPAC, as you said, already has recommended cutting back the benchmarks, not 9 percent. As you said, that would be too far, but 2 percent. See how the system adjusts. We've said that CMS will clearly – you may not want to talk about coding five years from now, but we can pretty well bet somebody is going to be talking about coding, including at CMS. So there will be these ongoing coding adjustments. But we will also continue to see MA program growth in all likelihood. Do you think we can evolve to that kind of a steady state with maybe periodic adjustments downward in the benchmarks, ongoing adjustments in coding? Do you think that will keep the system functioning for long? Well, so first let me say the main MedPAC recommendations, at least 2 percent, so I wouldn't say only 2 percent, but at least 2 percent, do some changes to coding, which we have discussed. And then there's some other things related to the STARS program. I don't think we're going to get sort of in a one single swoop to sort of a steady state. And honestly, if the Medicare Advantage plans can continue to perform better than fee for service if the playing field were level, if they can perform much better, I think we're going to continue to see growth in Medicare Advantage. And that's not going to be solved through any of the things that you said. That's going to need to be solved through bigger program reform. And that's why if your listeners wait for the June report, the Medicare Payment Advisory Commission, MedPAC, is having discussions about what that world might look like. I don't think that's a two-year world, three-year world. But I do think that if the plans can continue to provide A&B services more efficiently once payment is better aligned, and they continue to grow, we will have to think about bigger program things to just bring us back to the beginning. How do we deal with the world in which the Medicare Advantage program is no longer the barn, it's the house, or some version. I can't remember what you said, Susan. I'm sure you said it better than me. But that's going to require deeper, longer, bigger thinking. And beyond the scope of our terrific discussion so far today. But I want to thank you, Mike, for walking us through these incredibly important and complicated issues. Mike, thanks again. Thank you. And we will be able to talk about those other issues later for sure. So thank you. And thank you to your audience for listening.