Subscribe
Share
Share
Embed
Each week, Health Affairs' Rob Lott brings you in-depth conversations with leading researchers and influencers shaping the big ideas in health policy and the health care industry.
A Health Podyssey goes beyond the pages of the health policy journal Health Affairs to tell stories behind the research and share policy implications. Learn how academics and economists frame their research questions and journey to the intersection of health, health care, and policy. Health policy nerds rejoice! This podcast is for you.
Welcome to A Health Podyssey. I'm your host, Rob Lott. Like most public health programs, Medicare Advantage is a work in progress. That is, public officials and private stakeholders are constantly tweaking its rules and regulations, adapting its adjustments and adjusting its adaptations. This makes sense.
Rob Lott:Most first attempts at policy intervention are imperfect. There's always room for improvement. And besides, the world and the people in it are also constantly changing. And so too must the programs that serve them. Consider, for example, the tools that Medicare Advantage uses for risk adjustment, the mechanism to ensure that plans are adequately paid for covering the most costly beneficiaries.
Rob Lott:The main tool is essentially an algorithm that produces a risk score based on patients' diagnoses. Since CMS began risk adjustment in the early 2000s, the agency has adjusted that system at least half a dozen times. This aspiration to improvement and refinement is a good thing, but it doesn't guarantee that any given change will automatically have the desired effect. And so when a change is implemented, it's worth following up with questions like, did it work the way we planned? And if not, why not?
Rob Lott:Fortunately, these are the questions, we get to ask on today's Health Podyssey. I'm here with Niru Ghoshal-Datta, a PhD student at the University of California, Berkeley. With coauthors, he has a new article in the September issue of Health Affairs. Its title is also one of its main findings. Quote, new Medicare Advantage risk adjustment model had limited effects on premiums and cost sharing, but small effects on benefit generosity.
Rob Lott:I really like this paper, and I can't wait to learn more about it. Niru Ghoshal-Datta, welcome to A Health Podyssey.
Niru Ghoshal-Datta:Thanks for having me on, Rob. It's great to be here.
Rob Lott:Great. Well, let's just jump right in. Your paper, as we said, sort of looks at the impact of a change that CMS made to the Medicare Advantage risk adjustment algorithm. And the goal of that change, according to CMS, was to, quote, make the model less susceptible to discretionary coding. And I think that's probably a good place to start.
Rob Lott:Can you tell our listeners a little bit about discretionary coding? What is it and why is it a potential problem?
Niru Ghoshal-Datta:Sure. So I think when CMS is referring to, discretionary coding here, they're sort of broadly referring to cases where MA plans have some influence over what gets coded or documented in a patient's medical history. Now, and the reason this is an issue in MA is that the government pays, the MA plans more for covering sicker beneficiaries, so the plans have an incentive to code more to make their patients potentially look sicker than they actually are. Now, in some cases this could lead to coding that's more aggressive or even fraudulent on the part of MA insurers, but in other cases it might just, incentivize them to code more accurately. But in either case, what the effect of this is, is that there is higher coding in MA than there is in traditional Medicare, the public program, And this results in payments being higher in MA than in TM.
Niru Ghoshal-Datta:And this is one of the main reasons why it ultimately costs the government more to insure beneficiaries in MA.
Rob Lott:Okay. That makes sense. What specific changes were made in this new version of the model? I realize these are complex algorithms and there's a lot of detail here that we probably don't want to get too far into the weeds, but can you kinda give us a brief summary of, you know, what's new here?
Niru Ghoshal-Datta:Right. So CMS removed and reclassified a lot of the codes that were most susceptible to the sort of discretionary coding that you mentioned. So in some cases where they found that codes were differentially coded in the MA as opposed to TM, they eliminated the codes entirely, and in other cases, they got rid of levels of severity to these condition categories so that MA plans couldn't up code the severity of certain conditions. Now there are also some other tweaks to the model. They ran it on more recent data from traditional Medicare and this lowered the weights for some of the conditions.
Niru Ghoshal-Datta:But in aggregate, the effect of all of these, changes was to lower payments, to the MA programs, but that affected plans differently based on how much they actually use the codes that were affected in the new model. So that's the sort of variation that we leverage in our study to measure exposure to the model and exposure to those payment cuts.
Rob Lott:Got it. I want to zero in on one point you made, which was sort of eliminating these sort of severity levels of coding. And so I don't know if you, if there's an example, maybe you can point us to, but the idea that maybe there might be a, you know, a sort of a low, medium and a high version of, you know, respiratory disease and health systems were kind of defaulting to the higher level in order to, you know, ramp up the, you know, the risk adjustment there. And so by just doing away with it, they kind of remove that incentive. Is that sort of what's going on there?
Niru Ghoshal-Datta:Yeah. So I was going to say, to give an example of one of those conditions where they removed those levels of severity, An example is a diabetes category. So, diabetes had multiple sub codes within the HCC of diabetes, and, one of the codes was diabetes with chronic complications, and one of them is diabetes without complications. Now, CMS found that MA frequently coded the more severe level, of diabetes with complications because it received higher reimbursements for that code rather than the code without chronic complications. In the new model, they got rid of those levels of severity entirely, at least for some of those, subconditions.
Niru Ghoshal-Datta:And this sort of, lowered the ability of MA plans to have discretion, in what level of severity gets coded.
Rob Lott:Got it. Okay. So before this change took effect, did CMS make any, you know, projections or predictions about how the change would affect risk scores and ultimately costs?
Niru Ghoshal-Datta:Yeah. So back in 2024, when the model was first being implemented, CMS expected that it would reduce risk scores by about two percentage points and result in $7,000,000,000 of savings, the Medicare trust fund in that year alone. And the model was phased in over the course of three years from 2024 through 2026. And an additional third of the model was being factored in each year to calculate risk scores and calculate payments. So by 2026, when the model was fully implemented, it likely resulted in substantial payment reductions to the MA program and also substantial savings to CMS.
Niru Ghoshal-Datta:So, we tried to look at the impact of what those, payment reductions were on plan benefit offerings.
Rob Lott:Great. So as you say, you set out to estimate the sort of actual effects of the new model on premiums, cost sharing, also supplemental benefits that we know MA plans can provide and use use the sort of additional payment to to pour into or to fund those supplemental benefits. So what were some of your top line findings?
Niru Ghoshal-Datta:Yeah. So I think some of the key measures that we were looking at were really premiums and cost sharing. So we wanted to see whether plans passed, these payment cuts through to consumers in the form of higher premiums or cost sharing. And we found that they didn't for the most part, and that premiums and cost sharing were, largely unaffected, but we did find some effects on, the supplemental benefit offerings. So there's a wide variety of supplemental benefits that MA plans can offer that aren't included in the standard Medicare benefit package, and that can range, from things like dental, vision, hearing coverage to meals and transportation.
Niru Ghoshal-Datta:And, we looked at whether, these payment cuts affected the availability of these supplemental benefit offerings. And we found that there were reductions in some of these categories. So there were reductions in hearing coverage, I think meal coverage, gym memberships, and some other categories as well. And the effects on any of these individual categories was relatively small, but there's variation in what was reduced across which plans. And when you aggregate, all these reductions across all these benefits, they might end up aggregating to something that's more noticeable.
Niru Ghoshal-Datta:So I think our takeaway here was that plans didn't affect premiums or cost sharing, but they sort of reduced these supplemental benefits at the margin.
Rob Lott:Gotcha. And were any of these findings surprising to you? Were you expecting to see basically these results when you went into the research, or, how are you sort of, responding or reacting to those initial findings?
Niru Ghoshal-Datta:Yeah. I think they're broadly in line with what we would expect. I think we would expect MA plans to be able to absorb some of these payment cuts. And then if they were to pass through some of this beneficiaries, it makes sense that they would reduce these supplemental benefits before they raised premiums or cost sharing, which are typically more salient to consumers when they're shopping for MA plans in the MA market. So I think this isn't broadly in line with what we would expect, especially considered that these were relatively modest payment cuts, relative to the scale of differential payments in the MA program.
Niru Ghoshal-Datta:So we wouldn't expect large effects to begin with.
Rob Lott:Gotcha. And, you know, in general, the MA market is a competitive market, right? And that's sort of one of the appeals of the program is to sort of apply the force of competition to compel plans to improve their offerings. And I'm wondering if you have a sense of whether or not the changes differed in more competitive markets or not.
Niru Ghoshal-Datta:Yeah, that wasn't an explicit part of our analysis. I think we did some supplementary analysis of this and we didn't find much variation, but I don't think that was ultimately concluded in the paper. But, I think one thing that's important to note is that, sort of the competitive dynamics here, might differ from, previous payment changes. So most of the literature in this space has looked at the impact of, changes in county level benchmarks, which affect an entire MA market. And in this study, we looked at changes that affect an individual plan.
Niru Ghoshal-Datta:And I think that that can result in different competitive dynamics where a plan might not want to reduce their benefits if their competitors aren't doing the same thing. So, it might, sort of attenuate those effects in that way. We generally found that there weren't many differences. I think some of my co authors had previously done studies on those benchmark changes and they similarly found modest reductions in sort of the supplemental benefit categories. So we find that it's broadly consistent, and the competitive dynamics don't really differ too much here.
Rob Lott:Great. Well, that sort of leads me to another question, which is kind of thinking about other potential sort of secular trends that may have been at play here, and whether or not there have been factors other than this specific change that you were studying. Were you able to sort of separate those out and, how did that inform your thinking here?
Niru Ghoshal-Datta:Yeah. So we tried to isolate the effect of the risk adjustment payment changes in particular. So we controlled for any changes to county level benchmarks, which are those, you know, base rates that MA plans get paid. And then we also controlled for changes to star ratings, and that affects the quality bonus payments, which also changed in this time period. And then, outside of those factors, we controlled for just general year fixed effects, which controls for any secular trends that are common across all MA plans in a given year.
Niru Ghoshal-Datta:But there might have been other factors that could have differentially affected plans, such as increasing medical costs or the implementation of the IRA could have affected some of these Medicare Part D benefits. So those things were all happening in the background and they could have differentially affected plans, but they shouldn't affect our results unless they're somehow correlated with Mhmm. V 28 exposure.
Rob Lott:Got it. Okay. Were there any sort of big unanswered questions, you know, that maybe if you were able to apply the methods differently or if you had access to different data that, you'd love to answer, but you perhaps weren't able to hear, but maybe you still remain curious about, are there any of those sort of unanswered questions out there you'd like to perhaps explore in the future?
Niru Ghoshal-Datta:Sure. Yeah. I think there were definitely limitations to our analysis, in terms of what we were actually able to capture, in the benefit data that we had. So we were able to observe premiums and cost sharing, and then we were able to observe the availability of these supplemental benefits. But for most benefits, we weren't able to observe the actual generosity of what plans cover in those categories.
Niru Ghoshal-Datta:So take for instance, for the dental benefit, we didn't find that benefit to be reduced, but there could have been reductions in how much dental coverage plans actually cover, and we wouldn't be able to observe that in our analysis. So that would definitely be something to look out for in future analyses. There's also a range of other strategies that MA plans could use to potentially cut costs that we weren't able to observe. So plans could have narrowed their networks by excluding more providers from their networks. It could have been more aggressive about prior authorization or claims denials, and we wouldn't be able to observe any of that in the data that we had.
Niru Ghoshal-Datta:But it would definitely be interesting for future research to look into. And then in terms of takeaways and policy implications from the findings that we did have, I think our findings sort of suggest that supplemental benefits are likely to be hit the hardest when payments are cut to MA. And I think there needs to be a lot more work done to quantify the value of these supplemental benefits to both consumers and policymakers. And I think that's, definitely an area for future research and policy discussion to look into.
Rob Lott:Great. Well, good marching orders for future, research and future researchers, looking for for, additional study. Niru Ghoshal-Datta, thank you so much for taking the time to be with us here today. Thank you for your work on this paper, which readers can find in the September issue of Health Affairs. It's really important to track these changes after they're implemented and see what what kind of impact they have.
Rob Lott:So thank you for doing that, and thank you again for being here today.
Niru Ghoshal-Datta:Thank you for having me.
Rob Lott:And to our listeners, thanks for tuning in. If you enjoyed this episode, leave a review, recommend it to a friend, smash that subscribe button, and of course tune in next week. Thanks everyone.