Blog, The Hospital Finance Podcast®

Should CJR Include Risk Adjustment Factors? [PODCAST]

besler insights blog corner graphic

Hospital_Finance_Podcast small

In this episode, Dr. Chad Ellimoottil, Assistant Professor at the University of Michigan, discusses findings from a recent research study that looks at the potential use of risk adjustment factors in the CJR program.
Learn how to listen to The Hospital Finance Podcast on your mobile device.

Michael Passanante:   Hi, this is Mike Passanante and welcome back to the Hospital Finance Podcast. Today, I’m joined by Dr. Chad Ellimoottil. Dr. Ellimoottil is an Assistant Professor at the University of Michigan.

After graduating with a degree in Economics from Northwestern University, Dr. Ellimoottil worked at the Medicare Payment Advisory Commission. He subsequently completed medical school and urology residency at Loyola University Medical Center.

During his residency, he completed a Master’s degree in Health and Healthcare Research through Rackham Graduate School at the University of Michigan. He’s currently a practicing physician at the University of Michigan and in the Dow Division of Health Services Research. Dr. Ellimoottil’s research agenda is focused on evaluating alternative payment models including episode-based bundled payments.

We’ve asked Dr. Ellimoottil to join us on the podcast today because he is the lead investigator on a recent study that looked at the potential for including risk adjustment factors in Medicare’s Comprehensive Care for Joint Replacement Program. He’s going to discuss their findings and recommendations and share some thoughts on the direction of bundled payments, moving forward.

Dr. Ellimoottil, welcome to the podcast.

Dr. Chad Ellimoottil:   Thank you so much, Michael. I’m very happy to be here. It seems like these podcasts are very informative and I’m glad to be a part of it, part of the conversation.

Michael: We are glad to have you. I saw your study out in the news and I thought you’d be great to come on and talk to our audience who has heard about CJR from us in the past and I think you’ll add some interesting perspectives to that.

Dr. Ellimoottil: Sure, absolutely.

Michael: So let’s jump right in. My first question for you, we’ve covered as I mentioned various facets of CJR on previous podcasts. And for those of our audience who may not be familiar with your study, could you summarize the basis of the study and your subsequent findings?

Dr. Ellimoottil: Sure, absolutely. Our interest with this study was to evaluate a very specific decision that CMS made regarding the CJR program and it has to do with the accounting of the complexity of patients.

Our initial hypothesis or our study question was trying to understand what will be the financial implications on hospitals with Medicare’s current policy for CJR. I hate to say the phrase, but it does not include risk adjustment, but for the most part, it does not have a very robust way of accounting for patient complexity.

The way that we went about answering that question was looking at hospitals in the State of Michigan. This was a simulation, so we’re not looking specifically at the individual CJR hospitals, but we’re looking at a heterogeneous group of hospitals in the State of Michigan.

We ended up calculating their 90-day episode payments for joint replacement. There were about 23,000 patients that were in the study. And then we created a simulation where we looked at the cost of their payments or the cost of joint replacement for two years and then compared that to a simulated performance year.

Essentially what we ended up doing was we were able to calculate the reconciliation payments or as everyone knows, these are the bonuses and penalties that hospitals will receive at the end of the year. We were able to calculate those.

In using that as our substrate, we wanted to understand the impact of patient complexity. And so as step one, we wanted to see if there was any differences in patient complexity, so what we did was we created a variable or a proxy for patient complexity. And we used something that CMS already has that it has used for different purposes called the CMS HCC Risk Score and HCC stands for Hierarchical Condition Category.

Using that as our proxy for patient complexity, we first looked at how much does patient complexity vary across these hospitals. And we found that there was quite a bit of a difference in patient complexity.

And then we created two different simulation models. The first model was let’s set the target prices that are used to calculate our reconciliation payments. Let’s set those using hospitals-owned historical cost. And does patient complexity matter there?

What we found was that it doesn’t and that’s actually what we expected. That simulation itself is consistent with what you see in programs like BPCI, the Bundled Payments for Care Improvement Initiative, which is the target prices are essentially based on the hospital’s own set of patients. And what you find is that because those patients don’t vary from year to year, it doesn’t make much of a difference when you look at the reconciliation payments.

And then the second simulation essentially comparing our target price we used for the second simulation was the regional price. In this case, for this study, we used a state-wide price for CJR as I’m sure most of your audience knows it will be the regional prices based on Census Division, which for Michigan would include the entire Midwest.

Using that target price, we essentially ended up showing that when you’re comparing hospitals across a region, you see that patient complexity does differ quite a bit and it matters. And so step one was just showing that patient complexity does differ and then step two was actually showing what happens to hospitals if you did implement a risk adjusted target price.

What we ended up finding was that essentially I guess the difference is in the eye of the beholder, but substantial gains for hospitals that treat sicker patients and then losses for hospitals that treat healthier patients. It’s sort of re-calibrating that target price to each account for patient complexity. And that’s what we ended up finding.

The policy message from the paper was CMS should consider incorporating risk adjustment into the CJR bundle, but then also to future bundles because my suspicion is that future bundles will probably be based largely on how CJR is designed and we see that with the new cardiac bundle where the structure of it is identical to the way CJR is created.

Michael: That’s a great summary. I appreciate that. And I want to drill into that some more. Would you able to describe some of the factors you included in your calculations related to medical complexity?

Dr. Ellimoottil: Sure, absolutely. That’s a great question. We used, as I mentioned earlier, the CMS HCC Risk Score, which is an off the shelf risk adjustment tool. But there are certainly other factors that you can consider to measure patient complexity, but just to give you a summary of what we used.

The CMS Risk Score includes age, medical co-morbidities. It includes a very rudimentary measure of SES, which was dual eligible status. It also includes original reason for entitlement, so it includes disability status. And those are the main ones, age, sex, co-morbidities, dual eligible status and original reason for entitlement. Those were the variables that were put into creating the HCC Risk Score.

Some of the feedback that we received on this article as well, “Did you go far enough?” I mean even during the peer review process, some of the criticisms we received is that you should have included other variables, including a better representation of socioeconomic status, also functional status.

But really with this paper, what we were doing was showing how even just what’s available off the shelf and easy from administrative claim’s standpoint, non-controversial, I emphasize that really these variables are completely not controversial in terms of affecting or impacting payments and impacting risk for multiple different outcomes. And even using these non-controversial variables, you find signal when you’re looking at episode payments.

Michael: And the study looked specifically at patients in Michigan as you mentioned. Do you think that’s a good proxy for how medical complexity factors might impact CJR nationally or would other information need to be considered on a more regional basis?

Dr. Ellimoottil: That’s a really good question. There are several reasons that we used data in Michigan. Mainly it’s because it was a very robust data set, but Michigan itself has a pretty heterogeneous mix of hospitals and we’ve showed that empirically.

It may be a limitation in terms of we did not do this in national data, but I highly doubt that the findings would change. I think Michigan, for the most part, is a good reflection of the remainder of the United States. There are obviously some areas that will probably be affected more and some areas will be affected less. But I think in general, Michigan would likely be right in the middle.

Michael: This is a broader question for you outside of the study itself.

Dr. Ellimoottil: Yeah.

Michael: There was a study published recently that looked at the effect of socioeconomic factors as they relate to the Hospital Readmission Reduction Program, not CJR. That study found that socioeconomic factors did not greatly affect the readmission rates across hospitals.

Obviously you looked at medical complexity in your study, not socioeconomic factors. But do you think there’s an argument for including some blanket risk adjustment factor across the board in these types of CMS programs to cover a variety of risks? Or is this something that’s going to have to be a program-by-program look?

Dr. Ellimoottil: Yeah, that’s a great question too. I think ASPI is looking at SES for other CMS programs and the verdict is still out there. I heard them present at Academy Health, but that was just some preliminary findings.

I really don’t think that there should be blanket risk adjustment across all programs. I think it should go by a program. Truly, if you ask an economist or a statistician, they’ll tell you the same thing that you should go program by program because different programs have different outcomes. And the whole point of risk adjustment is to find risk factors that predict your outcome.

For payment like CJR, they may be different than what you would expect or the risk adjustments you would use for readmission or for whatever else or use of post-acute care or whatever other outcome you’re interested in looking at or mortality for example.

But there’s going to be a significant amount of overlap. We know that medical co-morbidities, for example, there will be significant overlap between what predicts cost and what predicts mortality and what predicts readmission.

I saw that paper by Dr. Krumholz’s group and I think it’s interesting because it adds to the national conversation on a very philosophical argument about using SES. Essentially one side of the story is don’t use SES because you wash away or you give a pass to hospitals for important factors that they should be considering and they are basically given the ability to perform, to have lower performance because of that specific patient population. But then on the other hand, we know that some of the factors that go into measuring SES are highly predictive of readmission, so maybe they should be included.

I think that obviously from a methodological standpoint, it’s not difficult to include it. So that philosophical argument is going to have to be disentangled at a much higher level – from a policy level.

Michael: Yeah. It’s great feedback because the idea of including these risk adjustment factors, whether they be socioeconomic, medical complexity or otherwise, it’s a hotly debated issue. And really the question is, is it something that you put out there?

Dr. Ellimoottil:  Yeah.

Michael: Is it a one size fits all or not? I see where you’re coming from on that. It makes sense.

Dr. Ellimoottil: And just to add a little bit to that, when you’re thinking of what variables to include in a risk adjustment model, you have to think about what is a variable that’s predictive of outcome that you think hospitals should get a pass on. That’s the way that I think about it.

And what I mean by that is should you pay a hospital that treats patients for joint replacement. And one hospital has an average of 65 and the other one has an average age of 40.

Intuitively, we know that the hospital that has an average patient age of 40 is going to have lower costs because patients that are 65 or 75 are going to require post-acute care. They’re going to require more care at home. They’re going to likely have more co-morbidities and end up coming back to the hospital more commonly.

We know that’s going to happen. Should you penalize a hospital that treats older patients? In that case, age would be an appropriate risk adjustment variable. You can give hospitals a pass on treating older patients because they are expected to be more expensive. And you shouldn’t penalize hospitals for treating those types of patients I think.

By not risk-adjusting for age, they could potentially lead to access issues where hospitals may decide not to treat older patients. And so I think in the grand scheme of risk adjustments, age for example is a very non-controversial variable.

Co-morbidities, for the most part are non-controversial. I mean if you look at the readmission program, the variables that Dr. Krumholz’s group created at Yale, a lot of them actually include co-morbidities. Almost all of them include co-morbidities. I think co-morbidities are also non-controversial. Age, co-morbidities, sex is not controversial, but then you get controversial when you start thinking about socioeconomic factors.

To go back to your original question on whether you should include a blanket risk adjustment, I don’t think any risk adjustment program should be blanket, but I do think that there are some variables that will end up being important for almost every program—age, sex and co-morbidities.

And that’s the whole point of this paper we’re showing. We looked at very, very basic risk adjustment variables and we found signal there. That’s why we think that that should be included. At least that minimum risk adjustment should be included in the CMS program.

Michael: Clearly, CJR is not going to be the last of these programs. You mentioned earlier the Cardiac EPM, which is out now, covers acute myocardial infarction and coronary artery bypass graft. And they’ve already expanded CJR to include surgical hip and femur fracture treatment. On a fairly new program, they’ve already expanded it.

Dr. Ellimoottil: Yeah.

Michael: This is the direction they want to go in. Do you have any thoughts on other models that they may introduce in the short term?

Dr. Ellimoottil: Yeah. I think that joint replacement and the cardiac models, everyone expected these to come out because they were the ones that Medicare had the most experience with.

I don’t know where they’re going to go next. I’m not sure if they’re going to include other medical conditions like CHF, pneumonia. I think they’ll end up using conditions that probably have the most success in BPCI.

Spine surgery, for example, was a procedure that a lot of hospitals picked. Obviously BPCI, you’re allowed to pick your conditions and spine surgery was one that many hospitals picked.

If I to predict where CMS is going to go in terms of additional conditions, I think that they’re probably going to pick the ones that were most successful in BPCI. Spine surgery is there and CHF, pneumonia. Those were conditions that may or may not have been successful, but at least they were picked by a lot of hospitals, so there’s some data to guide CMS as they design those programs.

But if you look at the cardiac program, at least the proposed rule is almost carbon copy in a lot of ways with CJR. I think the future programs are probably going to be very similar too with some minor tweaks here and there. But for the most part, they’ll be very similar.

Michael: And one of the things that I guess to remember at this point is that these are pilot programs.

Dr. Ellimoottil: Yeah.

Michael: That leads to my next question because the original pilot length of CJR was going to take us towards the end of 2020. Do you think there’s a possibility that they might curtail that timeline and establish it as a permanent program if they see that the results are panning out in a favorable way?

Dr. Ellimoottil: Yeah. I’m very confident that they’ll end up making it a permanent program because it’s a program that can’t fail. What I mean by that is that even if you’re not seeing significant savings from these programs, there are going to be significant changes that occur over five years in these hospitals in terms of care redesign.

Hospitals, never before, were really thinking about what goes on behind their four walls and I think this is forcing hospitals in a way to critically think about the patient in a 90-day unit. I think that there’s going to be a lot of spillover success just by implementing these programs.

The only way that it could fail miserably is if we get into year three, four and five and hospitals are going bankrupt from it. And that’d be a scary thing and I don’t know if that’s going to happen or not. But I think that there are pretty aggressive pricing changes that occur in year three, four and five when you go into regional based pricing and it’s yet to know what exactly is going to happen to those hospitals if they can actually keep their doors open with such wild swings and what’s expected from them for a joint replacement. But on the other hand, there are going to be hospitals that will end up doing very well for not doing much just because of their baseline costs.

We’ll see how those initial results pan out, but I think CMS made it clear that they want to expand alternative payment models. It’s built into MACRA. Physicians will end up doing better or getting more bonuses if they’re part of the alternative payment model route, so I think that CMS has a strong incentive to expand these programs.

Michael: Clearly, as you mentioned, hospitals are dealing with this now and they don’t have the benefit of a risk adjustment factor on CJR at the moment.

If you’re thinking in terms of hospital administration. What do you think the hospitals should be doing now to account for the potential of higher costs that could be associated with patients who have more complex medical conditions?

Dr. Ellimoottil: I think that the first step, and this is what we see going on with a lot of CJR hospitals, is that for some hospitals for the first time, they’re digging into their data. Now is the time to become very analytically savvy.

For these higher cost patients, the simple thing for hospitals to do would be to not treat those patients, but that would be a horrible thing. We wouldn’t want that to happen at all.

But maybe understand the root causes of these higher cost patients. If one of the root causes that your individual hospital is that the patients who are sicker come back to the hospital have higher readmission rates, try to figure out why that happens. It’s not because they’re sick. It’s not because they’re poor. Poor is not a reason to come back to the hospital. There’s got to be a root cause to it.

If they’re poor or they have a lower income and they can’t afford transportation to go to their follow-up appointment then perhaps creating an Uber program. That’s what we see actually in hospitals in Michigan where they’ve created transportation programs because the cost of a readmission in a bundled payment environment is much more costly than a $20 Uber ride.

Hospitals will start to become creative, but it all starts with identifying root causes. And that’s something that we’re doing in the State of Michigan. There’s a collaborative that I’m a part of called the Michigan Value Collaborative where we look at strategies, for example, for CHF to reduce CHF readmissions. It all starts with identifying what is the exact root cause. And a patient characteristic, in and of itself, there’s something that’s underlying with that patient characteristic that causes the readmission or causes the longer length of stay in post-acute care.

Without the benefit of risk adjustment, you really have to dig into the root causes and figure that out. You have to figure out what skilled nursing facilities. For skilled nursing facilities, it’s more complex than, “Yes, no. Go to a skilled nursing facility. Don’t go to a skilled nursing facility.” You have to dig deeper and look into intensity of skilled nursing facility care.

When your patient goes to a skilled nursing facility A, is the skilled nursing facility billing at the highest RUG or resource utilization group? And are they billing for twice as long as skilled nursing facility B? If that’s the case, then you have to approach that skilled nursing facility and say, “What are you doing that justifies this additional cost?”

Again, these programs are really causing hospitals to look outside of their own four walls and communicate with providers that they may have never communicated with before, depending on the hospitals. There are some hospitals that are very integrated. They’ve been doing this for 10 years, but I think that for newer hospitals that are part of these mandatory programs they’ll have to do that.

Michael: Clearly, times are changing for hospitals and they have to think differently and innovate differently. Thank you for your time today and coming by and shedding some light on these issues and providing, I think, a really unique perspective on CJR and bundled payments in general.

Dr. Ellimoottil: Sure, absolutely. I was very happy to be here.

Michael: Dr. Ellimoottil, thank you so much.

Dr. Ellimoottil: All right. Take care.

SUBSCRIBE for Weekly Insider Updates

  • Podcast Alerts
  • Healthcare Finance News
  • Upcoming Webinars

By submitting your email address, you are agreeing to receive email communications from BESLER.

BESLER respects your privacy and will never sell or distribute your contact information as detailed in our Privacy Policy.

New Webinar

Wednesday, December 14, 2022

live streaming

Partner with BESLER for Proven Solutions.