Skip to main content

Medicare Advantage is growing in popularity, but the program is also costly to taxpayers. 

According to analysis from Kaiser Family Foundation, enrollment in Medicare Advantage (MA) plans have doubled from 2011 to 2021. The plans are popular among beneficiaries, according to a survey from the Better Medicare Alliance. The research found universal satisfaction with Medicare Advantage’s coverage (98 percent), provider networks (97 percent), and handling of the COVID health crisis (98 percent).  

But Richard Kronick, PhD, Professor in the Department of Family Medicine and Public Health at the University of California, San Diego and a former Director of the Agency for Healthcare Research and Quality (AHRQ), is of the belief that Medicare overpays for these plans—and not by a small amount. According to Kronick’s own analysis, Medicare overpaid for MA plans by a total of $106 billion from 2010 to 2019.  

Kronick spoke with Health Evolution about how he arrived at this number, why there is increased scrutiny over Medicare Advantage plans, and how CMS could regulate the program better to reduce the amount these plans are being overpaid.  

In your analysis of Medicare Advantage how you determine that CMS is overpaying by $106 billion?  

Medicare pays Medicare Advantage plans based on the risk scores of Medicare Advantage members. Risk scores use diagnostic and demographic information to figure out how much care each beneficiary is likely to need. The risk scores start with the information that comes from fee-for-service Medicare beneficiaries. The average beneficiary in fee-for-service has a score of 1.0 construction. The diagnostic information for these risk scores for Medicare Advantage members gets reported by Medicare Advantage plans. So plans that have more people with cancer, heart disease, and serious illnesses get paid more money than plans with fewer of those beneficiaries. That’s how Medicare makes this work and tries to create a system where plans are not unhappy to have sick people.   

When this system was created, in the mid-2000s, Congress very quickly realized that the diagnostic information that was reported by plans would likely look different from the diagnostic information that was reported in fee-for-service. Congress realized if we create a system where plans get paid based on how sick the members are, there’s probably going to be a lot more quadriplegics written down. In 2005…Congress directed CMS to implement a coding intensity adjustment to adjust for the differences in coding between MA and fee-for-service. The purpose of this is to say that if you have a beneficiary who had a risk score of 1.0 in fee-for-service and she goes into MA and has a risk score of 1.1, the plan should get paid as if she were 1.0 because the system is calibrated based on fee for service. In 2010, CMS implemented a coding adjustment of 3.1 percent. In the Affordable Care Act and subsequent legislation, Congress directed CMS to enact statutory minimum adjustments so that in 2014 the adjustment had to be at least 4.4 percent and in 2018, 5.9 percent. CMS has the authority and the obligation to implement an adjustment larger than the statutory minimum if it determines that a larger adjustment is appropriate. But it has never yet implemented an adjustment larger than the statutory minimum.  

I did an analysis in trying to figure out how big the adjustment should be. I used a pretty simple approach based on the comparison of two numbers. One number is the relative risk of all Medicare Advantage to the risk of fee-for-service beneficiaries when that risk is measured using the diagnostic and demographic approach used to pay plans. The average risk score for Medicare Advantage beneficiaries is much higher than the average risk score for fee-for-service beneficiaries. And it has been increasing a lot since 2006. The numerator of this measure of coding intensity is the relative risk of MA using diagnostic and demographic information. The denominator is the relative risk of MA just using demographic information. Prior to the implementation of this risk adjustment system, the risk adjustment just used to be age, gender, Medicaid and institutional status to measure risk. Part of why Congress told CMS to use diagnostic information is that in the late 1990s and early 2000s there was very strong evidence that the demographic system overpaid MA plans because it was clear that MA was getting favorable selection and that healthy people were going into MA. There’s less favorable selection now than there was in 2000. In 2000, there were few in MA plans and now there’s close to 40 percent of eligible beneficiaries in MA plans. But there’s some evidence that MA is still maybe getting a little bit of favorable selection. If you’re really sick and you’re in fee for service and you’re using a lot of care, you’re not likely to join Medicare Advantage. There’s some evidence that some really sick people in MA switch back into fee-for-service. It’s probably still a little bit, a favorable selection but not too much on average. 

Through my analysis, I saw that the coding intensity adjustment should have been 20 percent, which is much higher than the 5.9 percent that CMS actually implemented for 2019. The coding intensity adjustment was increasing more than a percent per year over the previous 14 years. The difference between the adjustment that CMS implemented…and the adjustment I estimated using the demographic estimate of coding method results in the estimate that CMS overpaid by $106 billion over the decade.  

Taking advantage is a pejorative phrase. I’d say insurance companies are responding to the incentives that the system has created. Some might describe that as taking advantage. Others might say those are the expected consequences of the incentives.

Richard Kronick, PhD, University of California, San Diego

You’ve analyzed these plans for years, but lately there is a lot of scrutiny toward MA. Why is more controversy following these Medicare Advantage plans?  

There are couple of reasons. One is that as the number of beneficiaries in Medicare Advantage plans grow, the payments to Medicare Advantage plans become a bigger and bigger part of the overall Medicare program. People are concerned whether or not we are paying the right amount. If Medicare Advantage is smaller and its payments are too large, it’s not as big of a deal. As Medicare Advantage gets bigger and bigger, it becomes a much bigger deal. 

A second reason is that there has been an increasing number of whistleblower suits against the Medicare Advantage plans. The Department of Justice has joined in on a few. These suits are alleging a fraudulent behavior on the part of Medicare Advantage plans and that kind of behavior gets attention and interest.  

I want to emphasize that the work I’ve done is unable to differentiate between fraud and perfectly legitimate behavior on the part of Medicare Advantage plans. My suspicion is that a large part of these increases in risk scores that we have seen is not the result of fraud. It’s the result of legitimate behavior. Reading the DOJ complaints, it does look like there is fraudulent behavior at a variety of Medicare Advantage plans, but that’s not what I’m getting at. My work is trying to emphasize that CMS has not done what it should do in setting payments. CMS is given the statutory authority and obligation to promulgate coding intensity that adjusts for the differences in coding between Medicare Advantage and fee for service. It has not done that.  

So, are health plans taking advantage of the system or using the system as it’s designed? 

Taking advantage is a pejorative phrase. I’d say insurance companies are responding to the incentives that the system has created. Some might describe that as taking advantage. Others might say those are the expected consequences of the incentives. The courts will help us understand the line between legitimate and illegitimate behavior. There are areas where it’s not clear what is legal and what is not legal, but there’s certainly Medicare Advantage plans that are doing all they can to maximize the risk scores of their beneficiaries.  

The actor in the system that is not doing what’s intended is CMS. Or at least, CMS is not doing what’s written down in the law. CMS is told to adjust for differential coding, and it has not done that. Arguably, although the law tells CMS to do that, it may not be what Congress really wants to it do and CMS maybe has not fully adjusted for differential coding in part because it understands that there’s many members of Congress that don’t want it to fully adjust.  

Is it as simple as CMS making a more appropriate coding adjustment or are there other things that CMS can do to monitor this problem?  

I think the biggest challenge is for CMS to make an appropriate adjustment. What’s most needed is for Congress to make a clear statement in statute that CMS should adjust appropriately and put in a method of calculating the adjustment. Part of the problem is that…the last time CMS has said anything official about how to measure coding intensity was in 2010, which is a long time ago. I was a consultant to them and helped them develop that method and it was a reasonable way to start, but it’s far from an optimal method. There’s disagreement on how differential coding should be measured and I think if Congress were to wade in here and say what’s reasonable, it would remove any discretion from CMS about how to measure code intensity. Any time CMS has discretion, the pressures are for it to use that discretion in a way that will minimize the coding intensity.  

There are other things that CMS could be doing, such as more carefully and consistently auditing the behavior of Medicare Advantage plans. The Medicare Payment Advisory Commission has proposed for many years that diagnoses that Medicare Advantage plans collect in home-based health risk assessments be excluded from the calculation of the risk scores. That’s sensible and would help a little bit. But the problem is even with these kinds of actions, there’ll still be very big differences in coding between the two sectors. CMS needs to adopt a coding intensity adjustment that accounts for those differences.  

How will this play out in 2022 and beyond? 

The potential forcing event is the need to do something to make sure that the hospital insurance trust fund is not exhausted. The most recent estimates from the Medicare trustees and from the Congressional Budget Office is that the trust fund will be exhausted in 2026. Congress will not let that happen. I’m loathe to predict the actions of politicians but I’d be quite confident that Congress will not let the trust fund be exhausted. It will take some action to keep that from happening and those actions could include raising the revenues that go into the trust fund or reducing the expenditures going out of the trust fund or some combination of those things.  

We might see some changes in payments to hospitals, but the opportunities for savings are likely to be somewhat limited concerning facilities. You might see some changes in payments to skilled nursing facilities, but they were seen as heroic during the pandemic and any proposed changes to them will be tough and a relatively small part of the total anyway. So, changing the payments to Medicare Advantage plans is an opportunity to try and make sure that the trust fund is not exhausted.