AHIP's hooray for Medicare Advantage

Note: In today's New York Times Upshot, Reed Abelson and Margo Sanger-Katz take a deep dive into lawsuits and Inspector General reports detailing the way the largest Medicare Advantage insurers have inflated their payments from the federal government by gaming risk adjustment, the process by which they are compensated for patients who need more care than average. Abelson and Sanger-Katz note that this fall, CMS must decide whether to finalize a rule, first proposed in 2018, that would penalize upcoding more severely (indeed, the article seems designed to stiffen CMS’s spine). The AHIP-commissioned report (and AHIP’s spin on it) described below, purporting to show that MA plans provide superior value to the federal government than does traditional Medicare, should be understood in that context.

The industry euphemism for the systematic inflating of risk scores is “coding intensity” — long recognized by the Medicare Payment Advisory Commission (MedPAC) as a source of MA overpayment. See this post for an overview of MedPAC’s take on upcoding and what to do about it.

* * *

Bob Herman of Stat casts a cold eye on a new claim from AHIP that Medicare Advantage provides superior value to the federal government:

America’s Health Insurance Plans, the industry’s primary lobbying group, funded a new report that was conducted by actuaries at Wakely Consulting Group. AHIP claims the report proves Medicare Advantage...is “saving Americans billions of dollars every year.” The actuaries, however, never use that language in the report.

STAT spoke with several independent Medicare policy experts, all of whom said AHIP’s report was incomplete at best and refuted by other studies that analyzed the same data. 

AHIP's press release asserts that "in 2019, rather than being 2% more expensive than original Medicare, on an apples-to-apples basis, average MA spending was actually about 7% lower than original Medicare."

That conclusion is based on two claims grounded in the Wakely analysis: 

---

1) If you were to add an annual out-of-pocket cost cap to traditional, fee-for-service (FFS) Medicare (as Medicare Advantage plans must provide), the per-person cost would increase by 3.5%.

2) An "apples to apples" comparison of per-person costs in Medicare Advantage (MA) and FFS Medicare requires excluding from MedPAC's per-person cost calculation for FFS Medicare those enrollees who are enrolled only in Part A or only in Part B, rather than in both parts. Single-part enrollees in Part A in particular spend significantly less in Part A than enrollees in both parts. The Wakely report calculates that excluding single-part enrollees from the per-person FFS cost estimate would raise the estimate by 5.9%.*

Taking those two claims at face value, AHIP claims, means adding 9.4 percentage points to MedPAC's 2019 per-person cost estimate for FFS Medicare. The March 2019 MedPAC report pegged MA payments at 101% of FFS Medicare. Hence AHIP's claim that MA spending was 7% lower than FFS spending. (In 2022, MedPAC estimated MA costs at 104% of the cost of providing the MA population with FFS.)

Let's look at the two findings in turn. The first is a contrary-to-fact. The second is more arcane, ambiguous, and potentially impactful. 

"If you only had an OOP..."

In response to the first claim --  that adding OOP cap to FFS Medicare would all but wipe out the per-person spending difference with MA --  Jeannie Fuglesten Biniek, a senior Medicare analyst at KFF, told Stat's Herman, “That’s sort of a different point than when you look at actual spending. We’re still paying Medicare Advantage plans more. Maybe you’re getting more for it, but that’s a different question.”

That "different question" is worth asking. If MA costs 104% of the cost of FFS per person (MedPAC's 2022 estimate), and that extra 4% includes an OOP cap and, in total, $2,000 worth of average extra benefits by MedPAC 's 2022 estimate, MA does have a fair claim to offer savings to the federal government. 

The question is, should the federal government be incentivizing enrollees to opt for those extra benefits in exchange for a limited provider network, prior authorization, and higher risk of coverage denials? Do those 'managed care' techniques constitute a legitimate source of savings?

Put another way, would the extra money the federal government lavishes on MA be better spent improving benefits in FFS, thereby reducing MA's growing funding advantage?  On this front, AHIP has in a sense undercut its own argument. MedPAC has concluded in several annual reports that the payment formulas for MA are on various fronts too generous, and that the efficiencies MA does attain (spending about 85% per member of what FFS Medicare spends) should be constrained within payment at par -- 100% (or less) of the cost of FFS Medicare. If that payment discipline were were achieved, the extra 4% paid to MA plans could fund the FFS OOP cap, according to Wakely's calculations. That in turn would negate a key advantage held by Medicare Advantage. In fact it would reverse the payment ratio — MA would be paid 100% of current FFS costs, while FFS costs rise to about 104% of present costs. But the original premise of MA was that it would save a few percentage points of net spending.**

Apples to (risk adjusted or not) apples.

On the second issue, Wakely takes issue with a MedPAC calculation that has a potentially dramatic impact on the estimated per-person cost of FFS Medicare vs. Medicare Advantage. 

To calculate the per-person cost of FFS Medicare as a basis for MA plan benchmarks, MedPAC includes all FFS enrollees, including those enrolled only in Part A or only in Part B. While those enrolled in Part B only incur Part B costs comparable to those incurred by both-part enrollees, Part A-only enrollees spend far less in Part A than both-part enrollees. MedPAC notes that Part A-only enrollees are generally either employed and still enrolled in employer-sponsored insurance, or unwilling the pay the Part B premium, likely because they have high incomes and so pay higher income-adjusted premiums (as high as $578/month in 2022, while 95% of enrollees pay $170/month). Others are legally present immigrants who lack the required U.S. work history to qualify for free Part A coverage, and buy in. All of these groups are likely to be healthier than average FFS enrollees.

Back in 2017, MedPAC recommended that CMS drop single-part enrollees from the FFS spending estimate and calculated the impact of removing Part A-only enrollees (Part B was a wash) as follows:

We examined the Part A and Part B FFS spending for beneficiaries who were in Medicare FFS for all of 2014 and enrolled in either Part A (with or without Part B) or in Part A and Part B. We found that Part A spending for beneficiaries enrolled in Part A and Part B all year averaged 8 percent more than average Part A spending for beneficiaries enrolled in Part A (with or without Part B). Beneficiaries in Part A who choose not to buy Part B are, on average, healthier than those who buy Part B. We found that the average risk score of beneficiaries enrolled in both Part A and Part B is 6 percent higher than all beneficiaries enrolled in Part A (with or without Part B),without accounting for the effect of Medicare Secondary Payer status. Therefore, after risk adjustment, we found the difference in Part A spending between these two groups of beneficiaries is about 2 percent higher for those in both Part A and Part B. 

 As Part A accounts for about half of Part A/B spending (48% in 2022), MedPAC estimates that excluding the Part A-only enrollees would raise MA benchmarks by about 1%.

Wakely, using 2019 data as opposed to the 2014 data on which MedPAC's analysis is based, found a considerably larger spending gap between two-part vs. Part A-only enrollees:

Wakely independently calculated the cost difference with the 2019 100% FFS data. We found that Part A spending was 13.4% higher for beneficiaries that were enrolled in Part A and B compared to those only enrolled in Part A. This compares to the 8% in MedPAC’s study. We believe one cause of the cost differential increasing is that MA penetration has been increasing over time, leaving fewer beneficiaries in FFS. Since beneficiaries must be eligible for both Part A and B, those with Part A only enrollment comprise a greater percentage of the total remaining in FFS.

That is not too surprising: Two-part enrollees dropped from 87% of all FFS enrollees in 2015 ( See Table 13-8 in the 2017 report)  to 84% in 2019. If Wakely followed MedPAC's approach to risk adjustment, the higher differential by itself might raise the estimated impact of removing single-part enrollees from the calculation to perhaps 2% instead of 1%.

But Wakely argues that risk adjustment should not be part of the calculation:

It is not necessary to adjust for a difference in risk scores because the CMS HCC risk model is developed using only beneficiaries enrolled in both Part A and Part B , and thus already reflects for the higher average risk of those with both Part A and B. Therefore, it would be inappropriate to use risk scores produced by the model for beneficiaries who are only enrolled in Part A or only Part B since diagnoses will only be based on a subset of claims as compared with beneficiaries enrolled in both A and B. It is also inappropriate to apply a risk adjustment to estimated costs since only the cost difference would flow through to nationwide benchmark rates now based on only those enrolled in both A and B. 

I can't adjudicate this.  I have not fathomed why MedPAC subtracts the higher risk scores of both-part enrollees from the calculation. In August 2021, however, KFF produced its own analysis of relative MA and FFS costs that did exclude single-part FFS enrollees, comparing the risk-adjusted costs of MA and FFS enrollees county by county nationally and calculating "the ratio between the Medicare Advantage risk scores and the average HCC score for traditional Medicare beneficiaries in each county." KFF found that MA per-person costs were 103% of FFS costs. As MedPAC's 2021 report pegged MA spending at 104% FFS, KFF's results track with MedPAC's estimate that removing single-part FFS enrollees from the spending estimate would have a 1% impact -- though KFF used a different methodology.

In recommending the removal of single-part FFS enrollees from the FFS spending estimate, MedPAC's goal was to make benchmarks fairer for counties with a high concentration of single-part enrollees, not to recalculate the relative value of MA vs FFS. Jeannie Biniek, lead author of the KFF report cited above, wrote in an email to me,  “The benchmark is a mechanism through which MA payments can be affected, but it is not the appropriate way to compare actual spending. In fact, if MA benchmarks were calculated using the spending of people enrolled in both Part A and Part B, benchmarks would be higher. Holding all else equal, payments would be higher – and MA would be even further from producing savings for CMS. The current methodology puts downward pressure on MA payments, but despite that, payments to MA plans are higher per enrollee than would be spent in TM.”

The Wakely analysis suggests that on a risk-adjusted basis, MA payments to MA plans are not higher per enrollee than would be spent in FFS -- leaving aside the hypothetical addition of an OOP to FFS.

If I can get a better purchase on that argument, I'll return to it. If any readers can help, please contact me!

* * *

* Part A-only enrollees comprised 12% of FFS enrollees in 2019. How, then, could their removal from the total FFS spending calculation for Part A raise the per-person total by 13.4%? Tim Courtney, co-author of the Wakely report, told me by email that the cost for Part-A only enrollees is about 83% lower than Part A costs for those enrolled in both A and B.

** Last two sentences of this paragraph added on 10/24/22.

More on Medicare Advantage

To whose advantage is Medicare Advantage? Part 1

To whose advantage is Medicare Advantage? Part 2

Is avoiding overpayment of MA plans beyond U.S. government capacity?

Will Medicare Advantage swallow fee-for-service Medicare?