Editor’s Note: This perspective was originally published on Health Affairs Forefront on July 1, 2024.
In a recent Forefront article, two current and two former members of the Medicare Payment Advisory Commission (MedPAC) challenge the Commission’s estimate that favorable selection and upcoding increase payments to Medicare Advantage (MA) plans in 2024 by $83 billion over what Medicare would vhave spent if MA beneficiaries had remained in traditional Medicare (TM). The authors mischaracterize MedPAC’s upcoding and favorable selection findings; they also shift to discussing the value of MA from the perspective of “a holistic policy future for Medicare,” rather than analyzing MedPAC’s work on the dollars and cents of MA overpayments. After identifying several important areas where we agree with the authors—Brian DeBusk, Brian Miller, Craig Samitt, and Kenny Kan (DMSK)—we explain why many of their key assertions about the MedPAC analysis are wrong.
Areas of agreement
We agree with DMSK about the importance of beneficiaries being able to choose between TM and MA and also about other MA policy and technical considerations. Paying MA plans monthly capitation creates starkly different financial incentives than those created by fee-for-service (FFS) reimbursement in TM. The payments to MA plans for their enrollees do not vary based on providing more—or fewer—services. The wide range of spending for different beneficiaries requires adjusting fixed monthly MA payments to reflect the risk associated with varying medical conditions as measured by patient diagnoses, eligibility and demographics. Reporting more diagnoses can increase risk scores, and higher risk scores increase monthly capitation payments. Many MA plans provide financial incentives to physicians and use other mechanisms to generate additional diagnostic codes that increase revenue.
In contrast, risk adjustment plays no role in TM. FFS reimbursement increases in TM when physicians provide more services to patients, not when they more aggressively code diagnoses.
We agree that competitive bidding and shifting to larger geographic areas have the potential to improve MA rate-setting (although we would not have bidding affect what beneficiaries in TM pay, as in proposals for “premium support”). We concur with DMSK about improving the measurement of quality and reforming “star ratings” in MA, as well as providing beneficiaries with meaningful information about quality in both MA and TM. Similarly, the Centers for Medicare and Medicaid Services (CMS) should address “outright fraudulent coding” and “diagnostic upcoding, wherein complexity is falsely enhanced to drive [MA] payment,” by issuing and enforcing stringent regulations and implementing a “muscular approach” to Risk Adjustment Data Validation (RADV) audits.
We also support other measures to address upcoding, many of which have been recommended by MedPAC in recent years. In addition, we share the view that TM benefits should be improved, Medigap should be reformed, and “[i]ndiscriminate, across-the-board cuts” should be avoided.
What the authors got wrong
Two different questions: the value of MA versus the size of MA overpayments
The recent Forefront article by Michael Chernew and Paul Masi (CM), MedPAC’s Chair and Executive Director, respectively, posed a very specific question: In 2024, how much more were estimated payments to plans than what Medicare would have spent if MA enrollees had been enrolled in TM? In two painstakingly detailed, peer-reviewed, analytic chapters totaling 110 pages published in March 2024, MedPAC concluded that upcoding and favorable selection paid MA plans $83 billion (22 percent) more than what Medicare would have paid if MA enrollees were in TM. MedPAC detailed its methodologies, compared its approach to alternatives, and presented sensitivity analysis, and its findings are consistent with multiple analyses of upcoding and favorable selection published by independent experts.
DMSK asserted that they were responding to alleged flaws in the MedPAC analysis of comparative costs. In fact, that is not what they did. Instead, DMSK focused on “programmatic comparisons between FFS Medicare and MA,” arguing “analytical rigor in assessing MA’s value and performance compared to FFS Medicare is an absolute requirement.” Electing to assess the normative value of MA is entirely legitimate but differs fundamentally from comparing the level of payments for MA plans to TM. Any comparison of the values of MA and TM must consider the costs borne by taxpayers and beneficiaries, in addition to health outcomes and other factors.
The remainder of this article addresses key flaws in DMSK’s criticisms of MedPAC findings about MA overpayments from upcoding and favorable selection.
Coding intensity
In addition to noting the importance of eliminating “fraudulent coding” and “diagnostic upcoding,” our prior section on areas of agreement highlighted that reporting more diagnoses increases payments in MA but not TM. This is because diagnoses in FFS physician claims establish medical appropriateness but not payment. The “Coding Intensity Factor” (CIF) authorized by the Affordable Care Act and implemented by CMS reduces upcoding by 5.9 percentage points, which is less than one-third of total upcoding. After incorporating the 5.9 percent adjustment, MedPAC estimates the remaining upcoding overpays MA plans by 13 percent, or $50 billon.
To illustrate the difference, consider two medically similar hypothetical patients, one in TM and the other in MA. Each has two chronic conditions: diabetes without complications and congestive heart failure (CHF), diagnosed when each patient was hospitalized five years earlier. Currently, the CHF is stable and each patient is treated and appropriately managed by a general internist.
The internist’s office visit claim for treating the TM patient would include a diagnostic code for diabetes. Reporting CHF in addition would not affect FFS reimbursement because the office visit covers the services rendered, regardless of whether just the diabetes diagnosis is reported or both codes are entered. In contrast, for the medically similar MA patient receiving the same services during an office visit, having the internist report the heart failure diagnosis—or the MA plan generate it through chart reviews, health risk assessments, or artificial intelligence—would approximately triple the monthly capitation paid to the MA plan compared to reporting only the diagnosis for diabetes.
The increased capitation from reporting CHF as well as diabetes for the MA patient overpays the plan because MA rates and risk scores are based on TM spending and diagnoses. Not surprisingly, many MA organizations (MAOs) have recognized the financial benefits of upcoding, with six MAOs reporting risk scores that ranged from 29 to 52 percent higher than scores for comparable TM beneficiaries. Compounding the inequity, MedPAC reports that MAOs encompassing 18 percent of MA enrollees lost money by upcoding less than the 5.9 percent across-the-board CIF; on the other hand, MAOs with 82 percent of enrollees (including seven of the eight largest MAOs) received windfalls by upcoding more than the CIF.
Even though unreported diagnostic codes have no effect on TM payment, DMSK consider reporting only the diabetes code to represent TM “undercoding.” They conclude that “the MedPAC analysis likely overestimates coding intensity effects” by leaving “FFS undercoding unaccounted for.” The assertion is unsupported: DMSK analyze neither MedPAC’s modeling of upcoding nor its detailed comparison assessing differing upcoding methodologies. We know of no evidence, nor do DMSK provide any, that so-called TM undercoding (such as by not recording CHF for the hypothetical diabetic patient) leads to or is associated with undertreatment of TM patients.
Favorable selection
DMSK assert that that MedPAC “violates several analytical norms including use of a non-peer reviewed comparator model,” but the study they cite is one of several conducted by highly-respected academics about upcoding—not about favorable selection. MedPAC has conducted sensitivity analyses and references over a dozen peer-reviewed studies finding substantial favorable selection into MA and sicker beneficiaries being more likely to disenroll from MA, as well as describing in detail its modeling methodology and assessing other studies of favorable selection. Extensive evidence shows TM beneficiaries are, on average, sicker than MA beneficiaries, even after risk adjustment, not the other way around.
MedPAC has described its detailed analysis of data about the 17 percent of MA beneficiaries enrolled in employer-group waiver plans (EGWPs). Without presenting any analysis, DMSK contend that including EGWP beneficiaries in MedPAC’s estimate of favorable selection somehow “violates several analytical norms,” apparently because they “are not available to the general public and are not subject to CMS plan bidding.”
DMSK also fault MedPAC for not including end-stage renal disease (ESRD) beneficiaries in its estimates of favorable selection. Unfortunately, the longitudinal data needed to estimate favorable selection are quite limited because ESRD beneficiaries only became eligible to enroll in MA plans in 2021, and MedPAC explains: “As a greater share of beneficiaries with ESRD enroll in MA, we will continue to monitor the potential effects of these enrollees on MA payments.”
Conflating overpayments, MA plan profit, and ‘differential payment’
DMSK create a confusing, irrelevant strawman argument about how overpayments cannot be correct because they don’t equal plan profits, which neither MedPAC nor, interestingly, DMSK suggest. The simple and clear question addressed by MedPAC is how much more MA costs than what Medicare would spend in TM. MedPAC estimates $2,329 in average overpayment per MA enrollee, split between $2,142 in extra benefits and $187 in plan retention, which reflects both increased administrative costs (for example, to administer a dental benefit) and additional profit. DMSK assert not only that the MedPAC model is incomplete and overestimates the likely effects of coding intensity and favorable selection, but more importantly fails “to distinguish overpayment from differential payment.” Despite “differential payment” appearing three times, the definition remains unclear but apparently involves “holistic” analysis.
DMSK appear to explain their focus on “differential payment” by arguing “if that [$83 billion] entire amount represented plan profit,” the “excess spending” would exceed insurers’ combined pre-tax income, somehow “suggesting that the health plans are losing money on their remaining lines of business—an unlikely scenario.” Their article’s next two sentences say: “Just as not all differential payment is overpayment, neither is all differential payment contributing to plan profits. In fact, MA uses $2,328 [sic] in rebates per beneficiary annually to deliver additional benefits.”
The remainder of their paragraph argues for “targeted policy improvements that equalize treatment of MA and FFS” and against “blind cuts in MA” to avoid hurting health equity and lower-income and minority beneficiaries. In arguing against an imaginary strawman, DMSK mischaracterize the thoughtful analyses presented in the two MA chapters in the March 2024 MedPAC Report, which include neither “blind cuts in MA” nor any new MA reforms. As summarized in Table 12-7, “Commission recommendations [that] would fix many flaws in MA payment policies” are longstanding and first presented in Reports issued between March 2016 and June 2021.
Comparing the MedPAC estimate of $83 billion in overpayments to reported insurer profits is meaningless: As DMSK acknowledge, a very large portion of that amount goes to extra benefits for beneficiaries—not insurer profits. However, MedPAC warns that reported profits and medical loss ratios (MLRs) might be unreliable, especially for vertically integrated firms because “the margin for the insurance line of business might not reflect the margin for the parent organization.” MA plans have extensive opportunities to lower reported profits and evade MLR requirements through payments to related non-insurance organizations, such as medical practices owned by the parent organization, or through shifting administrative costs to capitated providers.
DMSK introduce speculation about the implausibility of $83 billion in MA overpayments based on insurer profits and then acknowledge that their contention makes no sense. To protect beneficiaries and taxpayers, DMSK could have more constructively focused on improving transparency about MAO ownership of related non-insurance entities and MAO evasion of MLR limits by reclassifying administrative costs as medical.
Medigap inflates TM Costs—and MA rates
DMSK discuss induced demand from TM beneficiaries having Medigap policies that entirely eliminate beneficiary cost-sharing despite legislative reforms markedly changing the Medigap market. Effective January 1, 2020, previously popular Plans C and F that eliminated beneficiary cost sharing are no longer offered to new beneficiaries, other plans have also been eliminated, and coverage of out-of-pocket costs is generally less comprehensive. Supplemental coverage for retirees offered by former employers has also shifted away from TM to EGWPs, which generally involve managed care.
Perhaps more fundamentally, demand induced by supplemental coverage increases TM expenditures, making DMSK’s point directionally wrong: Higher TM per-beneficiary cost associated with supplemental coverage increases MA rates, so eliminating Medigap (and other supplemental coverage) would lower MA rates.
Conclusion
When making recommendations, MedPAC operates by consensus, with Commissioners occasionally dissenting from specific recommendations. It is highly unusual for current and previous Commissioners to attack the analytical underpinnings of MedPAC’s work, which is generally highly regarded as thorough and methodologically solid. DMSK have misleadingly framed their article as “responding to MedPAC’s recent analysis claiming $83 billion in MA overpayments,” rather than acknowledging it poses a very different question that reflects their normative values. As we’ve highlighted, their critique baldly mischaracterizes the analysis and findings presented in Chapters 12 and 13 of the March 2024 MedPAC Report to Congress.
Authors’ note
Ginsburg served as a MedPAC Commissioner from 2016 to 2022 and Vice Chair from 2019 to 2022. All of us (jointly) were authors of a formal comment to CMS filed in response to the proposed rule for 2025 payment rates for Medicare Advantage plans.
Lieberman, S. M., Ginsburg, P. B., Berwick, D. M., & Gilfillan, R. (2024). Estimating Overpayments To MA Plans: MedPAC Critics Get It Wrong. Health Affairs Forefront (2024).
10.1377/forefront.20240627.228859
Copyright © [2024] Health Affairs by Project HOPE – The People-to-People Health Foundation, Inc.
Sign up for Schaeffer Center news
