Editor’s Note: This analysis is part of the USC-Brookings Schaeffer Initiative for Health Policy, which is a partnership between Economic Studies at Brookings and the University of Southern California Schaeffer Center for Health Policy & Economics. The Initiative aims to inform the national health care debate with rigorous, evidence-based analysis leading to practical recommendations using the collaborative strengths of USC and Brookings. This op-ed was first published in Health Affairs Forefront on February 24, 2022.
Health Affairs Forefront has published articles that advance two diametrically opposing assessments of issues regarding the Medicare Advantage (MA) program. In September 2021, Donald Berwick and Richard Gilfillan (B-G) argued that mounting MA overpayments, many from plan efforts to get physicians to code more diagnoses when they see a patient, simultaneously threaten Medicare’s fiscal soundness and health care delivery; for them, this “MA money machine” is reflected in the extremely high valuations obtained by start-up insurance companies focused on MA and primary-care physician practices that specialize in contracting with MA plans.
In his January 2022 response, George Halvorson (GH) categorically disputes many of the key B-G points, which he views as attacking the fundamentals underpinning MA, including the use of capitation as a payment approach. Most recently, Donald Crane (DC), president and CEO of America’s Physician Groups, advanced a viewpoint consistent with GH, albeit in a more measured narrative.
Agreement On MA Fundamentals
B-G, GH and DC all seem to embrace the fundamental policy premises underlying MA: allow beneficiaries to select organized delivery systems that Medicare reimburses based on capitation as alternatives to fee for service. However, B-G disagree forcefully with GH and DC about the extent of MA overpayments; the two sides also disagree on the social value of having taxpayers fund MA plans to offer more generous benefits than traditional Medicare, where beneficiaries typically purchase supplemental Medigap coverage at an average monthly cost of $150 (but premiums vary widely and can be much higher). B-G contend that the current system not only markedly overpays MA plans while adding little value but also creates damaging incentives that seriously distort health care.
All of the authors (including ourselves) have long believed in shifting the delivery of care away from unorganized fee for service towards organized systems of care that promote quality, permit measurement, and foster accountability. Clearly, MA has moved Medicare in that direction: between 2010 and 2022, beneficiaries in private plans in the 50 states plus DC increased by 17.8 million (163 percent), more than the 17.2 million increase in overall Medicare beneficiaries, with those in traditional Medicare falling by 0.6 million (-1.7 percent). Newer initiatives, such as accountable care organizations (ACO), are seeking to move delivery of care toward organized, accountable systems of care for those choosing to remain in traditional Medicare. And all the authors agree about the importance of promoting quality improvement throughout Medicare, as well as elsewhere.
Thus, many points in dispute strike us as B-G and GH/DC “talking past” the other. We should also point out that GH spent decades leading illustrious integrated care delivery organizations such as Health Partners and Kaiser Permanente, and the members of DC’s organization are integrated medical groups that deliver care to 25 percent of MA patients. The extent to which the remaining 75 percent of physicians who treat MA patients are effectively coordinating care, or are paid by MA plans on a capitated basis rather than fee for service, is an open question.
Quality In MA
Linking MA plan overpayments with a conclusion B-G attribute to the Medicare Payment Advisory Commission (MedPAC)—that MA has “no demonstrable clinical benefit to patients”—might be construed as fundamentally criticizing the underlying MA program. However, the attribution represents a misreading of MedPAC, which concluded that insufficient data limits the ability to meaningfully compare differences in clinical quality between MA and traditional Medicare, a failure associated with inadequate Centers for Medicare and Medicare Services (CMS) reporting requirements. Notably, peer-reviewed literature has found that MA does better on many, but not all, indicators of quality.
MA Profit Margins
B-G and GH differ starkly on whether Medicare is overpaying MA plans and on the need to reform payment approaches. B-G highlight the extreme valuations shown in recent transactions; GH counters that MA plan profits average only 4.5 percent of premiums, citing MedPAC’s analysis of the data plans file with CMS. Although 4.5 percent is consistent with what several industry sources confirmed to us as their estimates of average returns, two caveats apply. MedPAC noted that for-profit MA plans tend to have higher-than-average margins, and its data were for individual but not employer group MA plans. Employer-group MA plans are reportedly more profitable for insurers and comprise 18 percent of MA membership in January 2022.
How Medicare Sets MA Payments
A brief review of how CMS pays MA plans will inform why Medicare spends more on beneficiaries in MA than in traditional Medicare. Plans submit bids in June to CMS indicating the dollar amount it would take them to provide Medicare Part A and B benefits in the coming calendar year, including administrative costs and profit margins, to a member of average health status. CMS establishes a “benchmark” for each county reflecting Medicare spending for those beneficiaries not enrolled in MA. Benchmarks are adjusted according to a complex, legislatively mandated quartile system classifying each county’s per beneficiary Medicare spending into four categories, ranging from counties with low to high Medicare spending; benchmark rates are lowered relative to actual FFS Medicare spending for plans in counties with relatively high spending and are raised for plans in low spending counties. In addition, CMS increases payments by 5 percent to any plan with a quality rating at four or five stars; this increases MA costs, relative to what MA spending would be without the quality bonus and relative to the costs of serving the same beneficiaries in traditional Medicare, because the bonus payments for plans with higher star ratings are not offset by reductions in payments for plans with lower ratings.
Medicare’s monthly base payment—the amount before incorporating a star ratings bonus—will be the benchmark if a plan bid equals or exceeds a county benchmark. However, for the overwhelming majority of plans that bid below county benchmarks, the base payment is the plan’s bid. Plans bidding below benchmarks receive a “rebate” payment from CMS that increases their base payment (over and above any star ratings bonus). Rebates range from 50 percent to 70 percent of the difference, with higher star ratings entitling the plan to the higher percentages. Plans receiving rebates must “return” them to enrollees in the form of either premium reductions or additional benefits, which can be either lower cost sharing or services not covered by traditional Medicare.
Although plans submit bids to CMS for a standard beneficiary, the actual monthly payments for each plan member increase or decrease based on the risk scores of those who have enrolled. As a result, generating higher risk scores for MA members increases MA plan revenues. The patient diagnoses reported by physicians determine risk scores, and MA plans can directly or indirectly create incentives to affect physician coding, as detailed in a recent Department of Justice complaint against Kaiser Permanente.
Incentives To Add Diagnoses To Increase Risk Scores And MA Payments
With profit margins averaging 4.5 percent, a key question to ponder is why recent start-up valuations for health plans and primary care physician groups focused on the MA market average $87,000 per beneficiary, compared to a historical range for MA plans of $4,000 to $10,500? Although one factor may be a recent economy-wide tendency to assign historically high valuations to new and growing companies, B-G have a table highlighting the importance of more aggressive coding that boosts the risk scores assigned to their MA beneficiaries. Suggesting that start-up MA firms might have much higher profitability than other MA organizations, B-G point to their aggressive coding enhanced by pursuing new approaches to increase coding of diagnoses. The more rapid growth of start-up MA firms and the prospect of their approaches spreading to the rest of the industry represent important threats to B-G.
The essence of the coding issue is that physicians seeing patients with traditional Medicare have little incentive to list multiple diagnoses; they just need to have enough information on the claim to get paid for the service. On the other hand, MA plans get higher payments when their enrollees are classified as sicker, so they have strong incentives to get physicians to be more thorough in their coding. According to MedPAC, health risk assessments and chart reviews are the most important tools for plans to increase coding and get paid more. CMS currently reduces MA risk scores by the statutory minimum adjustment of 5.9 percent, but CMS has authority to impose larger reductions. MedPAC estimates that the differences are much larger—9.1 percent in 2019. As a result, the mandatory minimum subtraction from plans’ risk scores is currently too low by 3.2 percent, which contributes to the more generous benefits offered by MA plans.
GH views more generous MA benefits as a social good and believes that minimum Medical Loss Ratios (MLR) constrain plan profits. We share the skepticism raised in the peer-reviewed literature about the efficacy of MLRs as a constraint on plan profitability; tactics that can undermine the ability of MLRs to constrain MA plan profits include reclassifying administrative expenses as health care costs, which can occur when MA plans are tightly integrated with providers.
Alternative Estimates Of Effects Of Upcoding On MA Payments
Jacobs and Kronick measured coding intensity in a different way than MedPAC, comparing the CMS risk score for each contract with risk scores developed from prescription drug claims data. Coding intensity differs markedly across plans; plans that exhibit more aggressive coding—and receive higher Medicare payments—are able to both expand extra benefits offered enrollees and earn higher profits. Plans that are more aggressive than their peers likely will gain market share since they can offer more additional benefits. In addition, Jacobs and Kronick find that, if CMS imposed a larger coding intensity reduction, the adjustment would have a larger impact on plan profits than on enrollee benefits and premiums.
How Plans Increase Risk Scores
B-G outline some of the approaches used by the highlighted companies to increase risk scores. Clover Health pays physicians $30 per visit to use its Clover Assistant AI platform, which identifies coding opportunities. Approaches that pay medical groups a percentage of the monthly CMS capitation harness groups’ incentives to code more aggressively: higher coding increases the monthly payment, which the insurer and the medical groups can share. To the extent that a percentage of capitation contracts assign administrative activities such as utilization management, provider credentialing, or claims processing to a medical group, those expenses are considered medical and not administrative costs, undermining the efficacy of minimum MLRs as a hard constraint on profits.
B-G’s thesis comes down to the new organizations reaping the rewards from being better at increasing risk scores, allowing them to grow faster and earn more money. They show that for a primary care group contracting on a percentage of premium basis with MA plans, the returns per physician can be extremely high.
Social Value Of MA Overpayments
GH is clear about not being concerned with overpayments to MA plans if a portion of the supposed overpayments are going to MA enrollees and MA plans are promoting coordinated care. GH buttresses his conclusion by highlighting the attractiveness of MA to lower-income Medicare beneficiaries. But having the public sector overpay crowds out other governmental services, requires higher taxes, or increases fiscal deficits. Despite MA plans being able to deliver traditional Medicare benefits at an average of 87 percent of what spending would have been in traditional Medicare, MedPAC’s latest estimate is that MA payments exceed what the beneficiaries would have cost in the traditional program by 4 percent.
Given the magnitude of plan efficiencies in delivering Medicare benefits, we believe that many of the benefit improvements flowing to MA enrollees can be retained but more of these savings should go to taxpayers, as was envisioned when risk-based private plans were enacted in 1982. A policy that links deliberately overpaying MA plans to the availability of better benefits—by restricting better benefits to MA enrollees—essentially requires Medicare beneficiaries to leave traditional Medicare to share in the added benefits.
Another serious problem with overpaying MA plans is how it affects their behavior. The CEO of a large insurer participating in all insurance markets—not just MA—told one of us that the MA cuts included in the Affordable Care Act motivated insurers to get serious about providing value in their MA products; their success led to growth in MA, in contrast to predictions by the Congressional Budget Office and the Medicare Actuary that MA enrollment would decrease. When overpayment had been larger (actually much larger in some markets than it is today), many MA plans could prosper with an “arbitrage” approach of providing uncoordinated care financed by the overpayment. While the organizations that GH and DC are most familiar have cultures that are committed to approaches that realize value, it’s unclear the extent to which this permeates the rest of the MA marketplace. To the degree that financially winning under the current approach to paying MA plans results from gaming increased risk scores, it could slow the movement towards value even more.
Role Of Star Ratings
Both GH and DC wax enthusiastically about the system of star ratings for quality, one which classifies–and rewards—most plans for being above average! MA organizations have increased star ratings by combining geographic areas to enable the better-scoring plans to achieve the same bonuses as the plans that scored more poorly, making the quality information meaningless to beneficiaries. This loophole has only recently been addressed. The many Medicare quality incentive programs for providers are budget neutral, meaning the bonuses paid to relatively good performers are offset by the penalties for those that perform relatively poorly. Only the incentive program for MA is not budget neutral; it increases overall spending substantially.
Meaningful measures of quality in MA are limited because of the lack of data to compare enrollees’ experience to that of beneficiaries in traditional Medicare. Despite billions of dollars paid, the meaningfulness of the measures in the star rating system remains in question. In addition to making the system budget neutral, MedPAC has recommended redesigning the star ratings by replacing numerous process measures of quality with a limited number of outcome measures reported for both MA enrollees and traditional Medicare beneficiaries. MedPAC has also recommended accounting for differences in enrollees’ social risk factors by stratifying plan enrollment into groups with similar social risk profiles; this would avoid disadvantaging those plans with larger shares of higher-risk enrollees while maintaining transparency on actual quality.
Summing Up
Despite agreeing on the importance of shifting patients to organized systems of care and reducing reliance on fee for service, B-G diverges sharply from GH and DC on the appropriateness and value of MA as the program is currently implemented. The B-G critique centers on current rules fueling what they call the “MA money machine”, where excessive payments arise from aggressive coding and star rating bonuses. In their view, excessive plan reimbursement—and profits—attract MA-focused startups that threaten to spread undesirable behaviors as more established insurers and providers struggle to compete financially, a dynamic that inhibits needed changes. GH and DC adopt a very different—and much more sanguine—view of the current MA program, with GH being quite outspoken in defending extra payments as being socially beneficial and not leading to inappropriate profits.
We have particular concerns with the politics of MA overpayment. This overpayment leads not only to higher profits by plans but to extra (and increasing) benefits for enrollees—a potent combination that complicates reforming the program. The rapid and continuing growth of private plans—which in February 2022 enrolled 46 percent of Medicare beneficiaries—highlights the urgency of reform measures to address excessive payments driven by coding increases, star ratings, and the potential for unreliable county-level rates based on the costs of fewer and fewer beneficiaries in traditional Medicare.