Improving Medicare Advantage by Accounting for Large Differences in Upcoding Across Plans

Editor’s Note: This perspective was originally published in Health Affairs Forefront on February 3, 2025.

As a result of insurers “upcoding” patient diagnoses, the Centers for Medicare and Medicaid Services (CMS) overpaid Medicare Advantage (MA) by $50 billion (or 13 percent) in 2024, according to the Medicare Payment Advisory Commission (MedPAC). Upcoding in this context refers to the practice of reporting diagnostic codes for more serious—and more expensive—conditions than would have been included on a fee-for-service bill had the same patient been in traditional Medicare (TM).

The degree of upcoding varies a great deal across Medicare Advantage organizations (MAOs), the insurers sponsoring one or more MA plans. MedPAC estimated a 15-percentage-point variation in coding intensity among the eight largest MAOs and found 10 MAOs with coding intensity more than 20 percent higher than levels in TM. The rules for MA payments generate windfalls for half of MAOs but penalize the other half that limit upcoding.

Revenues inflated by aggressive upcoding permit an MA plan to offer better benefits than either TM or MA plans that upcode less. Because enhanced benefits attract additional enrollees, more aggressive MAO upcoding can yield both higher profits per enrollee and increased market share. In addition to increasing taxpayer costs and Medicare Part B premiums, upcoding undermines fair competition among MA plans and distorts beneficiaries’ annual choice between MA and TM.

The variation in MAO upcoding highlights issues with across-the-board policies. One prominent example is the current uniform 5.9 percent coding intensity reduction in MA risk scores—and thus payment rates—designed to compensate for higher coding intensity in MA as compared to TM; this uniform reduction ignores the financial windfall and competitive advantage for MA plans that upcode more aggressively than other MA plans. By contrast, policy tools calibrated to the specific degree of upcoding by each MAO have the potential to reduce both unfairness across plans and aggregate overpayments.

After providing background on MA payment and upcoding, we describe a two-pronged reform strategy to limit the competitive advantage for plans that aggressively upcode while lowering overpayments. We first explain how risk-adjustment calculations—and thus relative payment levels in MA and TM—could be improved by increasing the uniform coding intensity reduction applied to MA risk scores. We then offer a second step that would replace the uniform reduction with varying reductions tied to each MAO’s level of upcoding, thus alleviating the unfair advantage that MAOs with high levels of upcoding gain over other MAOs within MA.

Background On MA Payment And Upcoding

CMS pays an MA plan fixed monthly, per-beneficiary amounts (capitation). These amounts are computed by multiplying a plan’s standard rate for the county in which an enrollee resides by that enrollee’s risk score. CMS risk adjusts capitation to account for Medicare’s extremely skewed distribution of per-beneficiary spending: The lowest-spending 20.0 percent of beneficiaries account for only 0.7 percent of annual spending and the highest-spending 20.0 percent of beneficiaries account for 76.6 percent of spending. Risk scores adjust per-enrollee capitation so that plan payments approximate expected plan costs.

The CMS risk-adjustment model computes risk scores based on per-beneficiary expenditures in TM, demographic characteristics such as age and gender, basis for eligibility (for example, aged, disabled, dual eligibility for Medicare and Medicaid), and health conditions as measured by diagnoses. The risk score of a standard (average) beneficiary equals 1.0, and actual risk scores vary from 0.1 to more than 20.0.

Upcoding inflates enrollee risk scores; this increases plan revenues when Medicare multiplies capitation by each enrollee’s risk score. When plans code diagnoses more aggressively than what is done in TM, they inflate the severity of illness reported for their beneficiaries relative to TM. The resulting upcoding increases the risk score calculated for a beneficiary, and an inflated risk score directly increases the monthly Medicare payment to an MA plan for that enrollee.

The pervasiveness of upcoding led Congress to initially require an across-the-board minimum coding-intensity reduction of 3.41 percent in 2010, which subsequent legislation increased to a 5.90 percent reduction starting in 2018. The minimum coding-intensity adjustment reduces the monthly amounts paid to plans for each MA beneficiary by the applicable percentage. Although the statute delegates to CMS the authority to increase the coding adjustment factor above the minimum percentage, CMS has never done so—thus it remains a uniform 5.90 percent reduction to all monthly MA payments.

MedPAC in March 2024 reported that the minimum coding adjustment financially penalizes the half of MAOs that upcode by less than the mandatory minimum reduction. The extent of upcoding and the additional revenue for the other half of MAOs—those that upcode by more than the minimum coding adjustment, which on average are much larger and have 82 percent of enrollees—varies markedly, but aggressive upcoding can inflate revenue by 20 percent to 50 percent relative to TM (see Figure 12-7, p. 382).

MedPAC concluded that current policy on upcoding undermined fair competition, with the starkly different financial effects ranging from modestly underpaying to massively overpaying plans:

“These differences demonstrate that CMS’s across-the-board adjustment for coding intensity, which reduces all MA risk scores by the same amount, generates inequity across contracts by reducing net revenue for plans with lower coding intensity and allowing other plans to retain a significant amount of revenue from higher coding intensity.

These differences are large enough to give MA organizations with higher coding intensity a significant competitive advantage by [increasing plan revenue] and helping them to attract more enrollees.”

A Two-Pronged Strategy To Reduce Overpayments From Upcoding

Our first set of proposals would improve how the CMS risk-adjustment model incorporates reported diagnoses when computing risk scores and increase the current 5.9 percent minimum coding intensity adjustment factor. These long-standing, analytically justified reforms advanced by MedPAC and CMS would reduce overpayments from upcoding. Not surprisingly, though, refining risk adjustment continues to generate significant opposition, with insurers objecting to anything that reduces MAO revenue, even revenue from overpayments due to upcoding.

The second prong of our reform strategy would increase MA fairness by limiting the competitive advantage MAOs receive from aggressive upcoding. In the aggregate, the average reduction would equal the overall coding intensity adjustment factor. However, instead of a uniform flat reduction, the upcoding adjustment factor would vary for individual MAOs to offset their gains or losses in revenue. Relating the coding intensity adjustment to each MAO’s degree of actual upcoding would redistribute revenue and facilitate fairer competition among MA plans. Although the largest MAOs would strongly oppose this proposal because they upcode aggressively, half (or more) of MAOs would benefit from it.

Improving Risk Adjustment: Increasing Fairness Between TM And MA

The first prong of our strategy to address upcoding involves implementing MedPAC recommendations and CMS reforms to improve the calculation of risk scores. MedPAC recommends excluding diagnoses that are obtained from health risk assessments and using two years’ claims data when calculating risk scores; a recent Department of Health and Human Services Inspector General report documented how in-home health risk assessments increase upcoding. MedPAC also recommends increasing the minimum coding intensity adjustment to a reduction that “fully accounts” for the differences in coding in TM and MA.

In 2025, CMS should complete, on schedule, the third (and final) year of phasing in its new “v28” risk-adjustment model. The v28 model significantly improves the calculation of risk scores by excluding selected diagnoses that have little, if any, meaningful relation to a beneficiary’s health care or predicting high future health spending. Importantly, neither the CMS risk-model changes nor the MedPAC recommendations would affect diagnostic codes with meaningful implications for treatment or for recognizing truly ill beneficiaries.

Improving Risk Adjustment: Increasing Fairness Within MA

Instead of an across-the-board reduction in capitation, the second prong of our strategy would link each MAO’s reduction to its actual track record of upcoding. In the aggregate, the MAO-specific coding intensity factors would be adjusted to achieve the same total savings as the uniform coding intensity reduction.

Plans that upcode less than the uniform reduction factor would receive increased capitation. Plans that upcode by more than the uniform reduction factor would have their revenue decrease, reducing if not eliminating the reward for upcoding aggressively. Achieving the same total savings while offsetting the increased payments to MAOs that upcode little would require larger reductions to upcoding intensity factors for MAOs that upcode more aggressively. For example, if increased payments to MAOs with limited upcoding totaled 10 percent of the overall overpayment savings target, MAO-specific upcoding factors would have to increase by 10 percent: For a MAO with a 10 percent reduction factor, CMS would lower its capitation by 11 percent; for a MAO with a 50 percent reduction factor, CMS would lower its capitation by 55 percent.

In assessing the redistribution associated with shifting from a uniform to MAO-specific adjustment, MedPAC found MAOs with 82 percent of enrollees have upcoded by more than the minimum coding adjustment factor, and seven of the eight largest MAOs were among the most aggressive upcoders. The half of MAOs that upcoded relatively little accounted for only 18 percent of MA enrollees. Each MAO-specific upcoding factor would reflect its degree of upcoding. (As a technical matter, MAO-specific adjustments would also take into account enrollment to assure overall savings equal those from the uniform coding intensity factor.)

Operationalizing MAO-Specific Coding Intensity Adjustment Factors

One approach to operationalizing this policy would assign MAOs to one of three buckets. The first category would include plans that upcode by significantly less than the uniform reduction factor. The second category would include plans that upcode by amounts that do not differ significantly from the uniform reduction. The third category would include plans that upcode significantly more than the uniform reduction. In addition, the third category could be subdivided into plans that upcode by somewhat more than the average (for example, up to 5 percent or 10 percent more) and plans that upcode substantially more than the average.

CMS or Congress would set the uniform coding intensity adjustment factor. The CMS Actuary would determine the percentage adjustment factor (if any) needed to have the MAO-specific adjustments generate the same aggregate savings as uniform coding intensity adjustment factor. Initially, MAO upcoding amounts could be calculated employing MedPAC’s methodology in its March 2024 Report to Congress, although CMS could refine the MAO-specific coding intensity factor methodology over time.

Additional considerations could affect MAO-specific coding intensity adjustment factors. For example, the MAO-specific coding intensity factor could be increased (for example, by 20 percent) for plans that both upcode aggressively and engage in problematic practices—as identified by MedPAC, CMS, the CMS Office of Inspector General, and the Government Accountability Office—such as using home visits or AI to maximize the identification and reporting of diagnoses.

Conclusion

The MedPAC and CMS reforms would reduce upcoding by refining risk adjustment, which would both decrease overpayments and limit gains from questionable upcoding practices. Implementing the MedPAC recommendation to increase the uniform coding intensity adjustment above the statutory minimum would significantly reduce overpayments due to upcoding. But despite being analytically sound, the reforms would engender strong industry opposition, with insurers acting to mobilize beneficiary opposition by claiming that reducing their revenue will reduce the generosity of MA benefits.

Shifting from reducing MA capitation by an across-the-board coding intensity adjustment factor to MAO-specific adjustment factors would improve the financial outlook for many MAOs but reduce or eliminate the financial gain for MAOs that upcode aggressively. Redistributing the effects of the across-the-board coding intensity adjustment factor would improve equity and permit fairer competition among MAOs. It is unknown the extent to which “good guy” MAOs would support this approach to reform, but the policy has the potential to splinter uniform opposition by insurers.

Paying MA more than $100 billion—almost 30 percent—more than what MA beneficiaries would cost in TM might not be sustainable given the size of the federal deficit and the prospect of extending expiring tax provisions. Programmatically sound proposals cutting federal mandatory spending might attract more support under these circumstances than they have in the past. MedPAC has documented $83 billion or 22 percent in MA overpayments in 2024, with 13 percent generated by upcoding and 9 percent from favorable selection. In addition, two statutory features increase MA payments by another $23 billion or 6 percent above what MA beneficiaries would cost in TM. Requiring budget neutrality—the standard practice in TM—for “star ratings” bonuses and the “quartile” system for setting MA rates presents additional opportunities for eliminating MA spending that exceeds what costs would be for beneficiaries in TM.

Authors’ Note

The authors acknowledge financial support from Arnold Ventures.

“Improving Medicare Advantage By Accounting For Large Differences In Upcoding Across Plans”, Health Affairs Forefront, February 3, 2025 .DOI: 10.1377/forefront.20250131.366467