CMS Finalizes Medicare Advantage Risk Adjustment Rule

Health Highlights

The Big Picture

On January 30, the Centers for Medicare & Medicaid Services (CMS) released a long-awaited final rule addressing how it will calculate and collect overpayments from Medicare Advantage (MA) plans arising from risk adjustment errors discovered through data validation audits.

In the final rule, CMS announces that it will recover excess premiums paid to MA plans when it finds unsupported or inaccurate plan-submitted risk adjustment data. CMS will not only recover the specific premiums associated with a specific error but will also extrapolate a plan’s error rate across entire cohorts of beneficiaries, thereby magnifying the potential scope of recoveries. In a related decision, CMS will not be using an adjuster to establish equivalence in error rates between MA data and risk adjustment data reported in the fee-for-service (FFS) Medicare program. The agency has granted plans some relief, deciding not to use extrapolation for audits that look back at 2011 through 2017 and beginning the extrapolation-based recoveries for the audits looking at the 2018 contract year.

The decision is expected to draw significant objection and litigation challenges from MA plans, which stand to lose significant funds under the rule. CMS projects that recoveries for 2018 alone will add up to $479 million. There will likely be downstream impact to providers and vendors who have been delegated risk or participate in value-based contracts for MA populations, and who profit based on the performance of the population they manage. Payments in those arrangements are often tied to premiums, which may be impacted by closer scrutiny and increased recoveries based on risk adjustment errors.

Background: Medicare Advantage Risk Adjustment

Payments from CMS to MA plans are risk-adjusted. The monthly premium CMS pays to an MA plan is increased to account for each beneficiary’s health status and demographic factors, and the degree to which those differences are expected to cause increased medical spending. Through this risk adjustment process, the Medicare program attempts to minimize the process of “adverse selection,” under which MA plans avoid sicker beneficiaries and their higher medical costs.

CMS and the U.S. Department of Health and Human Services Office of Inspector General (HHS OIG) have increasingly come to view the risk adjustment program as a “major driver” of improper payments in the MA program. Out of concern that plans might improperly increase their payments through risk adjustment, CMS has imposed requirements for documenting and substantiating beneficiaries’ diagnoses in medical records and claims. Plans must repay funds to CMS when a diagnosis is found to be erroneous or unsubstantiated, or face sanctions and suits under the False Claims Act.

Federal regulators have increasingly audited MA plans for errors in risk adjustment in the past few years. The HHS OIG has been conducting targeted reviews of individual plans to search for incorrectly coded diagnoses. CMS’ main effort in this area is the Risk Adjustment Data Validation (RADV) audit program, dating back to calendar year (CY) 2011. Under RADV, CMS reviews medical records from a sample of up to 201 beneficiaries per MA contract and determines if any of the diagnoses submitted for those beneficiaries were erroneous.

RADV audits looking back at 2011, 2012, and 2013 have been conducted but are not yet finalized as CMS has been awaiting this final rule to determine how it will calculate the final tally of associated overpayments. The RADV audit looking at 2014 was launched in 2019, and the 2015 audit, delayed from intended start by COVID, took place in late 2020.

The issue left unresolved until this final rule is how CMS would treat the results of RADV audits and to what extent MA plans would be required to repay CMS for overpayments arising from erroneous diagnoses uncovered through RADV. Since 2012, CMS has contemplated extrapolating from the error rate determined in these audits. That is, rather than reverse and recoup payments associated with specific members whose diagnoses were erroneous or unsupported, CMS would recoup a much larger sum from plans on the presumption that the error rate from that small sample is representative of the error rate among all diagnoses submitted for all beneficiaries.

This proposal has been hotly contested since its original announcement. Nonetheless, in 2018, CMS formally issued a proposed rule under which it stated it intended to move forward with extrapolation-based recoupment for audits looking back at 2011, 2012, and 2013, and for all future years.

Under the 2018 proposed rule, CMS proposed to extrapolate an error rate determined through a RADV audit to an entire contract’s worth of beneficiaries’ diagnoses and recoup presumed overpayments from plans. It also proposed that it could instead identify sub-cohorts of beneficiaries within a contract (such as beneficiaries with a specific diagnosis), identify a sub-cohort-specific error rate, and apply that error rate contract-wide to the entire sub-cohort. And it reserved the right, in all cases, to review individual beneficiary records outside the RADV process when there is suspicion of error.

A second component of the proposed rule was a CMS decision not to apply a FFS Adjuster to the extrapolated error rate. The MA risk adjustment model is based on claims experience and costs observed in the FFS Medicare program. If the diagnoses submitted by physicians in that program contain errors, a FFS Adjuster to the extrapolated RADV error rate could be used to account for that underlying bias in the MA risk adjustment program. But around the time of the proposal, CMS released the results of a long-awaited study and asserted that errors in FFS claims data do not have any systemic effect on the MA risk adjustment model or payments made to MA plans. Accordingly, CMS did not propose to make any FFS adjustments to extrapolated RADV recoveries.

CMS did not immediately finalize the proposals. Instead, it first extended the comment period and then allowed itself multiple extensions on the grounds of “exceptional circumstances” against the usual statutory three-year deadline for publication of final Medicare rules.

The Final Rule

Extrapolation. In the final rule, CMS announces that it will extrapolate RADV audit findings to determine overpayments recoverable from MA plans due to RADV-detected errors.

CMS is not actually specifying any particular method for extrapolating. Instead, it will “rely on any statistically valid method for sampling and extrapolation that is determined to be well-suited to a particular audit.” This absence of a specific methodology appears calculated to give CMS flexibility in applying extrapolation differently in different circumstances, as CMS notes it “may include applying one or more RADV audit methodologies for any given RADV audit” or using discretion to not use extrapolation in a given circumstance. It may also be a calculated sidestep of commenter criticism of particular extrapolation methods (such as extrapolation based on sub-cohorts) and an attempt to avoid legal challenges on the basis of a particular method’s invalidity. Instead, individual plans will have to challenge the validity of a specific extrapolation method used in their own audit.

The extrapolation of RADV audit results applies retroactively but not so far into the past as originally proposed. The original proposed rule had indicated CMS would extrapolate results as far back as those from the RADV audits for 2011. Instead, CMS will only begin extrapolating with the RADV audit for 2018. For the audits looking at 2011 through 2017, CMS will collect only enrollee-level overpayments identified in those audits, or in audits conducted by the HHS OIG. This delay is justified by CMS based on operational concerns, such as controlling the number of appeals filed in the years following this final rule. It may also serve as an olive branch of sorts to MA plans: Plans need not fear substantial repayments for errors in the 2011 through 2017 period.1

Any retroactive imposition of extrapolation methods is likely to invite legal challenge. CMS states that it is “not imposing additional liabilities, penalties or retroactive application of new requirements or policy” and is merely recovering improper payments, but that claim will likely be litigated.

Any extrapolation methodology adopted by CMS for RADV audits will be focused on plan contracts that, through statistical modeling, are identified as being at the highest risk for improper payments.

Once assessed, CMS will recover contract-level payment adjustments through a lump-sum reduction in the plans’ monthly payments. CMS indicates that it will now begin sending recovery notices to plans for individual enrollee-level errors uncovered in the audits that examined 2011, 2012, and 2013.

FFS Adjuster. CMS is finalizing its proposal not to apply an FFS Adjuster to RADV audits.

In justifying this decision, CMS leans heavily on its interpretation of the statutes establishing the MA program. According to CMS, the statute’s mandate that MA rates have “actuarial equivalence” to those in FFS is applicable to rates alone and not to the obligation of an MA plan to report and return improper payments for diagnoses lacking medical record support. In so doing, CMS finds support in a recent decision from the U.S. Court of Appeals for the D.C. Circuit in UnitedHealthcare Insurance Company v. Becerra, where the court reached a similar conclusion in the context of False Claims Act litigation addressing the obligation to report overpayments. (For more on this decision, see the August 20, 2021, edition of Insights This Week.)

Notably, CMS does not cite its study on coding in FFS as a basis for its decision and explicitly disclaims dependence on it as grounds for finalizing the rule. This move may be an attempt to avoid a “battle of the studies” between CMS and advocates for an FFS Adjuster.

1 CMS estimates the forgone collections for this period at $683.2 million.



pursuant to New York DR 2-101(f)

© 2024 Manatt, Phelps & Phillips, LLP.

All rights reserved