Texas Court Strikes Down CMS’s RADV Rule – CMS’s Treatment of Actuarial Equivalence and the FFS Adjustor Does Matter, After All | Epstein Becker & Green
In what will create a glimmer of hope for beleaguered Medicare Advantage Organizations (MAOs) who are facing looming waves of Risk Adjustment Data Validation (RADV) audits, on September 25, 2025, a Northern District (ND) of Texas District Court invalidated the Centers for Medicare & Medicaid Services’ (CMS) 2023 RADV Final Rule, finding that the agency violated the notice-and-comment requirements of the Administrative Procedure Act (APA).
This ruling comes amidst a recently announced aggressive agenda by the current Trump administration to catch up on five payment years of RADV audits. This agenda has triggered visceral concern from MAOs, who are struggling to understand the potential breadth and impact of future extrapolated recoupments. Given the adverse residual impact to MAOs through CMS’s V28 risk adjustment (RA) model cuts, RADV audits landed at a less than opportune time for the industry. The Court’s ruling, while still subject to appeal and possible CMS rule redlining, is at least temporary good news for MAOs. But, for those popping corks for their victory lap, it ain’t over yet.
Background
In February 2023, CMS issued its RADV Final Rule, a regulation facially designed to strengthen oversight of MAOs. Though RADVs had, by 2023, been conducted for more than a decade in a limited fashion, the “new” rule allowed CMS to:
- Conduct extrapolated audits: Though CMS was vague as to the universe to which extrapolation would apply (contract or a sub-cohort of enrollees), the extrapolation methodology would involve sampling plan enrollees and extrapolating the results across a to-be-defined sampling frame to recoup overpayments; and
- Eliminate the longstanding Fee-for-Service (FFS) Adjuster: The FFS Adjuster is a statistical buffer meant to account for differences between traditional Medicare FFS data and Medicare Advantage (MA) data, which CMS ultimately designed to comport with actuarial equivalence (AE) requirements in the model. In short, the MA risk adjustment model was built on FFS-related data and costs. It is widely known there are defects in the reporting of diagnosis codes in the FFS model, which does not take into account what is actually documented in a medical record. RADV audits actually require diagnoses to be validated in medical records. So, the FFS Adjuster was meant to bring parity (a measure of AE) to a model based on unvalidated diagnoses to a RADV methodology that has the effect of requiring 100% chart validation of diagnoses.
CMS has long discussed the use of the FFS Adjuster, supporting the use of it in its original RADV final rule in 2012. However, at the height of False Claims Act (FCA) enforcement and Congressional and agency demands to police the risk adjustment MA payment program, CMS abandoned its narrative around the need for the FFS Adjuster. Its reasoning and empirical evidence justifying this departure have been widely criticized. But, by attempting to remove the FFS Adjuster, CMS granted itself the ability to demand larger repayment amounts dating back to 2018, since it would not need to discount for errors in FFS data when calculating recoupments. This proposal, combined with a broadly defined extrapolation methodology, raised the stakes for MAOs, including Humana, which brought the suit challenging the 2023 RADV Final Rule.
Humana Inc., et al. v Becerra, et al., No. 4:23-cv-00909-O, N.D. Tex. (2025)
Humana sued under the APA, arguing that CMS’s 2023 RADV Final Rule was both procedurally and substantively defective. The central contention was that CMS had changed its justifications between the proposed rule and the final rule in ways that deprived stakeholders of fair notice. Humana attacked the rule on multiple grounds, including: the rule’s retroactivity, i.e., a 2023 rule applying to bids for 2018; the arbitrary and capricious legal standard; and, most significant to this Court, the rule’s defects related to mandated APA notice-and-comment procedural safeguards.
In the 2018 RADV Proposed Rule, CMS justified eliminating the FFS Adjuster primarily on the basis of (1) an empirical study suggesting the adjuster was unnecessary, and (2) when the adjuster was applied, fairness concerns that audited plans were being treated differently than unaudited ones. Though not part of the Court’s rationale, notably CMS’s study has been challenged and criticized as being an unsound basis upon which to justify striking out what appeared to be a long-standing policy to achieve AE in risk adjustment payment. Significantly, in the 2023 Final Rule, CMS shifted to new rationale, arguing that “actuarial equivalence” requirements did not apply in the RADV context, and that a statutory “coding intensity adjustment” already addressed industry-wide coding variation, making the FFS Adjuster duplicative. CMS appeared unable to argue its “coding intensity” rationale before the Court, and therefore the Court focused solely on the legal attachment of AE requirements to RADV.
Humana argued that the abandonment of AE in RADV departed from expectations, blindsided MAOs, and robbed the industry of the ability to provide meaningful comments on CMS’s new position. Humana contended that CMS violated the APA and requested the 2023 RADV Final Rule be invalidated.
The Court’s Holding
Judge O’Connor of the ND of Texas vacated the 2023 RADV Final Rule on notice-and-comment procedural grounds. While appearing to nod favorably at Humana’s retroactivity arguments, the Court did not formally address either retroactivity and or the arbitrary and capricious arguments because it did not need to.
Judge O’Conner focused his scrutiny on CMS’s failure to comply with the so-called Logical Outgrowth Doctrine. Under the APA, agencies must publish a proposed rule, give the public adequate notice and opportunity to comment, and respond to significant comments before finalizing a regulation. This process is a fundamental tenet of administrative law and a check-and-balance on agency action. To ensure the connectedness of a proposed and final rule, a final rule must be a “logical outgrowth” of the proposal. Stakeholders in any given industry should be able to anticipate the agency’s rationale and have an opportunity to provide meaningful comments and feedback to the agency. Here, the Court found that CMS’s new rationales in the 2023 RADV Final Rule for removing the FFS Adjuster were sufficiently distinct from the initial justifications provided by CMS in the 2018 RADV Proposed Rule. The 2023 RADV Final Rule was, therefore, not a logical outgrowth of the 2018 RADV Proposed Rule. Stated plainly, CMS advanced a wholly new basis on which to remove the FFS Adjuster that stakeholders had no chance to address.
CMS argued that any procedural flaw with regard to its changed reasoning was harmless, since it had ultimate authority to interpret the statute, and the result was the same – removal of the FFS Adjuster. Judge O’Conner disagreed, emphasizing that notice and comment is not a box-checking exercise: stakeholders lost the opportunity to present data, arguments, and alternative perspectives on the new theories. And, while not ruling on retroactivity, the Court noted that this lost opportunity was particularly consequential given the retroactive application of the rule back to 2018. Because the defect was procedural and central to the rulemaking process, the Court vacated the 2013 RADV Final Rule in its entirety and remanded the rule to CMS for further proceedings consistent with the APA.
Well…What Now?
Given the national implications of the Court’s ruling, it is likely that CMS, acting through the Department of Justice, will appeal. This ruling, in a venue likely viewed as favorable to the plaintiffs, throws a wrench into RADV audits and how they play out, potentially a big wrench. That said, it is entirely unclear what CMS will actually do in response to this decision with regard to its current RADV strategy. CMS conducted RADV audits for years without a final RADV rule on payment recoupment. However, CMS has also not recouped any meaningful recoveries from RADVs ever. By remanding the rule back to the agency, the Court ordered CMS (should it choose to continue) to “go fix it,” consistent with the Court’s opinion. Regardless, this decision is welcome break and rare good news for health plans. MAOs have been on a multi-year rollercoaster. Yesterday the ride operator was ordered to pump the breaks.
[View source.]
