The Records Are Due August 28. The Extrapolation Answer Isn’t.
CMS’s PY2020 RADV audit guidance sets a hard medical-record submission deadline and a sampling method built to support extrapolated recovery, but five months after a federal court vacated CMS’s extrapolation authority, the agency still won’t say whether it will use it.
CMS’s payment year 2020 (PY2020) Risk Adjustment Data Validation (RADV) audit guidance, released March 20, 2026, requires Medicare Advantage organizations selected for audit to submit medical records by August 28, 2026. The sampling method CMS uses is explicitly designed to support extrapolated recovery. CMS has not decided whether it will actually extrapolate PY2020 findings across a plan’s full contract population, or recover only the sampled overpayments. That determination is still pending, five months after a federal court vacated CMS’s underlying extrapolation authority. CMS’s appeal of that ruling remains unresolved.
A new audit methodology, a fixed deadline, and one decision CMS keeps deferring
On March 20, 2026, CMS released its audit methods and instructions for PY2020 RADV audits, along with the list of Medicare Advantage contracts selected for audit. The methodology changed from prior years. CMS now draws sampled enrollees from the top quartile, rather than the top decile used in PY2018 and PY2019, of all RADV-eligible enrollees, ranked by CMS’s Center for Program Integrity predictive models for likely overpayment. Sample sizes run 35, 50, 100, or 200 enrollees depending on a contract’s size.
The Enrollee Data List became available in CMS’s Centralized Data Attestation Tool on April 3, 2026. The medical record submission window runs from April 13 to August 28, 2026, with hardship exception requests due September 11, 2026. Every part of that timeline is fixed and already in motion. The one piece CMS has not fixed is what happens to the evidence once it is submitted.
CMS’s own guidance: the PY2020 audits were “designed… to support the collection of extrapolated recoveries,” but CMS has “not decided whether it will make extrapolated recoveries… or only recover the specific overpayments associated with the sampled enrollees.”
CMS built the audit to support extrapolation. It just won’t commit to using it.
For risk adjustment and compliance leaders, the open extrapolation question matters directly. The same medical-record submission work due August 28 could produce two very different financial outcomes, depending on a decision CMS has not made and has not committed to a timeline for. Treating this submission as routine is defensible only for plans small enough that the extrapolation multiplier would not materially change the outcome.
A procedural defeat for CMS put extrapolation on hold, and CMS is appealing rather than conceding
Humana Inc. and Humana Benefit Plan of Texas, Inc. v. Becerra, No. 4:24-cv-00909-O (N.D. Tex., Sept. 25, 2025).
- Court findingCMS’s 2023 RADV final rule (CMS-4185-F2) was vacated. The final rule’s justification for eliminating the fee-for-service adjuster differed materially from what CMS had proposed, denying stakeholders a genuine chance to comment.
- Immediate effectExtrapolated recovery is paused. CMS currently recovers only sampled-enrollee overpayments across ongoing audits.
- CMS’s responseCMS is appealing the ruling rather than accepting it. PY2018 findings built on the vacated methodology may need to be revised depending on the appeal’s outcome.
- PY2020 postureRather than resolve the extrapolation question before PY2020 audits began, CMS built a methodology “designed to support” extrapolation and left the actual decision open.
Collect the evidence first. Decide the methodology later.
CMS’s own language is the tell here. The agency did not say it has deferred the extrapolation decision because the litigation is unresolved. It said the PY2020 audit was “designed… to support the collection of extrapolated recoveries,” while declining to commit to actually using that capability. That is a deliberate sequencing choice: collect the evidence first, under a fixed deadline, and preserve the option to decide the recovery methodology later, once the Humana appeal is resolved.
For a compliance program, the operational task does not change. Records are due August 28 either way. What should change is how finance and legal treat this filing.
CMS built the audit to support extrapolation. It just won’t say whether it will use it.
One open regulatory decision, six functions carrying the uncertainty
Risk adjustment / Coding
Must submit complete, defensible medical records for every sampled enrollee by August 28, regardless of which recovery methodology eventually applies.
Compliance
Needs a live monitoring process for CMS’s eventual extrapolation determination and the Humana appeal docket, not a one-time check.
Finance / Actuarial
Cannot size this year’s contingent liability with a single number. Reserve planning needs a documented range, not a point estimate.
Legal
Should evaluate hardship exception eligibility and whether a future extrapolation rule could apply retroactively, echoing the retroactivity dispute already in the Humana litigation.
IT / Data governance
Medical record retrieval and submission workflows need to be audit-ready for every sampled enrollee ID, with a timestamped submission log.
Executive leadership
This is a governance-level exposure question, not a routine coding audit, for any plan large enough that the extrapolation multiplier would matter.
The submission deadline is fixed. Ownership of the exposure question isn’t, yet.
CMS has been explicit that it built this audit to keep the extrapolation option open. That means no single function can treat this as someone else’s problem. The table below assigns ownership; the list after it sequences the work.
| Function | Required response | Evidence to retain |
|---|---|---|
| Risk adjustment / Coding | Retrieve the Enrollee Data List and gather, validate, and submit supporting medical records for every sampled enrollee before August 28, 2026. | Complete, timestamped medical record submission log tied to each enrollee ID. |
| Compliance | Track CMS’s extrapolation determination and the Humana appeal docket; document how a contract-wide exposure scenario would be assessed if extrapolation is later applied. | Internal monitoring log of appeal status and CMS guidance updates. |
| Finance / Actuarial | Model both a sample-only recovery scenario and an extrapolated, contract-wide recovery scenario for reserve planning. | Dated reserve-range memo covering both scenarios. |
| Legal | Assess hardship exception eligibility where applicable; evaluate retroactivity exposure if CMS reissues an extrapolation rule. | Hardship exception filing (if applicable) by September 11, 2026; legal memo on retroactivity exposure. |
Within 7 days
Confirm your organization’s PY2020 audit-selection status and retrieve the current Enrollee Data List from CDAT if selected. Assign a named owner for submission tracking.
Within 30 days
Complete a gap assessment of medical record availability against the sampled enrollee list. Escalate any records at meaningful risk of being unavailable before the deadline.
Within 60–90 days (post-submission)
Build both recovery scenarios into reserve and contingent-liability planning. Monitor the Humana appeal and any CMS announcement of its PY2020 extrapolation decision. These are editorial recommendations, not confirmed CMS requirements.
The ownership table above assumes four functions can work in parallel without a shared reference point. In practice, the function most likely to be under-resourced here is finance and actuarial, because a submission deadline reads as a compliance task, not a reserving question, until someone models both scenarios side by side.
PCOOB Weekly’s assessment is that the more useful governance move is not to guess CMS’s eventual decision, but to quantify the exposure now. Elevance’s own $935 million Q1 2026 accrual, disclosed as its best current estimate of exposure in an unrelated risk-adjustment certification enforcement matter (see the July 16, 2026 edition), shows plans already treat contingent risk-adjustment liabilities as material enough to put a number on before the outcome is final. The same discipline applies here: model a bracketed range now, not a single guess, so whatever CMS eventually decides updates a figure the plan has already reported rather than triggering a scramble after the fact.
Questions payer leaders should ask
- 01Has our organization confirmed whether we are selected for PY2020 RADV audit, and do we have the Enrollee Data List?
- 02Do we have a documented plan to submit complete medical records for every sampled enrollee before August 28, 2026?
- 03Have we modeled our potential exposure under both a sample-only recovery and an extrapolated, contract-wide recovery scenario?
- 04Who in our organization is tracking the Humana v. Becerra appeal, and how will we learn when CMS makes its PY2020 extrapolation decision?
- 05Does our audit-readiness program treat this submission as routine, or does the open extrapolation question warrant elevating it to a governance-level risk?
- 06If CMS applies extrapolation later, do we understand whether it could apply retroactively to this year’s submission?
Sources
- Crowell & Moring: CMS Releases PY 2020 RADV Audit Methods and Instructions: Key Takeaways for Health Plans (March 27, 2026)
- Groom Law Group: Court Rules That CMS Cannot Extrapolate Medicare Advantage Risk Adjustment Audit Results (October 8, 2025)
- Humana Inc. and Humana Benefit Plan of Texas, Inc. v. Becerra, No. 4:24-cv-00909-O (N.D. Tex. Sept. 25, 2025), as reported by Groom Law Group (September 25, 2025)
- CMS: Medicare Advantage Risk Adjustment Data Validation (RADV) Program — RADV Documents and Data (Accessed August 2026)
About PCOOB Weekly: PCOOB Weekly is an independent digital publication focused on U.S. healthcare payer compliance, operations, governance, oversight, and technology. It provides source-led analysis for Medicare, Medicaid, Commercial, and pharmacy benefit stakeholders.
Until next week, stay briefed.
Frequently asked questions
What is the deadline for PY2020 RADV medical record submissions?
CMS’s submission window for payment year 2020 Risk Adjustment Data Validation audits runs from April 13, 2026 to August 28, 2026, with hardship exception requests due September 11, 2026.
Has CMS decided whether PY2020 RADV findings will be extrapolated?
No. CMS has stated it designed the PY2020 audits to support extrapolated recovery but has not decided whether it will actually apply extrapolation across the full contract or limit recovery to the sampled enrollees, and has not given a timeline for that determination.
Why is the extrapolation question in doubt right now?
A federal district court vacated CMS’s 2023 RADV final rule, which had authorized extrapolation, in September 2025 on procedural grounds, finding CMS did not give stakeholders adequate notice of the final rule’s actual legal reasoning before it took effect. CMS is appealing that decision.
How does extrapolation change a plan’s financial exposure?
Extrapolation projects a sample’s error rate across an entire contract’s population rather than recovering only the specific overpayments found in the sample. CMS’s own guidance illustrates that a 5 percent sample error rate on a $1 billion contract could produce a $50 million recovery under extrapolation, compared with a far smaller sample-only recovery.
Does this affect PY2018 and PY2019 RADV audits too?
Yes. PY2018 audits, already under way on 60 contracts using the now-vacated extrapolation methodology, may need to be revisited. PY2019 audits, covering roughly 550 contracts, are proceeding under the prior rule with the fee-for-service adjuster and no extrapolation.