The answer first

CMS is now auditing every RADV-eligible Medicare Advantage contract, every year. In a May 2025 newsroom announcement, "CMS Rolls Out Aggressive Strategy to Enhance and Accelerate Medicare Advantage Audits," the agency committed to annual audits of all eligible MA contracts, roughly 550 plans instead of the roughly 60 it historically reviewed, while clearing the backlog of payment years 2018 through 2024 and expanding its medical coder workforce from about 40 to approximately 2,000 certified coders. Unsupported diagnoses found in sampled records can now be extrapolated across the contract under the RADV final rule, CMS-4185-F2 (88 FR 6643, February 1, 2023). And it all lands in CY2026, the first year the CMS-HCC V28 risk model is fully phased in per the CY2026 Rate Announcement. The risk model and the audit regime changed at the same time.

The scale shift, in four numbers

Seven payment years, cleared at once

The May 2025 CMS announcement does not just change the audit rate going forward. It commits to working through the accumulated backlog of payment years 2018 through 2024. That means a chart documenting a 2019 date of service is still audit-relevant today, and every one of those years can now generate record requests.

Why extrapolation changes the arithmetic

Under the RADV final rule, CMS-4185-F2 (88 FR 6643, February 1, 2023), CMS may extrapolate audit findings. A sampled record that fails to support a submitted diagnosis is no longer a one-record recovery. The error rate found in the sample can be projected across the contract's payments. One thin chart can now represent thousands of dollars of extrapolated exposure, which is exactly why plans are pushing documentation-support demands downstream to the provider groups that generated the diagnoses.

• Step 1CMS samples enrollee records from an audited contract
• Step 2Certified coders review whether the chart supports each diagnosis
• Step 3Unsupported diagnoses produce a payment error rate
• Step 4Per CMS-4185-F2, the error rate is extrapolated across the contract
• Step 5Plans recover from risk-sharing provider groups per their contracts

The V28 collision

Either change alone would be a compliance project. Together they mean the diagnoses your organization submits under a brand-new model in 2026 are being generated at the exact moment CMS has more audit capacity than it has ever had. There is no grace period in which V28 coding patterns settle before scrutiny arrives. If your documentation practices were built for the old sampling rate of roughly 60 plans a year, they were built for a world that no longer exists.

What this means for provider groups in risk arrangements

The audits target MA organizations, but the economics flow downhill. Provider groups in capitation, percent-of-premium, or shared-savings arrangements typically carry contractual responsibility for supporting the diagnoses they submit. Expect three things to rise sharply: chart-pull volume from your MA plan partners, requests to re-attest or supplement documentation for prior payment years back to 2018, and contract language that shifts extrapolated recovery exposure onto the group. Audit and enforcement teams should treat every submitted HCC as a claim that will eventually be read by one of those 2,000 coders.

Operator to-do list

1. Inventory your risk-sharing contracts. Identify every arrangement where extrapolated RADV recoveries can be passed through to your group, and quantify the payment years 2018 through 2024 exposure window.
2. Stand up a chart-retrieval response function. Plan-initiated pulls will arrive in waves. Define owners, turnaround targets, and a tracking log before the first request, not after.
3. Run an internal documentation-support audit on your highest-weight HCCs. Sample your own submitted diagnoses the way a CMS coder would: is the condition documented, evaluated, and supported in the encounter note?
4. Rebaseline coding workflows for V28. Confirm your coders and CDI staff are working from the V28 mappings for CY2026 dates of service, not muscle memory from the prior model.
5. Close the loop on prior-year gaps now. Where legacy records are thin, document what can legitimately be located and organized today, and never fabricate or backfill clinical content.
6. Brief leadership on extrapolation math. Boards and finance teams should understand that sample error rates project across contracts under CMS-4185-F2, so small sample failures are not small.

Sources

• CMS press release, "CMS Rolls Out Aggressive Strategy to Enhance and Accelerate Medicare Advantage Audits," cms.gov/newsroom, May 2025.
• RADV final rule, CMS-4185-F2, 88 FR 6643, February 1, 2023, authorizing extrapolation of audit findings.
• CY2026 Rate Announcement, CMS, confirming the CMS-HCC V28 model is fully phased in for CY2026.