RADV audits check whether coded diagnoses are supported by the record. This lesson covers how they work and the litigation that has reshaped them.
If risk scores drive payment and diagnoses drive risk scores, then verifying diagnoses is where program integrity lives. That verification is the Risk Adjustment Data Validation (RADV) audit, and it has changed substantially in the last two years, in both directions.
How a RADV audit works
CMS selects a sample of enrollees from a plan contract and asks the plan to produce medical records supporting the diagnoses it submitted. Diagnoses that the record does not support are unsupported, and the payment attached to them was not earned.
Recent operational detail: a January 2026 CMS memo confirmed that Payment Year 2020 audits began in February 2026, running on roughly a quarterly cadence, with a five-month record submission window and samples of 35 to 200 enrollees per contract.
The expansion
Historically CMS audited only a limited subset of contracts. Beginning in 2025, CMS announced it would audit all eligible MA contracts annually, a jump from roughly 60 plans per year to around 550. That is a step change in enforcement scope, and it means essentially every plan should now expect regular audit.
The extrapolation fight
The financial teeth of a RADV audit come from extrapolation: projecting the error rate found in a sample across the contract’s entire population, which turns a modest sample finding into a very large repayment. CMS estimated recovering about $4.7 billion through extrapolation between 2023 and 2032.
Worth remembering: that authority is currently in doubt. In September 2025 a federal district court vacated the RADV final rule that would have allowed CMS to extrapolate audit findings across a contract’s full population. The audits themselves continue, but extrapolated penalties are paused pending appeal. Anyone working in this area must check the current posture rather than assume extrapolation applies, because the answer has changed recently and may change again.
The pattern worth noticing
This is the second place in this course where a court has reshaped MA policy, and the Star Ratings lesson ahead is the third. It is a live illustration of the rulemaking course’s central point: when an agency imposes binding consequences, the authority and the procedure behind the rule matter, and parties with billions at stake will litigate both. Program integrity in MA is being worked out in courtrooms as much as in audits.
Key takeaways
- RADV audits sample enrollees and check whether submitted diagnoses are supported by the medical record.
- CMS expanded RADV from roughly 60 to about 550 contracts audited annually, with PY2020 audits beginning February 2026.
- A 2025 court decision vacated the rule permitting extrapolation, so audits continue while extrapolated penalties are paused; verify the current status before relying on it.
Sources
Check your understanding
What is the current status of extrapolation in RADV audits?
CMS expanded the audits substantially, but a 2025 district court decision vacated the rule that would have let CMS project sample findings across a contract's whole population. The audits continue; the extrapolated penalties do not, pending appeal.