A new HHS Office of Inspector General (OIG) audit puts a hard number on a coding problem risk adjustment leaders have tracked for years. CMS made an estimated $462 million in net overpayments to Medicare Advantage (MA) organizations in 2021, tied to acute stroke diagnosis codes the medical records did not support. The finding arrives as CMS expands Risk Adjustment Data Validation (RADV) audits and points to where auditors look next.
How the payment works
MS pays MA plans a monthly rate per enrollee, adjusted for health status. Sicker members carry higher risk scores and higher payments. Plans build those scores from diagnosis codes their providers submit. Some diagnoses raise payment more than others, so high-value codes draw scrutiny. Acute stroke is one of them.
What the audit found
OIG reviewed submissions for 97 enrollees in coordinated care and private fee-for-service plans. Every case failed. For all 97 sampled enrollees, the acute stroke codes plans sent to CMS lacked support in the medical records behind the physician data records. OIG extrapolated the sample and estimated $462 million in net overpayments for 2021 alone.
The pattern is precise. A plan submitted an acute stroke diagnosis on a physician record, but no acute stroke diagnosis appeared on an inpatient or outpatient hospital record during the same service year. An acute stroke produces a hospital encounter. When the hospital record shows no matching acute event, the physician code does not hold up.
OIG made one recommendation. CMS should build a procedure to block these payments when an acute stroke code appears on a physician record with no corresponding hospital diagnosis in the same year. CMS did not state whether the agency agrees.
Why this is not new
Auditors flagged this same scenario in earlier reviews of individual plans. This audit widened the lens across multiple organizations and found identical results. Acute stroke now sits beside acute heart attack and other acute conditions as a repeat target. The query is simple for auditors to run and difficult for plans to defend after submission.
What this means for health plans
- Extrapolation changes the math. CMS finalized a RADV rule allowing extrapolation of audit findings, so a small set of unsupported codes no longer means a small repayment. A handful of charts becomes a population-level estimate.
- The data signature is easy to find. A physician acute stroke code with no hospital acute stroke code in the same year is a clean flag. Your plan should run the query before CMS does.
- The burden sits with you. An acute stroke generates a hospital record. A chronic history of stroke uses a different code with different support rules. Coding the acute version off a single physician note invites the finding OIG describes.
The question to ask your team
If CMS pulled your acute stroke population today, how many codes would survive a chart review? A blank answer is your first project. Build the query, work the charts, and fix the upstream documentation gaps before the next audit cycle reaches you.
Risk adjustment accuracy and RADV readiness anchor the agenda at the RISE Risk Adjustment Forum, where plan leaders work through extrapolation risk, audit preparation, and the coding integrity practices these findings demand. Bring your acute condition data and leave with a plan.
Source: HHS OIG Report A-02-23-01020, issued May 28, 2026.