Coding
Medicare Advantage RADV audits and HCC coding: what is checked and what is unsettled
Medicare Advantage RADV audits check whether the diagnoses a plan submitted for risk adjustment are supported by medical records. The plan submits the records, and each audited condition category is scored as confirmed or discrepant, then converted into dollars. That is a different measure from a 95% code accuracy rate. The 2023 RADV rule was vacated in 2025 and is on appeal. This is general information, not legal advice. Review your payer contracts and the rules where you practice with your compliance advisor.
Published by OffshoreMeds. Facts last checked October 17, 2026. How we check our numbers.
The audit
What a RADV audit checks, who submits and how it is scored
| Step | What the sources say | Source |
|---|---|---|
| Step: Purpose | What the sources say: CMS checks that diagnoses submitted for risk adjustment are supported in the enrollees' medical records, and may collect overpayments for unsupported ones. | Source: CMS RADV program page |
| Step: Who submits | What the sources say: The Medicare Advantage organization and its providers submit a sample of medical records. In the payment year 2024 audit the plan uploads at least one valid record for each audited condition category. | Source: 42 CFR 422.310(e) |
| Step: A valid record | What the sources say: A legibly signed medical record from a face-to-face visit with a provider within the data collection year. | Source: CMS audit methods, 2026 |
| Step: The result | What the sources say: Each audited category is confirmed, confirmed higher, discrepant, discrepant lower or an administrative exception. CMS re-runs each sampled enrollee's risk score with only the validated categories and converts the difference to dollars; a net underpayment counts as zero, so an audit never adds payment. | Source: CMS audit methods, 2026 |
| Step: The appeal | What the sources say: A plan has 60 days from the audit report to appeal, may appeal one record for each audited category and carries the burden of proof. | Source: 42 CFR 422.311(c) |
The measure
Why a code accuracy rate is not a RADV result
- A RADV result is measured for each sampled enrollee and each audited condition category, in dollars. The 95% figure the Journal of AHIMA calls a de facto standard is a code-level benchmark (see how to audit medical coding accuracy). Neither measures the other.
- CMS says RADV audits are not specifically designed to detect fraud (88 FR 6644 and 6645), and a diagnosis not documented in the record is not a valid basis for risk adjustment payment.
- The acronym MEAT (monitor, evaluate, assess, treat), common in vendor content, did not appear in the CMS, OIG, Federal Register or court texts read. It is an industry convention, not a CMS test.
- Documentation comes from the treating provider’s signed note. Coders work from that record, and the diagnosis stays a clinical decision for the provider. See offshore medical coding staff for what coders do.
The model
The 2024 CMS-HCC model (V28) and the 2027 changes
| Payment year | Share from the 2024 model | Source |
|---|---|---|
| Payment year: 2024 | Share from the 2024 model: 33% | Source: CY2024 Rate Announcement |
| Payment year: 2025 | Share from the 2024 model: 67% | Source: CY2025 Rate Announcement |
| Payment year: 2026 | Share from the 2024 model: 100% | Source: CY2026 Rate Announcement |
| Payment year: 2027 | Share from the 2024 model: The 2024 model continues; CMS did not adopt the recalibrated model it had proposed | Source: CY2027 Rate Announcement, April 6, 2026 |
For 2027 CMS finalized excluding diagnoses from unlinked chart review records, with an exception for beneficiaries who changed Medicare Advantage parent organization, and excluding diagnoses from audio-only services when nothing else on the record is eligible. A diagnosis can still count if it is submitted on an encounter record or on a chart review linked to one. Payments for 2027 use diagnoses from 2026 dates of service (CY2027 Rate Announcement).
Status
The 2023 RADV rule: vacated and on appeal
| Date | What happened | Source |
|---|---|---|
| Date: February 1, 2023 | What happened: CMS published the final rule: extrapolation of RADV findings beginning with payment year 2018, and no fee-for-service adjuster in RADV audits. | Source: 88 FR 6643 |
| Date: September 25, 2025 | What happened: A federal district court in Texas vacated and remanded the rule on notice-and-comment grounds and did not rule on whether extrapolation or the missing adjuster is lawful. | Source: Court order |
| Date: November 2025 | What happened: The government appealed to the Fifth Circuit. | Source: CMS audit methods |
| Date: October 17, 2026 | What happened: No decision in that appeal was found on the court's opinions page, which listed opinions dated through October 16. | Source: Fifth Circuit opinions page |
- CMS says audits continue and that it has not yet decided whether it will collect extrapolated overpayments or only the amounts tied to the sampled enrollees (CMS audit methods).
- The eCFR text of 42 CFR 422.310(e) still reads that CMS may extrapolate for payment year 2018 and later, so the text has not been amended to match the court order.
- Check the court and CMS pages before relying on this table: the appeal can be decided at any time.
Oversight
What OIG and the regulations say about vendors
- A contractor or subcontractor that generates risk adjustment data must also certify its accuracy, completeness and truthfulness (42 CFR 422.504(l)).
- OIG’s voluntary guidance for Medicare Advantage organizations says risk adjustment should be a top priority in overseeing third parties, and asks plans to consider enhanced systems when vendors are paid in ways tied to risk adjustment data (OIG guidance).
- OIG’s 2026 audits of four plans found most sampled high-risk enrollee-years unsupported (A-07-24-01214, A-05-24-00010, A-03-22-00004, A-07-22-01207). Each sample was drawn from high-risk diagnosis groups, so the shares are not error rates for a plan or for the program.
Vendor questions
Questions to put to a risk adjustment coding vendor
No source says these are required. Each is tied to a rule or a document above.
- Will you certify in writing the accuracy, completeness and truthfulness of the data you generate (42 CFR 422.504(l))?
- How is each diagnosis tied to a face-to-face visit with a signed note from an acceptable provider type?
- For chart review diagnoses, how is each one linked to a submitted encounter record, given the 2027 exclusion, and how are unsupported diagnoses reported for deletion?
- Are you paid for each diagnosis or category captured, and what self-audit and reporting terms apply?
- Is your accuracy measure per code, per claim, or per category and enrollee, and what would it show if scored the way RADV scores it?
- Where do your coders sit, and what attestations do you give for offshore subcontracting (see offshore subcontractor attestation)?
Limits
What the sources do not say
- Whether extrapolation is lawful now, whether it is applied to payment year 2018 and later audits, and whether a fee-for-service adjuster applies, are not settled by anything read.
- Whether any findings or collections have issued for payment years 2018 or 2019 was not found, and no error rate is given because the OIG samples were targeted.
- No source read says whether offshore coders raise or lower RADV exposure. The CMS reviewer guidance found is a 2020 edition and was not confirmed as the version used for payment year 2024.
This is general information, not legal advice. Review your payer contracts and the rules where you practice with your compliance advisor.
The terms behind each seat
One flat monthly fee per seat, never a percentage of collections. A Business Associate Agreement is signed before anyone touches protected health information. Clinical decisions stay with your US-licensed providers. See all OffshoreMeds services.
Related guides
- Offshore medical coding staff
ICD-10, CPT and HCC coders working in your EHR, audited daily against a published accuracy standard.
- Auditing coding accuracy
OIG's sampling guidance, why no source sets an accuracy percentage, how to read the CERT rate and how to check a coder's credential.
- Offshore subcontractor attestation
What CMS requires of plans, what one plan's provider form asks and what a practice should have ready.
- Vendor questions before you sign
Questions on contract, access, location, payer rules, pay and exit, each tied to the rule or source behind it.
Get Started
Book a free 30-minute call.
Pick a time below. Have these ready: the roles you need, your specialty and your EHR. We’ll come back with a fixed seat quote, candidate profiles to interview and a go-live plan.
Calendar not showing? Open the booking page in a new tab.
Please keep patient information out of the booking notes. A Business Associate Agreement is signed before anyone touches protected health information. See how this site handles your details in the privacy policy.