Understanding RADV Audits: CMS Risk Adjustment Data Validation — A Complete Training Guide

Introduction

In Medicare Advantage (MA), risk scores directly drive plan payments — but accuracy is everything.
When CMS suspects that submitted diagnoses may not be fully supported by medical documentation, it initiates a Risk Adjustment Data Validation (RADV) audit.

The RADV audit process ensures that every dollar paid to MAOs aligns with the true clinical documentation in beneficiary medical records.

Understanding how RADV works — from sampling and validation to extrapolation and overpayment recovery — is essential for MA compliance, finance, and risk adjustment teams.

What Is RADV?

Risk Adjustment Data Validation (RADV) is CMS’s audit process used to confirm that diagnoses submitted for risk adjustment are supported by medical records.

The goal:
To verify that Medicare Advantage Organizations (MAOs) receive appropriate payments — not more, not less — based on accurately documented conditions.

RADV audits are required under:

• Social Security Act §1853(a)(1)(C)

• 42 CFR §422.310(e)

CMS performs these audits to detect errors, calculate payment discrepancies, and ensure compliance with risk adjustment data submission and documentation standards.

Types of RADV Audits

1. Contract-Level RADV (CL-RADV)

• Conducted at the contract level (H-number)

• Statistically valid sample of enrollees

• Extrapolation applied across contract payments

• Used by CMS as the primary audit mechanism since PY 2018

2. National RADV (NRADV)

• Nationwide review across multiple plans

• Used for model calibration, research, and quality control

3. Targeted or OIG-Led RADV

• Conducted by the Office of Inspector General (OIG)

• Focused on high-risk areas or conditions showing anomalous coding patterns

RADV Audit Flow: From Start to Recovery

1. Notification

• CMS notifies selected MAOs via official correspondence.

• Each contract receives a sample of enrollees for review.

2. Medical Record Submission

• MAOs submit medical records to the CMS Central Data Abstraction Tool (CDAT).

• Records must be legible, complete, and properly linked to each diagnosis submitted.

• MAOs typically have 12–16 weeks to submit documentation.

3. Validation

• CMS and its contractors review records against the submitted diagnoses.

• Each diagnosis is classified as:

  • Confirmed — Supported by the medical record

  • Discrepant — Documentation mismatch or incomplete

  • Invalid — No support found

  • Confirmed Higher/Lower — Hierarchy or specificity differs

4. Error Rate Calculation

• CMS computes an enrollee-level payment error by comparing the payment-weighted risk scores for confirmed vs. unconfirmed conditions.

5. Extrapolation

• The contract-level error rate is extrapolated from the audited sample to all members under that contract.

• This step began with Payment Year 2018, as finalized under the RADV Final Rule (CMS-4185-F2).

6. Appeal and Recovery

• MAOs may appeal findings within established timelines.

• CMS issues an Extrapolated Overpayment Recovery Notice for confirmed payment errors.

• Payment adjustments follow if no appeal is sustained.

How CMS Calculates Payment Error

Derived from the Payment Error Calculation Methodology (2025 update):

• CMS calculates each enrollee’s difference between original and validated risk scores.

• All enrollee differences are aggregated into a contract-level mean error.

• The mean error is multiplied by the total contract population to determine the estimated total payment error.

• No “fee-for-service adjuster” is applied — CMS now uses actual MA data only.

Formula (simplified):

Payment Error = (Σ Validated RAF - Σ Submitted RAF) / n * Contract Enrollee Count

Preliminary Findings and Discrepancies

During review, each diagnosis receives one of several possible results:

• ✅ Confirmed — Supported exactly as submitted

• ⚠️ Discrepant — Supported by record but coded to a different HCC

• ❌ Invalid — Unsupported diagnosis or incorrect documentation

• ⬆️ Confirmed Higher — More specific diagnosis found (e.g., diabetes with complications instead of unspecified)

• ⬇️ Confirmed Lower — Less specific or downgraded version validated

Each enrollee’s results contribute to the overall sample error rate, which is the foundation for contract-level extrapolation.

The Role of Overpayments & Self-Reporting

Per Section 1128J(d) of the Social Security Act, MAOs must report and return overpayments within 60 days of identification.

If RADV identifies overpayments, plans are required to:

• Validate internally whether additional contracts are affected.

• Correct and report identified overpayments to CMS.

• Maintain documentation demonstrating compliance.

Record Retention and Documentation Standards

Under 42 CFR §422.504(d) and (i), MAOs must:

• Retain supporting documentation for 10 years after the final payment determination.

• Maintain audit-ready records linking each submitted diagnosis to a valid medical record.

• Ensure each record includes patient name, date of service, provider signature, and credentials.

Incomplete or unsigned records are automatically considered invalid for RADV validation.

Sampling and Extrapolation Explained

CMS uses statistically valid sampling methods (as detailed in your uploaded Contract-Level RADV FAQ).

• Each sample is randomly drawn from the contract’s enrollee population.

• Stratification ensures representation across key variables (age, gender, HCC profile).

• Extrapolation applies only to Payment Year 2018 and onward.

CMS no longer applies the “FFS Adjuster” methodology — aligning MA audits more closely with commercial audit standards.

Appeals and CMS Process

1.  Request for Reconsideration (Reopenings)
   MAOs can request review of findings and submit additional documentation.

2. CMS Reconsideration Determination
   CMS issues an official determination following internal review.

3. Hearing Official Review (2nd Level)
   Independent Hearing Official reviews disputes on methodology or evidence.

4. Final Agency Decision
   The CMS Administrator or delegate issues a final decision.

Payment recovery follows, and MAOs must comply with debt collection procedures under 45 CFR Part 30.

Common RADV Error Themes

• Coding without documentation: Diagnoses reported but not supported by the chart.

• Incorrect provider type: Notes signed by unqualified providers.

• Date mismatches: Documentation not matching the service year.

• Illegible or incomplete records: Missing key identifiers or signatures.

• Unsupported hierarchical coding: Submitted at higher specificity than documented.

• Chart Review linkage errors: Missing or mislinked chart reviews to encounters.

Compliance Best Practices

• Maintain clear traceability: ICD → HCC → medical record evidence

• Audit internal submissions monthly against CMS reporting

• Train coders on RADV audit findings and documentation rules

• Automate error tracking from MAO-002 and RADV reconciliations

• Prepare an “Audit Binder” — a consolidated record of all documentation tied to each diagnosis submission

Key CMS References & Tools

• CMS RADV Resources Portal

• RADV Final Rule (CMS-4185-F2, January 2023)

• [RADV Sampling & Calculation Methodology Memo, May 2025]

• [CMS Medicare Managed Care Manual, Chapter 7, §40 – Risk Adjustment Data Validation]

Conclusion

RADV is not just an audit — it’s a system-wide check on data integrity and compliance.

Every plan’s financial accuracy, audit readiness, and CMS reputation depend on maintaining precise, well-documented diagnosis reporting.

In Medicare Advantage, documentation drives payment — and RADV ensures the documentation stands behind every dollar.

By mastering the RADV process — from EDPS submission to extrapolated overpayment reconciliation — your team can stay ahead of compliance risk and build a sustainable, audit-ready risk adjustment operation.