Written and maintained by CASRAI Editorial Board
Last updated
A Risk Adjustment Data Validation (RADV) audit is the primary federal program integrity mechanism used by the Centers for Medicare & Medicaid Services (CMS)—and separately by the U.S. Department of Health and Human Services (HHS) for Affordable Care Act (ACA) commercial risk adjustment—to verify that Hierarchical Condition Category (HCC) diagnosis codes submitted by Medicare Advantage Organizations (MAOs) for risk-adjusted capitation payments are substantiated by contemporaneous medical record documentation. Codified at 42 CFR § 422.310 and § 422.311, a CMS RADV audit compares an enrollee’s submitted diagnosis codes against abstracted clinical records from the data collection year to recalculate the enrollee’s Risk Adjustment Factor (RAF) score and recover improper overpayments.
What Is a RADV Audit?
In the Medicare Advantage (Part C) program, CMS pays participating health plans a monthly capitated payment per enrollee adjusted for health status using the CMS-HCC risk adjustment model (which maps ICD-10-CM diagnosis codes to Hierarchical Condition Categories to compute a patient-specific RAF score alongside risk-adjusted outcome models and comorbidity indices such as the Charlson Comorbidity Index). Unlike traditional fee-for-service (FFS) Medicare—where payment is tied to procedure codes (CPT/HCPCS) or inpatient MS-DRGs—Medicare Advantage payments increase directly when additional chronic or acute HCCs are reported on inpatient, outpatient, or professional encounter claims.
Because unsupported or unsubstantiated diagnosis codes inflate RAF scores and generate improper federal payments, CMS conducts two types of RADV audits under 42 CFR § 422.310(e) and § 422.311:
- National Sample RADV Audit: Conducted annually on a statistically representative sample of enrollees across all Medicare Advantage contracts nationwide to estimate the overall national Medicare Advantage improper payment rate for federal financial reporting under the Payment Integrity Information Act (PIIA).
- Contract-Level RADV Audit: Targets specific Medicare Advantage contracts (identified via risk-score growth analytics, Office of Inspector General [OIG] referrals, or stratified sampling) to audit up to 201 sampled enrollees per contract and recover contract-specific overpayments.
How the CMS Contract-Level RADV Audit Process Works (Step by Step)
A CMS contract-level RADV audit evaluates medical records from the data collection year (the calendar year immediately preceding the payment year [PY]) across five operational stages:
- Contract Selection and Stratified Enrollee Sampling: CMS selects target MAO contracts and draws a stratified random sample of up to 201 eligible enrollees per contract—typically divided equally across three risk-score strata (Low, Medium, and High RAF scores)—who were continuously enrolled in the contract during the data collection year and January of the payment year.
- Medical Record Request and the Up-to-Five Chart Limit: CMS issues an audit notice listing each sampled enrollee and the specific CMS-HCCs under review. The MAO (working with its network hospitals and physician groups) has a defined submission window (historically up to 25 weeks) to retrieve and upload medical records from the data collection year. Under CMS RADV Medical Record Reviewer Guidance, an MAO may submit up to 5 medical record encounter notes per audited CMS-HCC (or substantiate the HCC via a disease-hierarchy manifestation or higher/lower severity code within the same HCC family).
- Independent Certified Coder Abstraction: CMS medical record review contractors use certified coders to abstract ICD-10-CM codes from the submitted face-to-face clinical notes strictly according to the ICD-10-CM Official Guidelines for Coding and Reporting and MEAT documentation criteria (Monitor, Evaluate, Assess/Address, Treat). Discrepancies trigger secondary and senior coder adjudication reviews.
- HCC Discrepancy Calculation and RAF Recalculation: If none of the submitted medical records validate a reported CMS-HCC (and no hierarchical substitute HCC is found in the chart), the HCC is flagged as a RADV discrepancy and removed from the enrollee’s profile, reducing the enrollee’s payment-year RAF score. (Conversely, if a submitted chart substantiates an unreported HCC within the audited scope, CMS offsets the error calculation accordingly.)
- Overpayment Determination, Extrapolation, and Recovery: CMS calculates the net payment error for each sampled enrollee and determines the contract’s recovery obligation under 42 CFR § 422.311.
The CMS RADV Final Rule (88 FR 6643): PY 2018 Extrapolation, the FFS Adjuster, and Federal Litigation
On February 1, 2023, CMS published the landmark Medicare Advantage Risk Adjustment Data Validation (RADV) Final Rule (88 FR 6643), resolving rulemaking first proposed in November 2018 and establishing two major regulatory policies that subsequently became the focus of federal Administrative Procedure Act (APA) litigation:
| RADV Policy Dimension | CMS 2012 Methodology Notice | CMS February 1, 2023 RADV Final Rule (88 FR 6643) | Post-2023 Federal Litigation Status (Humana Inc. v. Becerra) |
|---|---|---|---|
| Enrollee-Level vs. Extrapolated Recovery | Proposed contract-level statistical extrapolation beginning with PY 2011 audits. | Non-extrapolated (enrollee-sample only) recovery for PY 2011–PY 2017; contract-level extrapolation authorized beginning with PY 2018 audits and onward. | Extrapolation authority for post-2018 audits challenged alongside the elimination of the FFS Adjuster. |
| Statistical Extrapolation Methodology | Specified stratified ratio estimator across Low/Medium/High risk strata. | Declined to codify a single fixed statistical formula; allows CMS to apply any statistically valid sampling and extrapolation method suited to the specific audit design. | Subject to administrative appeal review under 42 CFR § 422.311(c) and judicial APA scrutiny. |
| Fee-for-Service (FFS) Adjuster | Included an FFS Adjuster to offset unaudited diagnosis coding errors in traditional Medicare Part A/B claims used to calibrate the CMS-HCC model under the statutory actuarial equivalence mandate (42 U.S.C. § 1395w-23(a)(1)(C)). | Eliminated the FFS Adjuster entirely, asserting that statutory actuarial equivalence does not apply to RADV obligation to submit accurate, medical-record-supported diagnoses under 42 CFR § 422.310. | In Humana Inc. v. Becerra (N.D. Tex., Sept. 2025), the federal district court vacated the 2023 RADV Final Rule on APA procedural notice-and-comment grounds over removal of the FFS Adjuster, triggering ongoing federal appeals. |
Common Medical Record Discrepancies That Fail a CMS RADV Audit
Both CMS RADV audits and HHS Office of Inspector General (OIG) Medicare Advantage risk adjustment audits consistently flag six categories of clinical documentation defects that invalidate an HCC—defects that also distort longitudinal real-world evidence (RWE) cohorts derived from claims data:
- Problem List Rollover Without Contemporaneous MEAT Evidence: A chronic diagnosis (such as Major Depressive Disorder, Vascular Disease, or Chronic Kidney Disease Stage 3–5) appears in an auto-populated EHR ‘Past Medical History’ or ‘Active Problem List’ header, but the encounter note contains zero evidence that the condition was monitored, evaluated, assessed, or treated during that face-to-face visit (violating ICD-10-CM Official Guideline Section IV.J/K).
- Unacceptable Provider Type or Non-Face-to-Face Encounter: Diagnosis codes abstracted from standalone diagnostic radiology, laboratory, pathology, durable medical equipment (DME), or ambulance claims—or from encounters not conducted face-to-face (or via CMS-eligible synchronous video telehealth) by an acceptable physician or qualified non-physician practitioner specialty—are invalid for RADV validation.
- Missing or Invalid Clinician Signature and Credential Authentication: CMS RADV reviewer guidance requires every submitted medical record to bear an acceptable handwritten or electronic signature (e.g., ‘Electronically signed by [Name, MD/DO/NP/PA]’ with date/timestamp) from the rendering provider; unsigned drafts or notes signed solely by uncredentialed scribes fail validation unless remediated with a CMS-compliant attestation where permitted.
- ‘History of’ vs. Active Disease Coding Errors: Coding an eradicated malignancy as an active primary or metastatic cancer HCC when the clinical narrative states the patient has completed adjuvant therapy with no evidence of disease (which requires a personal history Z-code,
Z85.x, rather than an active C-code). - Contradictory Clinical Indicators Within the Same Encounter: An EHR template lists ‘Diabetes with chronic kidney disease’ (
E11.22) in the assessment, while the physical exam and laboratory review explicitly state ‘normal renal function, no nephropathy.’ - Wrong Date of Service or Wrong Patient Identifier: Submitting encounter notes whose date of service falls outside the January 1–December 31 data collection year (for outpatient/professional claims) or whose patient name/date of birth does not match the sampled beneficiary.
The Three-Stage RADV Dispute and Appeal Process (42 CFR § 422.311(c))
When CMS issues a final RADV Audit Report notifying an MAO of medical record discrepancies and a calculated overpayment recovery amount, 42 CFR § 422.311(c) grants the MAO the right to appeal both the medical record review determination (disputing whether a specific submitted chart substantiates an HCC) and the payment error calculation within 60 days of receiving the audit report across three sequential administrative levels:
- Stage 1 — Reconsideration (42 CFR § 422.311(c)(6)): For medical record review appeals, a three-person panel of certified medical coders (independent of the initial RADV review) re-evaluates the disputed chart against ICD-10-CM guidelines. For payment error calculation appeals, an independent CMS reconsideration official reviews whether CMS misapplied its stated payment error calculation methodology.
- Stage 2 — Hearing Officer Review (42 CFR § 422.311(c)(7)): If dissatisfied with the reconsideration decision, the MAO may request a hearing within 60 days before the CMS Office of Hearings.
- Stage 3 — CMS Administrator Review (42 CFR § 422.311(c)(8)): Either the MAO or CMS may request discretionary review by the CMS Administrator within 60 days of the Hearing Officer’s decision; the Administrator’s decision (or declination to review) constitutes final agency action subject to federal judicial review.
Frequently Asked Questions
What is a RADV audit?
A RADV (Risk Adjustment Data Validation) audit is a federal compliance audit conducted by the Centers for Medicare & Medicaid Services (CMS) under 42 CFR § 422.310 and § 422.311 to verify that Hierarchical Condition Category (HCC) diagnosis codes submitted by Medicare Advantage plans are supported by valid, signed face-to-face medical record documentation from the data collection year. If submitted medical records do not substantiate an audited HCC under ICD-10-CM Official Guidelines, CMS removes the HCC, recalculates the enrollee’s Risk Adjustment Factor (RAF) score, and recoups the resulting overpayment.
How many enrollees and medical records are reviewed in a CMS contract-level RADV audit?
In a standard contract-level RADV audit, CMS selects a stratified random sample of up to 201 enrollees per audited Medicare Advantage contract (divided across low, medium, and high risk-score strata). For each sampled enrollee, the Medicare Advantage Organization may submit up to 5 medical record encounter notes from the data collection year to substantiate each audited CMS-HCC.
What did the 2023 CMS RADV Final Rule change regarding extrapolation and the FFS Adjuster?
In the February 1, 2023 RADV Final Rule (88 FR 6643), CMS announced that it would collect non-extrapolated (sample-only) overpayments for Payment Years 2011–2017, begin applying contract-wide statistical extrapolation starting with Payment Year 2018 RADV audits, and eliminate the Fee-for-Service (FFS) Adjuster. In September 2025, the U.S. District Court for the Northern District of Texas (Humana Inc. v. Becerra) vacated the 2023 Final Rule on Administrative Procedure Act notice-and-comment grounds concerning the removal of the FFS Adjuster.
