What is RADV extrapolation?

RADV extrapolation means CMS applies the error rate found in the audit sample (typically 200 beneficiaries) across the entire MA contract population. Starting with payment year 2018 audits, this amplification makes even a small error rate result in significant payment recovery across tens of thousands of beneficiaries.

How many beneficiaries are selected in a RADV audit?

CMS typically selects approximately 200 beneficiaries per contract for a RADV audit. For each beneficiary, CMS reviews the submitted HCC diagnoses and requests the supporting medical records. The findings from this sample are then extrapolated to the entire contract population.

Can a coder be held personally responsible for RADV audit findings?

Coders are not personally liable for RADV payment recovery — that falls on the MA organization. However, coders who consistently submit unsupported diagnoses may face internal corrective action, and patterns of intentional upcoding can trigger separate Fraud, Waste, and Abuse (FWA) investigations by the OIG.

How long do MA organizations have to submit medical records for RADV?

Once CMS issues a RADV medical record request, MA organizations typically have 60-90 days to submit the requested records. Extensions may be granted in some cases, but failure to submit records results in automatic denial of the associated HCC diagnoses.

What is RADV extrapolation?

RADV extrapolation means CMS applies the error rate found in the audit sample (typically 200 beneficiaries) across the entire MA contract population. Starting with payment year 2018 audits, this amplification makes even a small error rate result in significant payment recovery across tens of thousands of beneficiaries.

How many beneficiaries are selected in a RADV audit?

CMS typically selects approximately 200 beneficiaries per contract for a RADV audit. For each beneficiary, CMS reviews the submitted HCC diagnoses and requests the supporting medical records. The findings from this sample are then extrapolated to the entire contract population.

Can a coder be held personally responsible for RADV audit findings?

Coders are not personally liable for RADV payment recovery — that falls on the MA organization. However, coders who consistently submit unsupported diagnoses may face internal corrective action, and patterns of intentional upcoding can trigger separate Fraud, Waste, and Abuse (FWA) investigations by the OIG.

How long do MA organizations have to submit medical records for RADV?

Once CMS issues a RADV medical record request, MA organizations typically have 60-90 days to submit the requested records. Extensions may be granted in some cases, but failure to submit records results in automatic denial of the associated HCC diagnoses.

RADV Audit Preparation 2026: Complete Guide for MA

Prepare for CMS RADV audits in 2026. Understand the audit process, documentation requirements, common denial reasons, and how to build an audit-ready coding program.

What Is RADV?

Risk Adjustment Data Validation (RADV) is CMS's audit program for verifying that diagnoses submitted by Medicare Advantage (MA) organizations for risk adjustment payment are supported by medical record documentation. RADV audits review a sample of beneficiaries and their submitted HCC diagnoses to determine whether each diagnosis meets coding and documentation standards.

When a RADV audit finds that a submitted diagnosis is not supported, the associated HCC and its RAF value are removed from the beneficiary's risk score. This results in payment recovery (clawback) from the MA organization. Beginning with payment year 2018 audits (conducted in 2024-2026), CMS applies extrapolation — meaning audit findings from the sample are applied across the MA contract's entire enrolled population.

RADV Audit Timeline for 2026

CMS is currently conducting RADV audits for payment years 2018-2020, with payment year 2021 audits expected to begin in 2026. The audit cycle typically follows this timeline:

Year 0 (Payment Year): MA organization submits encounter data with HCC diagnoses

Year +2 to +3: CMS selects audit sample (approximately 200 beneficiaries per contract)

Year +3 to +4: MA organization submits medical records for selected beneficiaries

Year +4 to +5: CMS completes coding review and issues preliminary findings

Year +5 to +6: Appeals process and final payment recovery

The long lag between the payment year and the audit means that documentation created in 2026 will not be audited until approximately 2029-2030. However, the documentation standards applied will be the standards in effect at the time the encounter occurred — not the standards at audit time.

Common RADV Denial Reasons

1. Missing MEAT Documentation (40-50% of denials)

The most common reason for RADV denials is a diagnosis listed in the assessment or problem list without any evidence that the provider monitored, evaluated, assessed, or treated the condition during the encounter. See our MEAT criteria guide for detailed examples.

2. Unsupported Specificity (20-25% of denials)

The ICD-10 code submitted requires a level of specificity that the documentation does not support. For example, submitting E11.22 (Type 2 diabetes with diabetic chronic kidney disease) when the note says "diabetes" without mentioning kidney disease.

3. One-Way Diagnosis (10-15% of denials)

A condition is documented during one encounter but never appears again in subsequent visits. For chronic conditions like diabetes or COPD, a single documentation instance raises questions about whether the condition is genuinely active. CMS does not require chronic conditions to be documented at every visit, but a complete absence in all other encounters within the data collection period weakens the audit position.

4. Copy-Forward Contamination (10% of denials)

Encounter notes that contain identical text across multiple visits suggest the documentation was auto-populated from prior encounters rather than reflecting current clinical assessment. RADV reviewers flag copy-forward patterns as potential documentation integrity issues.

5. Incorrect Code Assignment (5-10% of denials)

The ICD-10 code is simply wrong for the documented condition. This includes coding to a higher specificity than documented, assigning a combination code when only one component is documented, and using an outdated code for the fiscal year.

Building an Audit-Ready Program

Prospective Coding Quality

The best RADV defense is accurate coding at the point of submission:

Use current V28 mapping tools — verify every ICD-10 code maps to the intended HCC before submission using tools like the HCC Buddy encoder

Apply MEAT validation at chart review — before submitting a diagnosis for risk adjustment, confirm MEAT documentation exists in the encounter note

Code only what is documented — never assign a code based on clinical inference if the provider has not explicitly documented the condition

Track your coding accuracy — audit a sample of your own work quarterly and calculate your accuracy rate

Retrospective Audit Readiness

Prepare for eventual RADV review:

Maintain a complete medical record — all encounter notes, lab results, imaging reports, and specialist consultations should be accessible and organized

Implement chart addendum policies — if a provider needs to add missing documentation, the addendum should be timely (within 30 days of the encounter), clearly dated, and explain the clinical basis for the addition

Run internal audits — select 30-50 charts per quarter and review submitted HCC diagnoses against MEAT criteria; track denial patterns and address root causes

Provider Documentation Improvement

Work with providers to improve documentation quality:

MEAT education sessions — quarterly training on documentation standards with before/after examples

EHR template optimization — ensure templates prompt providers to address each chronic condition with at least one MEAT element

Real-time coding feedback — when chart reviewers identify documentation gaps, provide feedback to the rendering provider within 48 hours

RADV and Extrapolation

Starting with payment year 2018 audits, CMS applies extrapolation to RADV findings. This means the error rate found in the audit sample (typically 200 beneficiaries) is applied to the MA contract's entire enrolled population.

Example: If a contract has 50,000 beneficiaries and the RADV audit sample shows a 12% HCC diagnosis error rate, the payment recovery is calculated by applying the 12% error rate across all 50,000 beneficiaries — not just the 200 in the sample. This amplification makes even a small audit error rate financially devastating.

Extrapolation makes accurate prospective coding and documentation more important than ever. A single coder consistently submitting unsupported diagnoses can create an error pattern that, when extrapolated, results in millions of dollars in payment recovery.

Key Resources

CMS RADV Audit Methodology — available at cms.gov

OIG Reports on Medicare Advantage — the Office of Inspector General publishes regular reports on MA risk adjustment compliance

HCC Buddy — use the encoder and RAF Calculator to verify code-to-HCC mappings and model risk score impact

RADV Audit Preparation 2026: How to Build an Audit-Ready Risk Adjustment Program