What CMS Is Actually Looking For When RADV Comes Knocking
Roughly one in four Medicare Advantage organizations that receive a Risk Adjustment Data Validation audit discover that at least some of their highest-weighted HCC codes do not survive medical record review, even when those codes were assigned by experienced, credentialed coders working from the actual chart. That pattern reveals something important: RADV audit preparation is not primarily a coding problem. It is a documentation problem that coding gets blamed for.
Understanding that distinction is the difference between an organization that weathers a RADV audit and one that faces repayment demands and corrective action plans.
What RADV Actually Validates
The mechanics of a RADV audit are specific. CMS selects a sample of enrollees whose submitted diagnoses contributed to RAF scores, then requests the medical records that were used to support those diagnosis codes. An independent review contractor then evaluates each record against a defined standard, not the coding standard, but the documentation standard.
To pass validation, a submitted HCC diagnosis must be supported by a medical record from a face-to-face encounter with an eligible provider during the data collection year. The diagnosis must be documented by the treating clinician, and the documentation must show that the condition was monitored, evaluated, assessed, or treated during that visit. Coders and CDI professionals commonly refer to this framework as MEAT criteria, though CMS uses its own language in the RADV medical record reviewer guidance.
Critically, the acceptable provider and encounter types are defined narrowly. Emergency department visits, inpatient records, outpatient facility visits, and physician office encounters generally qualify. A phone encounter does not. A patient portal message does not. A note generated purely to transmit lab results to a patient does not.
Why the Source of the Diagnosis Matters
CMS requires that the diagnosis appear in the body of a clinician-authored note, with enough clinical context to demonstrate it was addressed at that visit. A diagnosis carried in the patient's problem list, printed on the encounter superbill, or pulled from a prior note does not, by itself, satisfy this requirement unless the current visit note also addresses it independently.
This is where the gap between coding accuracy and RADV defensibility opens widest. A coder who correctly pulls a diagnosis from a problem list, finds an ICD-10 code that maps to an HCC, and submits it has done technically correct work under standard outpatient coding guidelines. If the face-to-face encounter note for that date of service contains no reference to the condition beyond its presence on a printed list, that submission will likely fail RADV validation.
Why Coding Accuracy Alone Is Not Enough
This is the point most organizations underestimate, and it costs them.
A code can be clinically appropriate, correctly selected under ICD-10-CM guidelines, mapped to the right HCC, and still fail a RADV audit because the underlying documentation does not demonstrate the condition was actively addressed during the encounter. The audit is not checking your coders. It is checking your physicians' notes.
Consider a patient with type 2 diabetes with diabetic chronic kidney disease, mapping to HCC 18 in the CMS-HCC Version 28 model, a high-weight condition. If that combination is documented clearly in the assessment and plan section of the encounter note, with the physician noting current lab values, ongoing medication management, or a nephrology referral, the submission is defensible. If the only support is the diagnosis appearing on a printed problem list that was scanned into the chart without any clinician narrative, it is not, regardless of whether the code itself is accurate.
Organizations that have built their HCC programs around retrospective coding without concurrent documentation improvement consistently see this gap materialize under audit. The risk adjustment and HCC coding process must be integrated with documentation review, not layered on top of it after the fact.
The Most Common RADV Failure Patterns
Problem List Coding Without Current-Visit Support
This is the single most frequent failure mode. A chronic condition is accurately listed on the patient's problem list. The coder, following standard practice, includes it in the encounter coding. The encounter note itself contains no assessment, management discussion, or clinical reference to that condition during that specific visit. The RADV reviewer sees a problem list and a code but no evidence the physician engaged with the condition on that date.
Copy-Forward Documentation
Physicians under time pressure frequently carry forward prior visit notes with minimal modification. When an assessment and plan section is copied verbatim from a note dated six months earlier, a RADV reviewer cannot confirm the condition was actually addressed at the current encounter. Worse, copied documentation that contains information inconsistent with the patient's current status, such as a medication that was discontinued, or a complication that was resolved, can raise questions about the accuracy of the entire record.
Diagnosis Coded from Test Results Alone
Coding a condition from a laboratory result, imaging report, or pathology finding without a corresponding clinician assessment documenting that diagnosis is a documentation deficiency that RADV will catch. ICD-10-CM guidelines require that a physician clinically establish a diagnosis. A positive A1C does not code to diabetes without a physician's documented clinical interpretation connecting that result to a diagnosis. A chest X-ray report noting cardiomegaly does not code to heart failure without a clinician's assessment doing the same.
Visits That Do Not Qualify as Face-to-Face Encounters
Telehealth, telephone encounters, and patient portal communications are sometimes used to support diagnoses in ways that do not meet RADV's face-to-face requirement for all encounter types. Knowing which encounter types are eligible under the applicable RADV guidance for the audit period in question is a basic readiness requirement that many organizations skip.
What a Defensible Process Looks Like Before an Audit Happens
The organizations that perform best under RADV scrutiny are not the ones that scramble hardest after receiving an audit notice. They are the ones that have built ongoing processes that produce RADV-defensible documentation as a byproduct of normal clinical workflow.
Concurrent and Near-Real-Time Documentation Review
Retrospective chart review, typically conducted weeks or months after an encounter, can identify coding opportunities but cannot fix the underlying documentation before the submission deadline. CDI program support delivered concurrently, either during or immediately after an encounter, gives physicians the opportunity to amend or clarify documentation while the encounter is clinically fresh. This is the most effective intervention for closing the gap between what was coded and what will survive validation.
Physician Query Processes That Close Gaps Before Submission
A compliant physician query process, one that is non-leading, clinically grounded, and properly documented, allows CDI specialists and coders to ask clinicians to clarify ambiguous documentation before the diagnosis is submitted to CMS. The key word is before. A query that prompts a physician to add specificity to a note, clarify that a comorbidity was addressed during the visit, or confirm a clinical relationship between a documented finding and a diagnosis is a legitimate and important tool. Queries conducted after submission in response to an audit are significantly riskier territory.
Retrospective Chart Audits Designed for RADV Defensibility
Standard coding quality audits check for coding accuracy: correct code selection, proper sequencing, appropriate specificity. A RADV-readiness audit checks something different. It asks whether each submitted HCC is supported by a face-to-face encounter note containing an eligible provider's clinical documentation that addresses the condition. These are not the same test.
Periodic internal audits designed to simulate RADV review logic, pulling records for your highest-weighted HCC submissions and evaluating them through a RADV lens rather than a coding accuracy lens, will surface documentation gaps that a standard quality audit misses entirely. Download our free HCC audit checklist to build that review process into your regular compliance calendar.
A coding quality audit structured with RADV defensibility criteria built in gives your organization a realistic picture of which diagnoses are genuinely at risk before an external reviewer sees them.
Audit Readiness as an Ongoing Discipline
RADV audit preparation is not an event. It is a posture.
Organizations that treat it as an event, assembling records, briefing staff, and reviewing documentation in the weeks following an audit notice, are already behind. CMS's ability to look back at encounter data from prior contract years means that by the time an audit notice arrives, the documentation supporting those encounters is often a year or more old. Physicians have moved. Notes have been finalized. Amendment windows have closed.
The practical implication is that the work required to survive a RADV audit must happen at the time of the encounter, not at the time of the audit. This requires treating documentation quality as a clinical operations priority, not just a billing or compliance afterthought. It requires physician education that explains RADV mechanics in plain clinical terms, not just coding rules. And it requires accountability structures that measure and report on documentation quality alongside coding productivity.
For more on the broader patterns that compromise HCC program integrity, see our post on common HCC coding mistakes and our detailed guide to medical coding audit preparation.
When a Chart Will Not Hold Up: The Ethical Boundary
Not every documentation gap can or should be resolved in favor of keeping a code. If a chart genuinely does not support a submitted diagnosis under the applicable documentation standard, the appropriate action is to delete the code and, where required, submit a corrected claim or payment adjustment. Attempting to defend a submission that does not meet the documentation standard through retroactive note additions, misleading queries, or creative interpretations of ambiguous language creates legal exposure that dwarfs the financial impact of a straightforward repayment.
This is not a hypothetical risk. False Claims Act liability attaches to knowingly retaining overpayments. The compliance calculus is straightforward: a defensible process that produces accurate RAF scores is an asset; a process that inflates RAF scores through documentation that would not survive scrutiny is a liability.
Honest RADV readiness sometimes means accepting that a condition your physician genuinely manages was not documented in a way that CMS will accept, and building the process changes that prevent that gap from recurring.
Building the Program That Survives Scrutiny
The organizations that consistently perform well under RADV audit share a common characteristic: they have stopped treating HCC capture as a purely coding function and started treating it as a joint documentation and coding discipline, with defined accountability, physician engagement, and prospective review built into the encounter workflow.
If your program is still primarily retrospective, primarily coding-focused, or lacks a structured process for evaluating documentation against RADV criteria before submission, the gap between what you have submitted and what you can defend may be larger than your current audit results suggest.
Contact the MedCodex team through our risk adjustment and HCC coding services page to discuss how a documentation-forward HCC program can be built into your existing clinical workflow before your next audit window opens.