The Centers for Medicare & Medicaid Services (CMS) just issued a serious wake-up call to all Medicare Advantage (MA) plans: Starting now, every eligible MA contract will be audited annually, and CMS is rapidly closing the books on a backlog of Risk Adjustment Data Validation (RADV) audits stretching from 2018 to 2024.
This change marks a sharp pivot from a previously slow and selective audit approach to an aggressive, full-scope enforcement strategy. So, what’s driving this shift? And what exactly are “unsupported diagnoses,” the root cause of billions in overpayments?
Let’s break it all down.
What changed? A new era of RADV enforcement
For years, CMS lagged in RADV audits. The last significant recovery of MA overpayments came from audits on 2007 data. Meanwhile, federal estimates suggest MA plans may be overbilling the government by $17–$43 billion each year.
CMS is now scaling up fast:
-
- All MA contracts will be audited every year going forward.
- Audits from 2018–2024 will be completed by early 2026.
- The coding team is expanding from 40 to 2,000 staff.
- Each audit will now include 35–200 charts per plan, depending on size.
- Technology and AI will help flag suspicious/unsupported diagnoses in submitted claims.
The goal? To recover overpayments and hold plans accountable for the accuracy of the diagnoses they submit.
Why are unsupported diagnoses the target?
If you’re reading this, you probably already know that MA plans are reimbursed based on their members’ RAF scores, which are calculated using the Hierarchical Condition Categories (HCC) model, and that the plans get reimbursed more money to provide care for their sicker patients (i.e., those that have more conditions or more severe conditions).
But here’s the catch: Those diagnoses must be clinically documented and relevant to the patient’s care. If a diagnosis is not supported by medical documentation, it’s considered “unsupported”—and therefore ineligible for payment.
That’s why RADV audits zero in on one thing: Are the diagnoses you’re getting paid for actually supported in the chart?
It’s not enough for a condition to exist, it has to be documented with clinical relevance and intent to treat.
But in practice, that’s where things get confusing. Many unsupported diagnoses aren’t the result of fraud; they’re just the result of gaps in everyday workflows that go unchecked.
Let’s look at 4 common ways unsupported diagnoses show up in charts and how they can put your plan at risk.
4 Common use cases (and misuses) of unsupported diagnoses
Let’s explore where unsupported diagnoses appear and how they impact value-based care programs.
1. Auto-populated diagnoses (without clinical relevance)
Many EHRs allow auto-populating diagnoses from prior visits. While this speeds up documentation, it often leads to chronic or resolved conditions being pulled forward without evaluation or treatment, making them unsupported.
Example: A patient’s chart still lists “Major depressive disorder, moderate, recurrent episode” in the assessment and plan section of the note, but there’s no assessment, treatment, or mention in today’s note. That HCC should not be added to today’s note or submitted.
2. Historical diagnoses without ongoing impact
Diagnoses must be active during the encounter. Resolved conditions without current relevance can’t justify payment unless specified and supported.
Example: A patient had prostate cancer treated 5 years ago and is now cancer-free. If the patient is not on adjuvant therapy, then the active cancer diagnosis is unsupported and should be switched to history of prostate cancer.
3. Coding from lab values or imaging alone
Coders sometimes extract diagnoses from labs or imaging reports. However, a diagnosis needs to be made and documented by an eligible provider —typically MDs, DOs, NPs, or PAs—in a face-to-face visit, not inferred from a value. Diagnoses from scribes, MAs, or unlicensed staff aren’t valid.
Example: An eGFR of 55 appears in labs, but if the provider doesn’t document CKD stage 3a in the note, it’s not valid for risk adjustment.
Sign up to our DoctusTech VBC Hub to receive tailored content recommendations direct to your inbox.
Sign up to our DoctusTech VBC Hub to receive tailored content recommendations direct to your inbox.
4. Conditions without evaluation, assessment, or plan
Even if a condition is listed in the note, it’s unsupported if it isn’t evaluated or addressed. The recommended standard is the MEAT criteria (Monitoring, Evaluation, Assessment, Treatment).
Example: “Heart failure” is on the diagnosis list, but there’s no mention of symptoms, exam findings, or medication adjustments. That’s an audit risk.
What’s at stake?
CMS is not just validating diagnoses—they’re using audit results to extrapolate overpayments across entire plans. That means a few unsupported diagnoses in a 100-record sample could result in millions in repayment demands.
Plans that haven’t taken RADV seriously will be playing catch-up fast. The cost of non-compliance is no longer theoretical—it’s actionable.
How can you be prepared?
If you’re in population health, risk adjustment, or care delivery leadership, now is the time to:
1. Retrain providers on MEAT documentation
Help them understand which documentation is compliant for HCC coding and which isn’t.
2. Audit high-impact diagnoses internally.
Use retrospective chart audits to flag unsupported codes before CMS does.
3. Invest in Clinical Documentation Improvement (CDI)
CDI teams can help bridge gaps between documentation and coding accuracy.
4. Scrutinize vendor-coded charts
Not all external coders understand the nuance of HCC rules. Spot check their work.
5. Embrace RADV readiness as a continuous process.
With yearly audits now guaranteed, RADV prep isn’t a one-time event. It’s an operating requirement.
6. Leverage AI tools that do the heavy lifting.
AI-powered platforms like DoctusTech can transform your RADV readiness by checking 100% of charts in real time, ensuring HCCs are properly supported, and more.
Final thoughts: This is about accountability
CMS is sending a clear message: If you want to participate in a program that rewards managing patient risk, your documentation must prove it.
Unsupported diagnoses erode trust in the MA program and jeopardize funding meant for high-risk care. On the other hand, accurate documentation builds a sustainable model that rewards true clinical complexity and improves outcomes.
RADV audits aren’t just about clawbacks. They’re about accountability.
And for MA plans committed to integrity and quality care, that’s an opportunity, not a threat.
At DoctusTech, we help provider organizations and health plans improve HCC coding accuracy without adding an administrative burden to clinical teams.
With our mobile learning app, providers get brief, case-based training that fits into their day, reinforcing how to document and code diagnoses per CMS’s expectations. Our platform helps:
-
- Reinforce MEAT documentation standards using real-world scenarios
- Reduce unsupported diagnoses through more innovative education, not more paperwork
- Improve RAF accuracy while staying compliant with V28 and RADV rules
It’s not just about coding education. It’s about empowering providers to get it right the first time, which protects your organization from overpayment risk and helps ensure patients receive the care they need.
Want to see how it works? Schedule a demo and see how DoctusTech helps you stay ahead of CMS’s RADV expansion.