HCC Risk Adjustment Coding: The 5 Documentation Gaps That Cost Practices the Most

Published 15 September 2026 · 9 min read

By Elvis J., CPC, CPPM. 20+ years in medical coding, clinical documentation, and healthcare compliance. Last reviewed 15 September 2026.

Who this is for: leaders at medical billing companies, RCM vendors, coding platforms, and practice management software companies who sell to medical practices, especially those in Medicare Advantage risk adjustment. This covers the business and compliance side of HCC coding. It does not cover clinical diagnosis, treatment decisions, or how to appeal a specific audit finding.

A risk adjustment factor score is only as good as the documentation behind it. A practice can see a patient with five active chronic conditions and still capture risk adjustment credit for zero of them, not because the care did not happen, but because the note did not prove it happened in the way CMS requires.

A documentation gap is any place where a diagnosis affecting a patient's health status was never captured for HCC risk adjustment, usually because the medical record did not clearly show the diagnosis being monitored, evaluated, assessed, or treated at that visit. Five gaps account for most of the lost or clawed-back revenue: missing MEAT support, failure to recapture chronic conditions every year, coding a resolved condition as active, using an unspecified code when a more specific one applies, and copying a problem list forward without new evidence.

If you sell coding software, CDI tools, or billing services into value-based or Medicare Advantage practices, these five gaps are what your buyer is actually losing sleep over.

What Actually Supports an HCC Diagnosis

A diagnosis on a problem list is not, by itself, a billable HCC. Coders and auditors use the MEAT standard: the note has to show the condition being Monitored, Evaluated, Assessed or addressed, or Treated at that encounter (AAPC, 2026). MEAT is not written into the official ICD-10-CM coding guidelines. It does not need to be. Every Medicare Advantage plan, every CMS RADV auditor, and every clinical documentation improvement program in the country uses it as the working test for whether a diagnosis is supported.

A note that lists "diabetes, hypertension, CKD" with no other comment is a problem list, not an assessment. Official coding guidance is clear that a bare list does not meet the definition of an assessment and plan (AAPC, 2026). If a coder cannot point to a sentence showing the provider actually addressed the condition, the diagnosis does not count, no matter how real the condition is.

Why 2026 Raises the Stakes

The model that turns a diagnosis into a dollar amount changed underneath the industry, and 2026 is the year the change finished. CMS is now using 100 percent of the risk score calculated under the 2024 CMS-HCC risk adjustment model, commonly called V28, for non-PACE Medicare Advantage organizations, completing a three-year phase-in that started at 33 percent in 2024 (CMS, CY 2026 Rate Announcement).

V28 was not a patch on the old model. It was rebuilt on more current claims data and reorganized which diagnoses map to which HCC categories. Some conditions that used to carry weight no longer do, and the total number of HCC categories dropped. A practice that has not updated its documentation habits since the old model, or a coding platform that has not updated its logic, is now scoring against rules that shifted under it.

The 5 Documentation Gaps That Cost the Most

1. No MEAT support in the note

This is the most common gap and the easiest one for a RADV auditor to find. The condition is real, the patient has it, and the provider even thinks about it during the visit, but nothing in the note shows monitoring, evaluation, assessment, or treatment. A diagnosis copied into an assessment line with no supporting sentence is a diagnosis a coder should not code and an auditor will not honor.

2. Chronic conditions not recaptured every calendar year

HCC risk adjustment does not carry a diagnosis forward from last year's chart. Every chronic condition that still affects the patient has to be documented and coded again at a visit during the current calendar year, or the risk score drops to zero for that condition on January 1, regardless of how long the patient has had it. Practices that treat chronic disease management as a formality rather than an annual documentation event lose recapture on conditions that never went away.

3. History-of conditions coded as active

This is the single gap CMS auditors have spent the most money chasing. In one federal audit, all 97 sampled Medicare Advantage enrollees had an acute stroke diagnosis code submitted to CMS that was not supported by any inpatient or outpatient record showing an acute stroke in that service year, and OIG estimated the resulting overpayment at $462 million for a single payment year (HHS-OIG, Report A-02-23-01020, issued May 2026). The pattern behind that number is simple: a patient had a stroke years ago, it still appears on the problem list as "stroke," and it gets coded every year as if it were still an acute event instead of a history-of condition with no current active management. The same failure mode shows up with old cancers, past heart attacks, and resolved fractures.

4. Unspecified codes where a more specific one applies

Risk adjustment models reward specificity because more specific diagnoses are tied to more predictable costs. A coder who defaults to an unspecified code when the documentation actually supports a more specific one, a complication, a manifestation, a particular stage, can undercode the true risk of the patient. This is not about inflating anything. It is about making sure the code matches what the note actually says, including complications the provider documented but the coder did not carry through to the final code set. Coders should reference the specific condition category, not quote code descriptors verbatim from the ICD-10-CM manual, since those descriptors are part of the classification system and should be applied, not copied into patient-facing or marketing material.

5. Copy-forward documentation with no new evidence

Electronic health records make it simple to pull last visit's assessment into this visit's note with one click. That habit is where a lot of MEAT support quietly disappears. If the copied text describes what was true three months ago and nothing in today's note shows the condition was actually addressed today, an auditor reading the chart cannot tell today's encounter from a template. The diagnosis might still be real and still be relevant, but the note no longer proves it for this date of service.

What This Costs, in Real Numbers

The $462 million figure above is one audit, one diagnosis category, one payment year, across multiple Medicare Advantage organizations, not a number every practice should expect to see (HHS-OIG, Report A-02-23-01020, 2026). What it shows is the scale CMS is willing to pursue when a documentation gap becomes a pattern across a population, and RADV audits are how CMS finds that pattern. A gap that looks small at the level of one chart becomes a nine-figure number at the level of a health plan's full membership, which is exactly why plans push documentation standards down to the practices and vendors that feed them data.

The same upstream logic applies to why clean claim rate predicts cash flow better than denial rate: the earlier a problem is visible in the workflow, the cheaper it is to fix. A documentation gap caught at the point of care costs nothing. The same gap caught in a RADV audit two years later costs a repayment.

What This Means If You Build or Sell to Practices

If your product touches coding, CDI, or risk adjustment reporting for practices in value-based care or Medicare Advantage, these five gaps are the actual buying triggers behind a purchase decision, more than any feature list. This is also the kind of specificity that belongs in your own content built around what your buyers actually get audited on, not just your product docs.

• Sell against the specific gap, not "better coding." A practice administrator understands "your notes aren't recapturing conditions every January" faster than "our AI improves documentation quality."
• Make sure your own logic reflects V28, not the retired model. A platform still scoring against the old HCC categories will quietly mismatch what CMS actually pays for in 2026.
• MEAT support is a feature request waiting to happen. Prompting a provider in real time when a diagnosis on the assessment lacks a MEAT element is one of the more defensible product claims in this space, because it is checking documentation mechanics, not making a clinical judgment.
• Do not promise an audit-proof result. RADV outcomes depend on medical record review by CMS contractors. The honest claim is that better documentation habits reduce exposure, not that any tool eliminates audit risk.

Frequently Asked Questions

What does MEAT stand for in HCC coding?

Monitor, Evaluate, Assess or Address, and Treat. A diagnosis needs documentation showing at least one of these four elements at the encounter to count toward HCC risk adjustment (AAPC, 2026).

Does an HCC diagnosis carry over automatically from last year?

No. Chronic conditions have to be documented and coded again at an encounter during the current calendar year, or the condition drops out of the risk score at the start of the next year.

What is the current CMS-HCC risk adjustment model?

CMS is using 100 percent of the risk score calculated under the 2024 CMS-HCC model, commonly called V28, for non-PACE Medicare Advantage organizations in 2026, completing a phase-in that began in 2024 (CMS, CY 2026 Rate Announcement).

Why do history-of conditions get flagged in audits so often?

Because they are easy to miscode and easy for auditors to catch. A past stroke or resolved cancer coded as a current, active condition is one of the most common patterns CMS audits have found, including a 2026 federal audit that found this exact pattern in 100 percent of its sampled cases for one diagnosis category (HHS-OIG, Report A-02-23-01020, 2026).

The Bottom Line

None of these five gaps require better clinical care. They require better documentation of care that is likely already happening. That distinction matters for anyone selling into this market: the pitch is not "treat patients differently," it is "prove what you already did." Products and content that make that distinction clearly tend to land better with practices than ones that treat documentation and clinical quality as the same problem.

If you sell into healthcare

Enovadi Systems writes content for companies that sell into healthcare, built on 20+ years of coding, billing, and compliance experience rather than a general content template. Talk to us about content built around what actually gets audited.