A chart shows a patient with diabetes, chronic kidney disease, and congestive heart failure. All three conditions are documented in the patient’s history, but the note for this particular visit clearly shows active management of only two of them.
The diagnosis itself may be accurate. The problem is that the current medical record does not provide enough support for reporting it for risk adjustment purposes that year. A true diagnosis with weak supporting documentation and an inaccurate diagnosis can both fail a review for the same underlying reason: the chart doesn’t back up the code.
This is what makes risk adjustment coding harder than most people expect going in. It’s less about knowing which ICD-10 code maps to which HCC category, and more about building documentation habits precise enough to survive a review that happens months or years after the visit.
How MEAT Helps Coders Evaluate HCC Documentation
MEAT (Monitor, Evaluate, Assess, or Treat) is a commonly used framework for reviewing whether documentation shows that a reportable condition was addressed during the encounter, rather than simply carried forward on a list. A condition listed on a problem list without any of those four elements documented for the current encounter is generally not a strong candidate for risk adjustment coding, regardless of how clinically real the condition is.
This trips up practices constantly because it’s easy to assume that referencing a condition is the same as documenting it. Reviewing recent labs tied to kidney function, adjusting a medication dose for heart failure, or evaluating a diabetic patient’s neuropathy symptoms during the visit all support MEAT. Simply carrying a diagnosis forward on the problem list, with no visible connection to what happened during that specific encounter, does not.
Coding Historical Conditions as Active
This is one of the most consistently flagged errors in recent federal audits, and it’s a genuinely easy mistake to make: a condition that resolved in the past gets coded as if it’s still active. A stroke from several years ago gets submitted as an acute stroke. A cancer that was treated and resolved gets coded as a current malignancy.
The scale of this problem became public in a notable way in 2026. In one OIG audit, none of the high-risk acute stroke diagnosis codes submitted for the 97 sampled enrollees were supported by the associated medical records. Based on those findings, OIG estimated approximately 462 million dollars in potential net overpayments to Medicare Advantage organizations for 2021. That’s a documented pattern serious enough to draw federal scrutiny, and stroke and myocardial infarction diagnoses have been particularly prone to this history-versus-active confusion because the acute event and its long-term aftermath often get described in similar language in the chart.
The fix is a coding habit, not a coding rule: confirm whether a condition is currently active and being managed, or whether it’s a historical event with lingering effects, before assigning a code that implies it’s still acute.
Hierarchy Suppression Errors
HCC models are built around hierarchies, where a more severe version of a condition suppresses payment for a less severe version of the same condition. When conditions fall within the same HCC hierarchy, the more severe category can supersede the less severe category in the risk score calculation. Coders therefore need to understand the hierarchy rather than assuming that reporting every historical severity level will increase risk adjustment value.
This isn’t really a documentation problem so much as a coding logic problem: understanding which conditions in a patient’s chart interact hierarchically, and coding to the correct, single, most specific level rather than layering in every version of a diagnosis that’s ever appeared in the record.
Why Do HCCs Need Annual Recapture?
HCCs don’t roll over automatically from one year to the next. For diagnoses that are relevant to the applicable risk adjustment model, conditions generally need current-year documentation and qualifying diagnosis data to be captured for that year’s risk score, even if nothing about the patient’s condition has changed. A diabetic patient managed consistently for the past five years still needs that diagnosis reflected in this year’s documentation, not just referenced as an ongoing historical fact.
Practices that don’t build recapture into their annual workflow tend to see RAF scores drift downward over time, not because patients are getting healthier, but because conditions that are still very much present simply aren’t being re-documented on schedule.
Defaulting to Unspecified Codes
When documentation supports a specific diagnosis but the coder (or the EHR’s suggested code) defaults to an unspecified version instead, that unspecified code may fail to map to a payable HCC at all, or it may provide less risk-adjustment value than a more specific code the documentation actually supports. This tends to happen under time pressure, when a specific code requires a bit more digging through the note than an unspecified one does.
CMS-HCC V28 Is Fully Phased In for 2026
For CY 2026, CMS has completed the phase-in of the 2024 CMS-HCC model, commonly referred to as V28, for Medicare Advantage organizations other than PACE. CMS now calculates 100% of the risk score using this updated model, which replaced the prior model’s condition groupings and weights. Coders and practices still working from older reference materials or habits built around the previous model’s category structure risk misclassifying conditions that were regrouped or reweighted under V28. Keeping coding references current with the active model, rather than relying on training or cheat sheets that predate the phase-in, matters more this year than in a typical year specifically because of that changeover. It’s worth noting V28 applies to Medicare Advantage; other programs, including ACOs and PACE, work under their own applicable risk adjustment models.
A Quick Prevention Checklist
| Issue | Prevention Habit |
| Documentation doesn’t show the condition was addressed | Confirm the note shows monitoring, evaluation, assessment, or treatment for the specific encounter, not just a problem list entry |
| Historical conditions coded as active | Verify current activity status before coding an acute or active version of a past condition |
| Hierarchy-related coding errors | Code to the correct, single, most specific severity level rather than layering multiple versions of one condition |
| Missed current-year diagnosis capture | Build a systematic yearly review to confirm active conditions relevant to the model are re-documented and re-coded |
| Unspecified code defaults | Query the provider for more specific documentation before defaulting to an unspecified code |
| Outdated model references | Confirm coding references reflect the current CMS-HCC V28 model rather than a prior version |
Conclusion
The instinct in risk adjustment coding used to be about capture: finding every codable condition in the chart. That instinct hasn’t disappeared, but it’s no longer sufficient on its own. Recent federal audit activity reinforces an important point: a diagnosis submitted for risk adjustment must be supported by the medical record and applicable coding and risk-adjustment requirements, tied to a specific encounter and coded at the correct level of severity.
Practices and coders who stay ahead of this aren’t necessarily the ones finding the most diagnoses. They’re the ones building habits that make every diagnosis defensible: checking for MEAT elements before coding, confirming active versus historical status, respecting hierarchy rules, and treating annual recapture as a scheduled task rather than an afterthought.
