The CMS-HCC V28 Model Transition: What Changed and How to Prepare Your Coding Team
CMS phased in the V28 risk adjustment model beginning with payment year 2024, blending it with V24 at a 33/67 ratio in year one, 67/33 in year two, and moving to full V28 weighting for payment year 2026. That phased schedule is not a grace period. It is a window during which coding teams working from V24 assumptions are silently miscapturing risk on a significant share of their Medicare Advantage population.
The CMS-HCC V28 transition is not a cosmetic update. The model was reconstructed from the ground up using more recent Medicare claims data, updated ICD-10-CM coding conventions, and a clinical philosophy that rewards specificity over breadth. If your coders and CDI staff are still working from V24 crosswalks, old suspecting output, or training materials that predate the transition, your RAF scores are drifting away from actual patient acuity right now, and you may not see it until a plan-level reconciliation or RADV audit surfaces the gap.
What Actually Changed at the Mechanical Level
Expanded HCC Category Count and Restructured Groupings
V24 contained 86 HCC categories. V28 expanded that to 115. The increase is not just addition. CMS split existing categories, consolidated others, retired some entirely, and introduced new ones that reflect clinical distinctions that V24 treated as equivalent.
Diabetes is the clearest illustration. Under V24, diabetes with chronic complications mapped to HCC 18. Under V28, the model separates diabetes with kidney complications, with neurological complications, with ophthalmologic complications, and with circulatory complications into distinct categories, each with its own coefficient. A coder who documents and captures diabetes generically, the way that was acceptable under V24, will assign the wrong category under V28, or assign a lower-weighted one, or miss the capture entirely if the ICD-10-CM code they select no longer maps to an active V28 HCC.
Code-Level Crosswalk Changes
This is where the operational damage happens. The V28 model remapped ICD-10-CM codes across the restructured HCC hierarchy. Some codes that previously triggered a high-weight HCC under V24 now map to a different, lower-weight category. Some codes were dropped from HCC mapping entirely. Some codes that had no HCC mapping under V24 now map to a V28 category.
CMS published the full crosswalk files, but a published file is not a trained coder. The practical risk is that coders relying on institutional memory, old quick-reference sheets, or EHR templates that haven't been updated will select codes that feel correct but crosswalk differently than expected under the new model.
Coefficient Changes Within Retained Categories
Even where a category name stayed the same or similar, the relative additive RAF coefficient assigned to it changed. Categories that were weighted heavily under V24 may carry a lower coefficient under V28, and vice versa. Accurate capture of the right ICD-10-CM code matters more than it ever did, because the financial consequence of choosing the wrong code within a condition area is now larger in some cases and smaller in others, depending on which side of a newly split category boundary your code lands on.
Why This Creates a Silent RAF Accuracy Problem
The problem is not that coders are making obvious errors. The problem is that a coder applying V24 logic under V28 looks like they are working correctly. They are documenting a condition, selecting a plausible ICD-10-CM code, and submitting a claim. The code exists. It may even map to some HCC. But it may map to the wrong one, or map with the wrong weight, or not map at all when the correct clinical specificity would have mapped to a meaningful category.
For a Medicare Advantage plan or a capitated provider group, this kind of systematic undercapture is invisible in day-to-day operations. It shows up at reconciliation, at RADV audit, or when a plan compares its risk scores against benchmarks and finds unexplained deterioration in RAF relative to prior years. By that point, missed submissions for prior periods may be unrecoverable, and the compliance question of whether suspended or incorrectly captured codes were submitted in good faith becomes relevant.
Teams doing risk adjustment and HCC coding for Medicare Advantage populations cannot afford to treat this as a background concern. The model is the payment mechanism.
What Has to Change on the Ground for Your Coding Team
Replace Every V24 Reference Tool
Quick-reference HCC cheat sheets, laminated crosswalk cards, EHR diagnosis code favorites lists, and any internal training document that maps clinical conditions to HCC categories needs to be reviewed against the V28 mappings and either updated or retired. This sounds straightforward. In practice, many organizations do not know how many informal reference tools their coders are actually using, because most of them were created by individual coders or CDI specialists over years and live in desk drawers, shared drives, or personal note files.
An inventory of existing reference materials is a prerequisite to replacing them. You cannot update what you have not identified.
Retrain on Category Logic, Not Just Code Lists
Giving coders a new crosswalk list and calling it training does not work. The V28 model reflects clinical reasoning: that diabetes with renal manifestations is materially different from diabetes with peripheral neuropathy, that both are different from uncontrolled diabetes without complication, and that the clinical documentation has to support the specific ICD-10-CM code that reflects the specific complication. Coders need to understand why the categories changed, not just which codes changed, so they can apply the logic when they encounter clinical documentation that doesn't fit neatly into a reference list example.
This is also where CDI program support becomes operationally critical. If the physician documentation says "diabetes with complications" without specifying which complication, a coder cannot assign the specific V28 code that reflects renal versus neurological involvement. CDI staff need to know what clinical specificity the V28 model requires before they can query providers effectively.
Update Suspecting and Gap-Closure Logic
Any suspecting tool, whether built into your EHR, purchased from a vendor, or maintained internally as a query list or analytics output, that was built against V24 HCC mappings is producing incorrect suspecting output today. A suspect that fires based on a V24 HCC assignment may point to a code that no longer maps to the same category. A suspect may fail to fire at all for a condition that V28 now recognizes in a separate, clinically distinct category that did not exist in V24.
For health plans and larger provider groups, re-running suspecting models against V28 mappings is not optional. It is the mechanism by which you identify conditions that are documented in the record but not captured at the specificity the new model requires.
You can find specific guidance on what a current-model audit should examine in our free HCC audit checklist.
How to Sequence the Transition Correctly
Sequencing matters because retraining coders on V28 logic before auditing your current coding state means you will not know what baseline error patterns you are correcting. You will also waste training resources on coders who may not be making the errors you assume they are making, while the actual problem patterns go unaddressed.
Step One: Audit Current Coding Against V28 Mappings
Pull a sample of claims coded over the past twelve months for your highest-acuity condition categories, diabetes with complications, chronic kidney disease, heart failure, COPD, major depressive and bipolar disorders, and run them against the V28 crosswalk. Identify where codes that were submitted under V24 assumptions map differently under V28. Quantify both undercapture and, critically, any overcapture that creates compliance exposure.
A coding quality audit structured around the V28 model gives you the baseline you need before you spend a dollar on retraining.
Step Two: Target Retraining at Identified Gaps
Once you know which condition areas and which coders or CDI staff are showing V28 discordance, retraining becomes focused rather than generic. Condition-specific education on diabetes complication specificity is more effective than a general "V28 overview" webinar, because coders can connect the training directly to the error patterns they have actually been making.
Step Three: Re-Run Suspecting Logic and Re-Audit
After retraining and reference tool updates, re-run your suspecting output against V28 mappings and pull a post-training audit sample. The goal is to confirm that corrected coding behavior is holding and that your suspecting tools are now surfacing the right gaps, not the V24 gaps you used to chase.
For more context on maintaining RAF accuracy through model transitions, see our related post on HCC RAF score accuracy and the documentation-side implications covered in our piece on Medicare Advantage RAF documentation.
The Risk of Doing Nothing
RAF scores that quietly drift from actual patient acuity do not announce themselves. A plan or provider group with a high-acuity Medicare Advantage population may see stable or even slightly improving risk scores while the true capture rate is declining, because the blend weighting during transition years partially masks V28 miscapture. When the blend reaches full V28 weighting in payment year 2026, the full impact of uncorrected coding practices lands at once.
Beyond payment consequences, there is a compliance dimension. Submitting diagnosis codes that do not reflect the most specific, supportable clinical picture, because your tools and training were built for a model that CMS has retired, is a documentation integrity issue. RADV auditors examine whether submitted codes are supported by the medical record and whether the correct HCC was assigned. An audit conducted under V28 criteria against claims coded under V24 assumptions will find discordance.
The CMS-HCC V28 transition required CMS to rebuild the foundation of how clinical complexity is measured and paid. Coding teams that treat it as a reference update rather than a workflow change will feel that difference in their RAF accuracy and eventually in their revenue.
If your organization is ready to assess where your coding team stands against V28 requirements and build a structured remediation plan, contact MedCodex Health to discuss a targeted V28 readiness engagement.