V28 was treated as a remapping exercise. It was a change in what documentation has to prove, and most organizations never revisited the practices under it.
Most organizations closed out the V28 transition the same way. Finance modeled the RAF impact, coding updated its target lists, provider education refreshed its slides, and the project moved to done.
That project answered which conditions still pay. It did not answer the harder question underneath the model change, which is what documentation now has to prove, and that one is still open at most plans and provider groups two years in.
The tell is simple. If your V28 response produced a new code list but no change to how a condition gets documented at the visit, you resized the target and left the process that misses it intact.
A Remap Is Not a Response
V28 did two things at once and most transition work only absorbed the first.
It removed a large share of diagnosis codes from risk-adjustment mapping and re-based the coefficients, which is the part everyone modeled. The financial exposure was legible, the remediation was obvious, and the work was mostly analytical.
It also raised what a payable condition has to demonstrate, which almost nobody modeled, because the exposure is invisible until an auditor or a rejected diagnosis surfaces it. Conditions that used to clear on a problem list entry or a medication reference now need specificity that a lot of routine documentation was never built to produce. The condition is present, the clinician knows it is present, and the note says something that mapped under V24 and does not map now.
The reason this stayed hidden is that it does not fail loudly. Under V24 a vague diagnosis still landed somewhere. Under V28 it lands nowhere, and the member simply carries a lower score with no error message anywhere in the process to explain it.
Where the Unfinished Work Actually Sits
Three places, all of them upstream of the coding department that usually inherits the problem.
Documentation templates and EMR defaults still reflect V24 economics. Any smart phrase, problem list convention, or note template built when a condition was payable in a general form is now producing documentation that no longer supports capture. These artifacts persist quietly because nobody owns them and they were correct when they were written.
Provider education still teaches condition lists rather than specificity. Telling a clinician which conditions matter was sufficient guidance under a model that paid for the condition. Under a model that pays for the documented characterization of the condition, a list is not the instruction that changes behavior.
Recapture logic still trusts last year. Suspect models built on prior-year HCCs carry a hidden assumption that what was captured before is capturable again. Across a model transition that assumption breaks in both directions, surfacing conditions that no longer map and missing ones that now require documentation the prior year never demanded.
Want to see where your V28 exposure actually sits?
The Measurement That Exposes It
Recapture rate by condition, compared against itself across the transition years, is the cleanest diagnostic available and most organizations already have the data to run it.
Chronic conditions do not resolve. So when a condition class recaptures at a materially lower rate than it did under the prior model, and the clinical population has not changed, the drop is documentation and coding practice rather than clinical reality. Looking at it in aggregate hides this, because the aggregate is dominated by conditions the model did not touch. Looking at it by condition class points directly at the documentation habits that stopped clearing the bar.
The second diagnostic is narrower and more uncomfortable. Take conditions your coders confirmed and that did not survive into the risk score, and sort them by provider and by note template. Failures cluster. They almost always trace to a small number of documentation patterns rather than to distributed clinician behavior, which is why broad provider education underperforms as a response.
Why This Compounds Into 2027
An unfinished V28 response was survivable while retrospective review could still catch the difference. That is changing on a published schedule.
The CY 2027 Rate Announcement ends unlinked chart reviews beginning with the 2027 payment year, so a chart review diagnosis has to tie back to an encounter CMS accepted, and the July HPMS memo on CY 2027 risk adjustment implementation sets out what has to be in place. A documentation practice that produces conditions the model no longer recognizes can no longer be patched by a sweep, because the sweep now depends on an accepted encounter that was never generated.
RADV moving toward every contract every year, with findings extrapolated across the contract, closes the other side of it. Fewer conditions carry payment under V28, which means each surviving capture bears more weight, and a capture built on documentation that barely clears the specificity bar is exactly the kind an auditor removes.
The organizations that finish the V28 work are not the ones with better code lists. They are the ones that changed what the note says.
How Invent Health Handles Model Change
Invent Health runs risk adjustment as one connected loop rather than three tools that integrate, and it starts with analytics, not coding.
Risk Analytics decides what is worth working. Suspect logic runs across historical, lab, pharmacy, comorbidity, and chart-derived signals, drawing on EMR data through CCD and FHIR. Conditions are ranked by how much they move the score and how likely they are to close, and internal capture is validated against CMS-recognized HCCs in the MOR with RAF progression tracked across payment cycles in the MMR, which is what makes recapture decay visible by condition rather than only in aggregate.
The Coder Workbench confirms the condition is real. AI-assisted coding reads the clinical documentation, shows the supporting evidence, and recommends ICD-10-CM codes for a certified coder to confirm and sign. Invent Health’s NLP detects that supporting evidence with 85 percent or higher accuracy out of the box, and the Coder Workbench averages 95 percent coding accuracy. It is a two-pass, coder-in-the-loop model, and every code stays linked to the chart for audit-defensible lineage.
Encounter Submissions makes sure it counts, generating and validating clean submissions, 837P, 837I, and DME for Medicare EDPS and Edge Server XML for ACA, flagging missing and non-risk-eligible diagnoses before a file leaves, then reconciling MAO-002 and MAO-004 for Medicare and the Edge Server reports for ACA.
The platform supports CMS-HCC V24 and V28, ESRD, RxHCC, and HHS-HCC for ACA, with new CMS or HHS models typically live within 30 to 60 days of advance notice. For organizations working with delegated groups and IPAs, the same reporting rolls up to the group level, which is where documentation practice actually changes.
Check the Condition Classes, Not the Aggregate
Run recapture by condition class across the transition years on a population you have already closed. Where a chronic class dropped and the clinical picture did not, the documentation practice underneath it is still producing V24 notes.
Frequently asked questions
What changed in the V28 risk adjustment model?
V28 removed a large share of diagnosis codes from risk-adjustment mapping, restructured and renumbered condition categories, and re-based coefficients. It also raised the documentation specificity required for many conditions to map at all, which is the change most transition projects measured financially but never addressed operationally.
Why did RAF scores drop under V28?
Partly because fewer diagnosis codes map to payable categories and coefficients were re-based. The less visible driver is documentation that satisfied V24 in a general form and no longer maps under V28, so conditions the population genuinely has stop producing score with no error surfacing anywhere in the process.
Does V28 change documentation requirements?
Effectively yes. The clinical standard for supporting a diagnosis did not change, but the specificity a note must contain for a condition to map did. Documentation templates and problem list conventions built under V24 often produce entries that no longer support capture.
How do you tell V28 impact apart from clinical change?
Compare recapture rates by condition class across transition years. Chronic conditions do not resolve, so when a class recaptures materially lower without a corresponding change in the population, the cause is documentation and coding practice rather than clinical reality.
Why is aggregate RAF a poor measure of V28 readiness?
Because the aggregate is dominated by conditions the model did not meaningfully change, which masks sharp declines in the classes it did. Condition-level analysis points at the specific documentation habits that stopped clearing the bar; the aggregate only says the number moved.
What should provider education cover under V28?
Specificity rather than condition lists. Telling clinicians which conditions matter was adequate when the model paid for the condition. When the model pays for the documented characterization of the condition, education has to address how the condition is described in the note.
Do EMR templates need updating for V28?
Frequently, yes. Smart phrases, note templates, and problem list conventions written under V24 economics still generate documentation shaped for a model that no longer exists. These artifacts rarely have an owner and are seldom revisited after a model transition.
How does V28 interact with suspect and recapture logic?
Suspect models built on prior-year HCCs assume what was captured before is capturable again. Across a model transition that assumption fails in both directions, surfacing conditions that no longer map while missing ones that now require documentation the prior year never demanded.
Does ending unlinked chart reviews affect V28 remediation?
Yes. Starting with the 2027 payment year, chart review diagnoses must tie to an accepted encounter, so documentation problems can no longer be patched retrospectively when no qualifying encounter exists. Remediation has to move upstream to the visit.
Why does V28 increase audit exposure per capture?
Because fewer conditions carry payment, each surviving capture represents more revenue. With RADV findings extrapolated across a contract, captures resting on documentation that barely clears the specificity threshold become the expensive ones to lose.
Do provider groups and health systems face the same V28 issues?
Yes, and often more directly, since they control the documentation itself. IPAs, medical groups, and health systems carrying risk own the templates and clinician workflows where V28 exposure originates, whether or not they hold the CMS contract.
How does Invent Health support a model transition?
Capture is validated against CMS-recognized HCCs in the MOR with RAF tracked across payment cycles, which surfaces recapture decay by condition. The Coder Workbench ties each confirmed code to its chart evidence, and the platform supports V24 and V28 with new models typically live within 30 to 60 days of advance notice.
