The Centers for Medicare & Medicaid Services (CMS) introduced CMS-HCC Version 28 (V28) in 2024 as the next evolution of Medicare Advantage (MA) risk adjustment. For many organizations, the “new CMS-HCC model” felt more theoretical than operational, something plans were preparing for but not fully experiencing. That changes in 2026, when V28 becomes operational at scale.
CMS has now fully implemented V28 for MA payment, replacing Version 24 after nearly a decade of use. The transition period is over. Risk scores and therefore revenue, are now calculated entirely using the new model. This isn’t just a technical update. It changes how patient illness is measured, how documentation affects payment, and how plans and providers need to operate going forward. Complicating matters, CMS has recently proposed a new V28 recalibration in the 2027 Advance Notice for Medicare Advantage that would again adjust the calibration factors.[1] Although this proposal has not yet been finalized, it reinforces a key message: the old playbook for managing risk no longer works. Organizations must adapt, because the rules will continue to evolve
In this post, Rick Gordon, Director of Government Programs at MedInsight, and Marcos Dachary, Chief Market Strategist and General Manager, Payer Solutions at MedInsight, break down what’s changed, why it matters, and what to do next.
First, a quick refresher: How risk adjustment works
Medicare Advantage plans are paid based on how sick their members are expected to be. CMS uses diagnosis codes submitted by providers to calculate a risk score, often called a risk adjustment factor (RAF) score. In general, the more accurately documented a population’s disease burden is, the higher the payment.
CMS periodically updates the risk model to better reflect real medical costs and coding accuracy. The 2024/v28 update is one of the most significant in years.
According to CMS, the goal was to improve payment accuracy by removing diagnoses that were weak predictors of cost and requiring more clinically specific documentation.[2]
CMS-HCC V28: What actually changed
Here’s the simplest way to understand V28:
- Some diagnoses that used to count toward payment no longer do.
- Others now require more precise documentation.
- And the value of certain conditions has shifted.
CMS reviewed thousands of diagnosis codes and rebuilt how conditions group together. The result:
- Some codes were removed to improve accuracy and fairness
- Disease severity matters more
- Documentation must be clearer and more specific
- Average risk scores may decline even if patients haven’t changed
CMS made these adjustments using newer fee-for-service Medicare data to better align payments with actual costs. [3]
Why 2026 feels like a sudden change: Performance may look different
Despite the noise, the model didn’t start this year, it was phased in beginning in 2024, but CMS blended old and new models during the transition.
In 2026, the blend disappears. Plans are now seeing the full impact:
- No legacy model protection
- Full effect of removed codes
- Full recalibration of payment weights
Many organizations are interpreting this as performance decline when, in reality, it is the result of changes in the V28 methodology.
Research published in Health Affairs shows that updates to risk adjustment models often redistribute payments across plans even when patient health stays the same, meaning results can shift simply because measurement changed.[4]
Why more analytics may not be the answer for MA risk adjustment
One reaction we’ve seen repeatedly is a push to build bigger internal reporting ecosystems.
More dashboards, more enterprise data initiatives, and more cross-line-of-business analytics platforms meant to serve everyone at once. It’s a logical response, but it may not be the right one.
Most plans already know what has happened. They can see RAF movement and identify broad trends. Another reporting layer rarely changes the outcome.
Where organizations are struggling is the last mile. The messy, operational part:
- Which members still need outreach?
- Which providers have open actions?
- What work is done versus pending?
That’s not a reporting problem. It’s an execution problem. Execution improves when internal teams spend less time building infrastructure and more time reducing friction for providers and members. It is more effective to offload the operational complexity and stay focused on delivering care.
The question few plans ask out loud: Do we really have the data?
Most organizations believe they fully understand their members, but with data existing in separate streams, they might not see the full picture. For instance, Medicare populations may move between states, and data from rural providers, labs, behavioral health providers, and pharmacies can further fragment information. In fact, PBM fragmentation alone can obscure meaningful clinical signals. Even a small percentage of missing encounters creates blind spots and V28 magnifies those gaps because documentation precision now matters more.
A useful leadership question isn’t whether analytics are sophisticated. It’s whether the underlying data is complete.
Before you React: What’s really driving declining risk scores?
A common question we hear is: “Are our risk scores dropping because we’re underperforming?”
The honest answer is: it depends.
Risk scores can change because of:
- Model redesign (structural change)
- Documentation gaps (operational issue)
- Population differences (membership change)
- Normalization adjustments (policy change)
Without separating these factors, organizations may invest heavily in the wrong solution.
For example, increasing chart reviews won’t recover revenue tied to diagnoses CMS removed from eligibility.
The biggest shift in 2026: Timing matters more than volume
Historically, many plans relied heavily on retrospective chart reviews to capture missed diagnoses. However, under V28, this approach has become more nuanced. The new model places greater emphasis on clinically relevant, encounter based documentation and reduces the impact of certain diagnoses that were more easily captured through retrospective methods. As a result, retrospective review alone, particularly when volume driven, delivers lower incremental value.
In simple terms: Earlier, accurate documentation now matters more than end-of-year cleanup.
This aligns with broader CMS policy goals emphasizing accuracy and defensibility over coding intensity.[5]
How these changes are impacting providers
Clinicians are being asked for more documentation specificity, often without understanding why requirements changed.
When providers don’t see the connection between documentation and patient management, engagement drops.
Organizations adapting successfully are translating model changes into provider-level insights showing how documentation affects their own patient panels rather than delivering generic education.
Why risk adjustment changes under V28 are reshaping providers and quality teams
Historically, risk adjustment and quality teams ran parallel efforts with separate outreach lists and workflows. From a provider’s perspective, that can feel like multiple organizations asking for different things about the same patient.
Leading plans are aligning these efforts using shared activity tracking to create one view of what actions are pending for each member and provider panel. When providers see a single, coordinated request instead of competing workflows, engagement improves quickly. Administrative fatigue drops and work actually gets completed.
Why not every plan is affected the same way
Another important reality: there is no universal impact.
MedPAC has noted that risk adjustment updates redistribute payments depending on population characteristics and coding patterns.[6]
Impact varies based on:
- Dual-eligible populations
- Chronic disease mix
- Specialty utilization
- Historical coding practices
Two similar-sized plans can experience very different results.
What plans should focus on now to prepare for risk adjustment
Most organizations don’t need more activity: they need more clarity.
Three practical priorities stand out:
- Understand what changed vs. what broke.
Separate model impact from operational performance. - Reevaluate old strategies.
Past chart review success does not guarantee future ROI. - Treat risk adjustment as an operational function.
It now affects forecasting, care management, provider engagement, and compliance not just coding teams.
The bigger picture: What the V28 means for the future
The 2024 CMS-HCC model signals a broader shift in Medicare Advantage. Risk adjustment is moving away from retrospective revenue recovery and toward accurate, ongoing representation of patient health.
Organizations that adapt operationally will stabilize faster. Those relying on legacy approaches may experience continued margin pressure.
The opportunity still exists but it now depends on precision rather than scale.
How MedInsight can help organizations adapt to V28
At Milliman MedInsight® (MedInsight), we work with plans and risk-bearing providers to answer the questions many teams are asking right now:
- How much of our RAF change is model-driven?
- Where are real documentation opportunities?
- Which providers or specialties are most affected?
- How should strategy change under V28?
To address these questions, the MedInsight Risk Adjustment Platform unifies claims, clinical, and CMS data in a single environment and enriches it with HCC insights and risk attributes for precise stratification. Advanced AI-driven recapture analytics convert that data into tailored reports and streamlined coding workflows to boost accuracy and efficiency.
As CMS-HCC V28 increases complexity, embedded AI within the MedInsight Risk Adjustment Platform helps organizations improve accuracy, move faster, and address documentation gaps that affect risk adjustment performance by:
- Identifying high‑confidence diagnostic suspects with clinical evidence: AI analyzes claims, encounters, and clinical notes to surface likely HCCs supported by defensible documentation, helping teams focus on conditions that truly impact risk scores.
- Accelerating HCC opportunity identification across medical records: Automated record processing reduces manual review effort and surfaces potential HCCs faster than traditional workflows, enabling teams to act earlier in the measurement year.
- Applying NLP aligned to full CMS coding guidelines, not just conditions that are Monitored, Evaluated, Addressed, and Treated (MEAT): Natural language processing (NLP) evaluates documentation against complete CMS requirements, improving coding defensibility and reducing reliance on shorthand criteria that increase audit risk.
- Streamlining chart review from minutes to seconds: AI‑assisted chart summaries allow coders to validate findings quickly, replacing time‑intensive longitudinal reviews with fast, targeted confirmation that preserves accuracy and throughput.
By combining actuarial insight with integrated analytics, we help organizations move from uncertainty to a clear action plan.
A simple next step: What MA Plans can do now
If your organization is unsure whether recent performance changes reflect model redesign or operational gaps, that uncertainty itself is a signal.
A focused V28 impact assessment can quickly clarify exposure, priorities, and next steps.
The rules have changed. Understanding them clearly is now a competitive advantage. Connecting with a MedInsight expert is a practical place to start.
What resources are available on risk adjustment?
Schedule a demo: Connect with one of our healthcare analytics experts to schedule a personalized demo and see the platform in action.
Visit our website: Learn more about our Risk Adjustment Suite and Risk Adjustment Platform, a comprehensive, end-to-end risk adjustment solution.
Visit us at Becker’s: Join us at the Becker’s Spring 2026 Payer Issuers Roundtable held from April 13-14, 2026, in Chicago and connect with our subject matter experts to discuss your priorities and challenges.
