The CMS-HCC V28 transition is the most significant change to Medicare risk adjustment in a decade. As of Payment Year 2026, non-PACE Medicare Advantage plans and MSSP ACOs operate under a 100% V28 model — the old V24 model is fully retired.
If your organization is still using manual RAF calculations or a vendor that hasn't fully updated its blend schedule logic, you're at serious risk of miscalculating member revenue and misaligning care management priorities.
What Changed from V24 to V28
The 2024 CMS-HCC model (V28) made several structural changes compared to the 2020 model (V24):
- •86 new HCC categories added, 67 HCC categories removed or consolidated
- •Coefficient recalibration across all remaining HCCs — most decreased, reflecting more accurate cost prediction
- •Demographic risk scores updated with new age-sex interaction factors
- •Disease hierarchies revised — 190 hierarchy rules govern which HCCs are suppressed when a more severe condition is present
The net effect for most portfolios: a 5-15% decrease in raw RAF scores at the HCC level before normalization, partially offset by the updated normalization factor.
The Blend Schedule You Must Implement Correctly
CMS did not require an overnight switch. Instead, they published a phased transition:
| Payment Year | V28 Weight | V24 Weight | |---|---|---| | PY2024 | 33% | 67% | | PY2025 | 67% | 33% | | PY2026 | 100% | 0% |
PACE organizations are on a different schedule: PY2026 uses 90% V28 / 10% V24, per the 2026 CMS Final Rate Announcement (April 2025).
This means your scoring engine must:
- •Calculate the patient's RAF under each model separately
- •Apply the correct blend weights for the payment year
- •Use the per-segment normalization factor appropriate to each model (CNA for community non-dually eligible, etc.)
Getting this wrong by even 0.5% across a 10,000-member population equals $500,000+ in revenue miscalculation.
Normalization: The Most Misunderstood Variable
Normalization converts a patient's raw RAF score into the payment-relevant score. Two normalization concepts cause the most confusion:
Per-segment CNA (Coding Normalization Adjustment): V28 uses 1.015 for the CNA used in per-segment score calculation. This is NOT the same as the aggregate model normalization factor.
Aggregate model normalization: This is the factor published in the CMS Final Rate Announcement each April. For PY2025 it was 1.045; for PY2026 it is 1.067 (per the April 7, 2025 announcement).
Many legacy systems hardcode these values and fail to update them when CMS publishes final rates. The correct pattern is to store them in a versioned YAML configuration, validate against regulatory bounds on ingest, and populate the scoring engine dynamically.
What RADV Auditors Look For
The Risk Adjustment Data Validation (RADV) audit process specifically evaluates whether your submitted HCC codes are:
- •Supported by a face-to-face encounter (telehealth requires POS 02 and modifier 95 or GT for V28 purposes)
- •Documented with MEAT criteria (Monitoring, Evaluating, Assessing or Addressing, and Treating)
- •Traceable to a source record with an immutable audit trail
V28 expanded the MEAT documentation requirements. Clinical documentation now needs to explicitly link the HCC to current treatment — not just mention the diagnosis historically.
A 97.3/100 RADV readiness score requires that every submitted HCC has:
- •A face-to-face encounter within the measurement year
- •MEAT documentation scored by an evidence engine
- •Bronze-layer provenance (content SHA-256 hash linking back to the source clinical record)
The Automation Imperative
Manual V28 compliance is not sustainable. The combination of:
- •HCC coefficient changes (86 new, 67 removed)
- •Disease hierarchy rule updates (190 rules)
- •Per-year blend schedule adjustments
- •Annual normalization factor updates
- •PACE vs. non-PACE differentiation
...creates a compliance surface area that requires programmatic enforcement, not spreadsheets.
The organizations winning on risk adjustment in 2026 are those who automated the blend calculation, moved to real-time streaming RAF updates (so care management can act within seconds of a new diagnosis), and built audit trails that survive RADV scrutiny without scrambling.
AssureLogix's CMS-HCC V28 engine handles V24/V28 blending, PACE overrides, normalization factor injection from annual YAML configs, and RADV audit trail generation automatically. Book a demo to see the scoring engine in action.