V28 Risk Model Goes Full-Weight in PY 2026, Compressing Medicare Advantage Risk Scores by $11 Billion

From modeling the V28 transition’s impact on plan-level risk scores across three payment years, the pattern is consistent: plans with heavy reliance on telephonic health risk assessments face the steepest revenue compression. Payment year 2026 is where the full weight lands. CMS now calculates Medicare Advantage risk scores using 100% of the 2024 CMS-HCC model (V28), ending the three-year blended phase-in that gave plans a partial V24 cushion. The V24 model is retired. The buffer is gone.

The trade press has largely covered this transition as a coding compliance challenge, focused on which ICD-10-CM codes survive and which do not. That framing misses the actuarial core of the problem. V28 at full weight is a revenue event, not a coding event. It compresses risk scores, reduces plan payments, and forces actuaries to recalibrate bid assumptions, benefit designs, and provider contracts simultaneously. CMS projects a 3.12% average risk score reduction across the MA population, translating to approximately $11 billion in net savings to the Medicare Trust Fund. For plans already operating on thin margins after years of benchmark compression, that 3.12% is the difference between a sustainable benefit package and a market exit.

As we documented in our analysis of the 2.9 million forced disenrollments in 2026, carrier exits surged tenfold from a 1% historical average to 10% of the MA market. V28 at full weight is one of the primary drivers of that exodus.

The V28 Phase-In Timeline: Three Steps to Full Weight

CMS structured the V28 transition across three payment years to give plans time to adjust coding workflows and financial models. Each increment removed more of the V24 cushion, and each hit harder than the one before it.

Payment Year	V28 Weight	V24 Weight	Incremental Revenue Impact
PY 2024	33%	67%	Moderate; V24 still dominant
PY 2025	67%	33%	Steep; V24 cushion drops to one-third
PY 2026	100%	0%	Full impact; no V24 offset

The blending mechanics matter because the V24 model assigned higher coefficients to many conditions than V28 does. During PY 2024, plans still received 67% of the V24 risk score for each enrollee, which masked much of V28’s compression effect. By PY 2025, the V24 weight dropped to 33%, and plan actuaries began seeing the steeper cliff in their encounter data runs. The financial impact was not linear across years; the jump from 67% to 100% V28 weight is the most punishing increment because the remaining one-third of V24 cushion disappears entirely.

CMS applies a normalization factor of 1.067 to the V28 model for PY 2026, which partially offsets the risk score compression but does not eliminate it. The statutory minimum MA coding pattern adjustment of 5.90% also remains in effect, reducing payments to account for the diagnostic coding intensity differences between MA plans and fee-for-service Medicare. Together, these adjustments define the effective payment rate that plan actuaries must build their 2027 bids against.

Milliman identified the V28 transition as one of six critical issues for MA plans in 2026, noting that full V28 implementation, combined with new RxHCC normalization factors that affect MA-PD populations independently, creates compounding financial pressure that plans cannot offset through coding optimization alone.

2,027 Fewer Valid Diagnostic Codes

The V28 model reduced the number of valid ICD-10-CM diagnostic codes that map to HCCs from 9,797 to 7,770, a net reduction of 2,027 codes. CMS removed approximately 2,294 codes from the mapping and added 268 new ones. The net effect is a 20.7% reduction in the diagnostic codes that generate risk-adjusted payments.

This is not a uniform trim. The codes removed disproportionately affect conditions where the distinction between complicated and uncomplicated presentations previously generated different HCC mappings. Diabetes is the most prominent example. Under V24, a provider documenting diabetes with peripheral vascular disease captured a combined risk adjustment factor (RAF) of approximately 0.590 (0.302 for the diabetes HCC plus 0.288 for the vascular HCC). Under V28, the same patient generates a RAF of approximately 0.166, because V28 constrains diabetic disorders so that uncomplicated and complicated presentations contribute similar risk scores.

The coding reduction forces plans to retrain providers on the revised code-to-HCC mappings. Providers accustomed to documenting certain complication codes that previously generated higher risk scores must now understand which codes still map to HCCs and which no longer do. For plans that relied on retrospective chart review to capture missed diagnoses, the narrower code set means fewer opportunities to recover risk score value through documentation improvement programs.

3Gen Consulting’s V28 coding impact analysis found that plans with mature prospective coding programs, where providers document conditions accurately at the point of care, experienced less disruption than plans dependent on retrospective review. The distinction matters because CMS has signaled, through both the V28 code reductions and the proposed 2027 chart review exclusion, that it intends to narrow the gap between MA and fee-for-service coding intensity over time.

HCC Expansion from 86 to 115: Where the Coefficients Moved

While V28 reduced the number of valid diagnostic codes, it simultaneously expanded the number of hierarchical condition categories from 86 to 115, a net increase of 29 HCCs. The expansion creates more granular clinical groupings, which in theory allows the model to differentiate more precisely between patients with varying levels of clinical complexity. In practice, the expansion redistributes coefficient weight in ways that benefit some plan populations and penalize others.

Several specific coefficient movements stand out for their actuarial significance:

Pressure ulcer categories (newly created): V28 introduced three new HCCs for pressure ulcers with RAF coefficients significantly higher than most existing categories. HCC 379 (pressure ulcer with necrosis through to muscle, tendon, or bone) carries a coefficient of 1.965. HCC 381 (full thickness skin loss) carries 1.075. HCC 382 (partial thickness skin loss) carries 0.838. Plans with skilled nursing facility populations or complex wound care programs stand to capture substantial risk score value from these new categories.

Diabetes coefficient restructuring: The constraining effect on diabetes is V28’s most widely discussed change. Type 2 diabetes (uncomplicated) saw its coefficient increase from 0.105 under V24 to 0.166 under V28, a 58% gain. But the combined coefficients for diabetes with complications dropped sharply because V28 eliminates the additive effect of complication-specific HCCs. The diabetes-plus-CHF interaction, for example, dropped from a 0.121 RAF add-on to 0.112. For plans with large diabetic populations, the net effect depends entirely on the prevalence of documented complications within their enrollee base.

Substance use disorder categories: V28 added new HCCs for substance use disorders, creating coded pathways that did not exist under V24. Plans investing in behavioral health integration may capture risk score value from these new categories, though the documentation requirements are still being clarified through CMS guidance.

Eliminated interactions: V28 removed the interaction between immune disorders (HCC 47) and cancer (HCCs 8-12). Plans with oncology-focused populations that previously benefited from this interaction will see a direct risk score reduction.

Quantifying the 3.12% Average Risk Score Reduction

CMS projected a 3.12% average reduction in Medicare Advantage risk scores when V28 was first applied at 33% weight in PY 2024. That projection, documented in the CY 2024 Advance Notice, translates to approximately $11 billion in net savings to the Medicare Trust Fund. At full weight in PY 2026, the per-plan impact varies substantially depending on population acuity mix, coding practices, and reliance on supplemental data sources.

The 3.12% figure is an average across all MA contracts. Plans with sicker populations, particularly those with high concentrations of diabetes, cardiovascular disease, and cancer, face steeper compression because V28 reduced the coefficient spread between complicated and uncomplicated presentations of these conditions. Plans with relatively healthier populations may see minimal impact or, in some cases, slight risk score increases from the new HCC categories.

The revenue impact per member per month (PMPM) depends on the plan’s county-level benchmarks and star rating quality bonus. For a plan operating in a county with a $1,000 benchmark and a 5% quality bonus, a 3.12% risk score reduction translates to roughly $33 PMPM in lost revenue. At scale, a mid-size plan with 100,000 members faces an annualized revenue reduction of approximately $39.6 million from risk score compression alone, before accounting for normalization adjustments or coding pattern corrections.

Milliman’s State of the 2026 MA Industry report quantified the broader market effect: total value added across the MA program declined over 7% from 2025 to 2026, the largest single-year decline in the program’s history. Part C benefit value decreased approximately $17 PMPM year over year. Total member premiums (Part C plus Part D) increased $1.64 PMPM before the Annual Election Period, marking the first time in at least a decade that aggregate premiums rose.

As our analysis of the KFF 2026 MA Spotlight data documented, the benefit design trade-offs are already visible. Average monthly plan premiums declined from $16.40 to $14.00, but supplemental benefits eroded across OTC allowances (down from 73% to 66% of plans), meal benefits (down from 65% to 57%), and transportation benefits (down from 30% to 24%). Plans are cutting supplemental benefits to preserve the zero-premium positioning that MA enrollees have come to expect.

Health Risk Assessments and Telephonic Coding Under Pressure

V28’s code reduction hits hardest for plans that built their risk capture strategy around telephonic health risk assessments (HRAs) and retrospective chart reviews. These supplemental data sources accounted for approximately $7.5 billion in MA payments in 2023, according to Georgetown University’s Center on Health Insurance Reforms. CMS data shows that 62% of MA enrollees have at least one chart review record, and chart reviews increased payments for roughly one in six MA enrollees.

The V28 model was explicitly calibrated to reduce the value that plans can extract from supplemental coding activity. By eliminating 2,027 diagnostic codes from the HCC mapping, CMS narrowed the universe of conditions that chart reviewers and HRA vendors can capture. The codes most frequently identified through retrospective review, particularly the complication-specific diabetes and cardiovascular codes, are precisely the codes that V28 either eliminated or constrained.

MedPAC has been tracking MA overpayments driven by coding intensity for years. Its March 2026 report estimated total MA overpayments at $27 billion in 2023, with $24 billion attributed to chart review and coding intensity practices. MA payments run approximately 22% higher than original Medicare for equivalent enrollees, a gap that V28 was designed to narrow.

Plans that invested in prospective coding improvement, training providers to document accurately during patient encounters rather than recovering missed codes afterward, are better positioned under V28. Plans dependent on supplemental HRA vendors and retrospective chart review firms face a structural revenue decline that no amount of vendor optimization can fully recover, because the codes themselves no longer exist in the model.

How Plan Actuaries Are Adjusting 2027 Bids

The June 2026 bid submission deadline forces MA plan actuaries to build their CY 2027 bids on the first quarter of actual V28 full-weight encounter data. This is a narrow experience base for a model transition of this magnitude. The key adjustments fall into three categories: revenue forecasting, benefit design, and provider contracting.

Revenue forecasting: Plan actuaries must project 2027 risk scores using 2026 encounter data that reflects the full V28 code set for the first time. The challenge is separating genuine acuity changes in the population from the mechanical effect of the code-to-HCC mapping changes. A plan that sees a 4% risk score decline in Q1 2026 needs to determine how much of that decline reflects the model transition (which will persist) versus temporary provider coding adjustment (which should improve as providers adapt to the new mapping).

CMS finalized a 2.48% rate increase for CY 2027 in its April 6 Rate Announcement, a significant reversal from the 0.09% increase proposed in the January Advance Notice. As we analyzed in our coverage of the CMS risk model deferral and its bid pricing impact, the 239 basis point swing from advance notice to final rate was the most significant revision in recent memory and reflects Congressional pressure to cushion the V28 transition. Combined with expected risk score growth of approximately 2.45%, the effective growth rate reaches roughly 4.98%, though that still trails prevailing medical cost trends in most markets.

Benefit design: Plans are reducing supplemental benefits to offset revenue compression. Milliman reported 231 fewer $0 MA-PD plans in 2026 versus 2025, a 9.5% decrease. Medical maximum out-of-pocket (MOOP) limits increased from approximately $5,100 to $5,440 (7% increase). Medical deductibles jumped from $33 to $57, a 72% increase. Part D deductibles rose from approximately $230 to $375, a 60% increase, with deductible prevalence climbing to 83% of members.

The Better Medicare Alliance reported national median MOOP increased from $5,400 to $5,900, a 9.3% increase and $900 more than 2024. Plans offering Social Determinants of Health (SSBCI) benefits fell from 16% to 12%. Fitness benefits declined from 96% to 93%.

Provider contracting: Plans are renegotiating provider contracts to share the documentation burden and risk score impact. Value-based arrangements that tie provider compensation to risk score accuracy are becoming more common, though they introduce complexity around which party bears the risk of V28’s mechanical code reductions versus genuine documentation gaps.

CMS 2027 Proposals That Compound V28’s Revenue Pressure

V28 at full weight is not the only revenue headwind facing MA plans. CMS proposed two additional changes in its January 26, 2026 Advance Notice for CY 2027 that would compound the compression if finalized.

Chart review exclusion ($7.12 billion impact): CMS proposed that MA organizations can no longer submit diagnoses from “unlinked” chart reviews for risk score adjustments. Unlinked reviews are those conducted outside of an actual clinical encounter and not associated with a specific service or care event. CMS estimates this exclusion would reduce MA payments by $7.12 billion, equivalent to a 1.53% decrease in MA payments relative to 2026. As we detailed in our analysis of the CMS chart review ban, 58% of MA contracts submitted unlinked chart review records in 2022, making the revenue exposure widespread across the industry.

Audio-only telehealth exclusion: Diagnoses from audio-only encounters, identified by modifiers 93 and FQ, would no longer count toward risk score calculations. CMS excluded these encounters from the updated V28 model calibration. Plans that expanded telephonic assessment programs during and after the pandemic face direct risk score reductions from this change.

The combined impact of V28 at full weight, the chart review exclusion, and the audio-only telehealth restriction creates a layered revenue compression that is structurally different from a single rate cut. Each layer targets a different risk score capture mechanism. V28 narrows the code set. The chart review exclusion restricts the data source. The audio-only exclusion restricts the encounter modality. Plans that relied on all three supplemental coding channels face compounding losses.

The CY 2027 Advance Notice estimated the combined risk model revision and normalization impact at negative $15.22 billion. When CMS finalized the rates at 2.48% growth instead of the proposed 0.09%, it softened but did not eliminate the blow. The April 6 final rule retained the chart review exclusion and audio-only restrictions while providing a larger benchmark increase than initially proposed.

Carrier-Level Responses and Market Restructuring

The V28 transition is driving measurable strategic shifts across the largest MA carriers.

UnitedHealth Group now projects 1.3 to 1.4 million member exits from MA in 2026, revised upward from an initial projection of 600,000. UnitedHealth described its Q1 results as “unusual and unacceptable” and is implementing benefit reductions, particularly in PPO plans, while prioritizing profitability over membership growth.

CVS Health (Aetna) dropped 90 MA plans for 2026 and is operating prescription drug plans in 100 fewer U.S. counties than 2025. Its 2025-2026 margin recovery strategy focuses on benefit cuts and portfolio optimization rather than enrollment expansion.

Humana reduced county coverage to 85% (down from 89% in 2025) but improved its Star Ratings positioning, with 4.5-star plans increasing from 3% in 2025 to 14% in 2026. Humana saw a 22% enrollment increase in areas with positive relativity shifts, while carriers with negative relativity shifts (UnitedHealth, CVS, Elevance, Centene) experienced 4% to 33% enrollment declines.

Overall MA enrollment is projected to drop from approximately 35 million to 34 million in 2026. If confirmed, this would be the first year-over-year enrollment decline in roughly two decades, a remarkable reversal for a program that grew continuously through multiple payment reform cycles. As our reporting on the D-SNP enrollment tripling documented, carriers are pivoting from standard MA toward special needs plans that carry higher risk scores, better margins of 7.5% versus 3.6% for standard MA (MedPAC), and more favorable regulatory treatment.

The SNP Pivot and Its Risk Adjustment Implications

The shift toward Special Needs Plans deserves separate actuarial attention because it intersects directly with V28’s coefficient redistribution. C-SNPs (Chronic Condition SNPs) saw enrollment surge 49% in the most recent period, with plan count tripling since 2020. D-SNPs (Dual-Eligible SNPs) tripled from 2.2 million to 6.0 million members since 2018. SNPs now represent 23% of total MA enrollment.

Under V28, SNP populations may actually benefit from some of the new HCC categories. The expanded pressure ulcer HCCs (with coefficients up to 1.965) and new substance use disorder categories align with clinical profiles common in SNP populations. Plans that restructure their product portfolios toward SNPs can partially offset V28’s compression on standard MA products by capturing higher risk scores in populations where the new model categories create value.

The risk for CMS is that the SNP pivot becomes a risk score optimization strategy rather than a care model innovation. If carriers are shifting membership into SNP vehicles primarily to capture higher risk scores under V28, the program integrity concerns that motivated V28 in the first place may simply migrate from one product category to another.

Why This Matters for Plan Actuaries

V28 at full weight represents the most significant single-year change to MA plan economics since CMS introduced the hierarchical condition category model. Patterns we have tracked across three payment years of the transition point to several actuarial implications that extend beyond the immediate bid cycle.

First, the experience data from PY 2026 will set the baseline for all future risk score projections. Plan actuaries who set trend assumptions during the blended V24/V28 period need to rebase their models on clean V28-only experience. Trend rates calibrated to the blended period will overstate expected risk scores going forward.

Second, the interaction between V28 and CMS’s proposed chart review and audio-only exclusions creates a compounding effect that is difficult to model without plan-specific coding source data. Actuaries need to decompose their risk score into its component data sources (encounters, chart reviews, HRAs, telehealth) and model the revenue impact of each proposed restriction independently before aggregating.

Third, the RADV audit extrapolation methodology, though currently vacated by courts, remains a background risk. CMS has appealed the vacatur, and if extrapolation is reinstated, plans that responded to V28 by intensifying their remaining coding channels could face retroactive audit exposure. The actuarial reserve for RADV contingent liability requires reassessment in light of both the court decision and CMS’s stated intent to appeal.

Fourth, the shift to 65% clinical weight in CMS Star Ratings for 2027 compounds the V28 pressure by tying quality bonus payments more tightly to clinical outcomes. Plans that lose risk score revenue through V28 and simultaneously lose Star Ratings performance from the measure restructuring face a double compression that cannot be solved through coding optimization.

The plans that will navigate V28 most effectively are those that treated the three-year phase-in as a transition period rather than a reprieve, investing in prospective provider documentation, encounter-based coding accuracy, and product portfolio restructuring toward populations where V28’s new categories create genuine actuarial value.