Special needs plans (SNPs) drove 85 percent of the net increase in Medicare Advantage (MA) enrollment over the past year, according to KFF, while MA enrollment overall grew 3 percent.
Nearly 8.2 million people are enrolled in SNPs in 2026, or 23 percent of all MA enrollees, up from 21 percent in 2025. Dual eligible special needs plans (D-SNPs) hold 78 percent of SNP enrollment, roughly 6.4 million people. Growth of this size inside a single product type resets where risk adjustment teams spend their attention.
The growth story is familiar. The operational story gets less airtime. A dual eligible member is risk-adjusted under Medicare rules, and the same member sits inside a state Medicaid program with its own risk adjustment model, its own submission calendar, and its own documentation expectations. Two payers, two methodologies, one chart.
For organizations built around Medicare, the reasonable first question is why any of this belongs on a Medicare team’s agenda. The answer starts inside the Medicare payment itself.
Medicaid status is already inside your Medicare payment
The risk model reads it. The CMS-HCC non-ESRD model scores community-dwelling beneficiaries in separate segments by dual status: full benefit dual, partial benefit dual, and non-dual, with institutional residents scored separately again. Medicaid eligibility is a factor in the score. The same member, with the same conditions and the same documentation, generates a different Medicare payment depending on dual status. Determining dual status is a state Medicaid function, not a Medicare one.
State data quality is therefore a Medicare revenue issue. Dual status shifts month to month as members gain and lose Medicaid eligibility, and state eligibility files drive the determination. A lag or an error in the state file moves the Medicare risk score for the affected member. Plans usually discover the problem at reconciliation, well after the payment cycle it affected.
Star Ratings picked up the same dependency. Beginning with the 2027 Star Ratings, the Excellent Health Outcomes for All (EHO4all) reward, the renamed Health Equity Index, measures plan performance for enrollees who are dually eligible, receiving the low-income subsidy, or disabled. Bonus revenue now turns in part on outcomes for a population defined by Medicaid and low-income status.
Enrollment rules are converging too. Under the CY2025 MA final rule, D-SNPs affiliated with a Medicaid managed care organization must operate with exclusively aligned enrollment by 2027, covered in more detail below.
Three separate Medicare revenue mechanisms, risk scores, Star Ratings bonus, and enrollment eligibility, now depend on Medicaid data a Medicare team does not generate and often does not see. The question is no longer whether Medicaid belongs on the agenda. It is how much of the Medicaid side the organization actually controls.
Here is where the two payment systems diverge, why multi-state operation multiplies the exposure, and what regulators are signaling on both sides.
Two models, two sets of rules
The models differ. MA runs on the CMS Hierarchical Condition Categories (CMS-HCC) model. Most state Medicaid programs run on the Chronic Illness and Disability Payment System (CDPS), developed at the University of California San Diego. Thirty-two state Medicaid agencies, including the District of Columbia and Puerto Rico, use CDPS, according to the Institute for Medicaid Innovation. Other states use Clinical Risk Groups, DxCG, or the Johns Hopkins ACG System.
The inputs differ. CDPS+Rx, the variant used in states such as California, scores members from ICD diagnosis codes and National Drug Code pharmacy data together. CMS-HCC scores from diagnoses. A pharmacy fill with no matching diagnosis carries weight in one system and none in the other.
The math differs. CDPS weights combine additively across major body-system categories. CMS-HCC applies hierarchies and interaction terms across condition groups. Identical documentation produces different revenue behavior in each program.
The calibration population differs. CMS-HCC reflects treatment patterns among Medicare fee-for-service beneficiaries. CDPS+Rx version 7.0 was rebuilt on Medicaid managed care data from 2017 through 2019, correcting a long-standing gap between the population used to build the model and the population it pays for. Teams reasoning about Medicaid capture from Medicare intuition start from the wrong baseline.
State variation is the multiplier
No national Medicaid risk adjustment standard exists. Each state sets its own model, its own model version, its own submission windows, and its own reconciliation approach through the managed care contract.
Version drift. States adopt CDPS updates on their own schedules. A plan operating in several states runs several model versions at once, with different weights attached to the same condition.
Timing. Medicaid submission deadlines and lookback periods follow state contract terms rather than a national calendar. A risk adjustment operation built around the Medicare sweeps calendar will miss Medicaid deadlines set by a different clock.
Reconciliation. Some states true up capitation against submitted encounter data on a defined cycle. Others do not. The financial consequence of an incomplete Medicaid encounter file therefore varies by market.
For a single-state D-SNP, this is a second rulebook. For a plan expanding a D-SNP footprint into new states, each new market adds another.
Regulatory pressure is arriving from both directions
On the Medicare side, alignment is becoming mandatory. As noted above, the CY2025 MA final rule requires D-SNPs affiliated with a Medicaid managed care organization to operate with exclusively aligned enrollment by 2027. Members whose Medicare and Medicaid coverage sit with different organizations will need to be moved or exited. Enrollment, care management, and data flows all shift as a result, and risk adjustment operations inherit the consequences.
On the Medicaid side, oversight is tightening on its own track. CMS issued an informational bulletin on March 11, 2026 directing states to strengthen managed care monitoring. Three items land squarely on plan operations:
- All Medicaid managed care contract and rate submissions moved to the CMS MC-Review system as of July 1, 2026
- States must ensure plans report encounter data to T-MSIS including allowed and paid amounts by September 2026, with new fields for state-directed payment detail
- States are expected to monitor prior authorization denial rates across plans and investigate outliers
Encounter data completeness has always mattered for Medicaid rate development. It now sits inside a federal reporting structure with named deadlines.
Integration on paper is not integration in practice. Milliman’s review of the 2026 D-SNP market found most integration requirements continue to be met through coordination rather than genuine integration. Organizations meeting the letter of the requirement through coordination agreements still carry the operational seams, and those seams are where documentation and data completeness problems live.
Where the seams show up
Three patterns recur in growing D-SNP operations, along with the practices strong programs use to close them.
Ownership sits between teams. Medicare risk adjustment reports to one leader, Medicaid encounter operations to another, and no single owner holds the dual eligible member end to end. Programs handling this well name one accountable owner for the dual population and give the role visibility into both data streams.
Provider documentation is built for one payer. Network providers document to Medicare expectations because Medicare is the volume. Conditions carrying Medicaid weight and no Medicare weight go uncaptured, and the Medicaid rate cell absorbs the difference. Provider education built around the dual population, rather than around Medicare alone, closes most of the gap.
Expansion moves faster than infrastructure. A new state market arrives with a new model version, a new submission calendar, and a new set of contract terms. The programs keeping pace treat each new state as an operational build, with the model version and submission calendar loaded before the first member enrolls.
Questions to consider
- How quickly does a change in a member’s state Medicaid eligibility reach the team responsible for your Medicare risk scores?
- Who in your organization owns the dual eligible member across both Medicare and Medicaid risk adjustment, and is the answer a person or a committee?
- Do you know which CDPS version, or which alternative model, applies in each state where you operate a D-SNP, and are the weights loaded into your analytics?
- Would your September 2026 Medicaid encounter submissions withstand the same scrutiny you apply to a Medicare sweep?
- If exclusively aligned enrollment takes effect in 2027 across your affiliated markets, which members move, and who has modeled the risk score impact?
RISE will examine dual eligible risk adjustment across Medicare and Medicaid, along with RADV readiness, state Medicaid variation, and SNP compliance exposure, at The 27th Risk Adjustment Forum, October 27-29, in Orlando.