CMS has issued its CY2027 proposed rule for Medicare Advantage. It includes a "request for information" about future strategic directions for the program.
###
CMS issues four major Medicare rules each year. In the spring, we have the Inpatient Rule, which finalizes in August, ahead of the October fiscal year. In the summer, we have the Physician and the Hospital Outpatient rules, which publish November 1, ahead of the new calendar year.
And around November, we get the Medicare Advantage proposals, which finalize in the spring, and of the next MA contract year.
Find the MA press release here:
https://www.cms.gov/newsroom/press-releases/cms-proposes-new-policies-strengthen-quality-access-competition-medicare-advantage-part-d
Find the fact sheet here:
https://www.cms.gov/newsroom/fact-sheets/contract-year-2027-medicare-advantage-part-d-proposed-rule
Find the actual proposed rule here (paginated publication on 11/28):
https://www.federalregister.gov/public-inspection/2025-21456/medicare-program-contract-year-2027-policy-and-technical-changes-to-the-medicare-advantage-program
###
The word "coverage" occurs 726 times, but I don't see the words LCD or NCD this year. Prior Authorization 17 times, denial 3 times. Artificial intelligence, twice.
The request for information on "Future Directions in Medicare Advantage" starts on inspection copy page 6-11.
Comment to January 26, 2026.
###
AI CORNER
###
CMS’s RFI on Future Directions in Medicare Advantage (MA) is framed as a broad re-think of the program’s payment and benefit architecture, with CMMI explicitly positioned as the vehicle for testing ideas that would require statutory waivers. It mixes some narrow technical questions with several genuinely foundational ones about risk adjustment, quality bonuses, benefits, and data infrastructure.
1. Overall purpose and levers for change
CMS emphasizes that MA now covers over half of Medicare beneficiaries and that the current architecture (payment, risk adjustment, quality bonuses, and benefit design) may need modernization to support competition, equity, and value, while limiting gaming and inappropriate spending.
They highlight two distinct channels for change:
-
Conventional rulemaking – e.g., annual rate announcements, refinements to the HCC model, and regulatory changes that are “national in scale.”
-
CMMI models under §1115A – explicitly called out as the vehicle for testing more ambitious payment and benefit designs, especially those that need statutory flexibilities or waivers (for example, departures from current bidding rules or benefit rules that cannot be done in ordinary rulemaking). The Value-Based Insurance Design (VBID) model is cited as the MA-specific precedent.
There’s a clear tension in the text: CMS wants to address long-standing critiques (coding intensity, opaque quality bonuses, low-value supplemental benefits) but signals that many of the more disruptive ideas will likely be piloted first via CMMI rather than imposed system-wide.
2. Modernizing risk adjustment
Risk adjustment is treated as a central lever for MA’s future, not just a technical side issue. CMS underscores that:
-
RA is integral to plan payment, competition, and incentives for enrollee selection, care management, and coding behavior.
-
Current diagnosis-based HCC models, layered on demographic factors, have produced coding-related overpayments, differential impacts across plan types, and heavy administrative overhead.
The RFI requests comment on several structural questions, many of which could lend themselves to CMMI testing:
-
Re-balancing goals: How to simultaneously promote competition, maintain a level playing field (especially for smaller or less-resourced MA plans), reduce administrative burden, and ensure accurate payments—especially for high-need and underserved populations.
-
Alternative model designs:
-
Whether to move toward models that rely less heavily on diagnosis coding and more on other predictors of risk (utilization, pharmacy, functional status, social risk, etc.).
-
Whether to exclude or de-emphasize diagnoses from certain sources (e.g., “chart reviews,” home assessments) or require corroboration via follow-up encounters.
-
Potential “next-generation” models, including those that could use machine learning/AI as the prediction engine instead of standard linear models, with a request for input on safeguards, transparency, and fairness if AI is used.
-
-
Temporal rules for diagnoses: Over what time windows diagnoses should count; how to treat persistent but stable conditions; and how to treat conditions that have resolved but remain in the record.
-
Guardrails against gaming and upcoding: CMS openly invites comment on mechanisms to dampen coding incentives while preserving incentives for genuine care management and early detection.
Analytically, this section signals a willingness to consider fundamental re-engineering of RA, not just coefficient tweaks—especially if CMMI models can demonstrate alternative approaches that reduce coding games without destabilizing the program.
3. Re-designing quality bonus payments (QBP)
The QBP section is less about adding new Stars measures and more about questioning whether the entire bonus framework is fit for purpose.
Key strategic themes:
-
Lag and misalignment – CMS notes the multi-year lag between measurement periods and when QBP affects plan payment and bids, creating a disconnect between current quality performance and current financial incentives.
-
Concerns about gaming and opacity – Building on MedPAC and other critiques, CMS acknowledges that QBP may have become a blunt, expensive instrument with unclear marginal benefit for quality.
-
Role for CMMI – CMS explicitly contemplates a CMMI model to test alternative QBP structures that could:
-
Tighten the link between recent performance and payment.
-
Better target high-value improvements (e.g., outcomes, equity, patient experience).
-
Reduce opportunities for gaming via low-value process measures.
-
Although Star Ratings are referenced, the emphasis is squarely on the structure and timing of quality bonuses, not on the mechanics of individual measures.
4. Well-being and nutrition: reframing benefits
The RFI devotes a full section to “Well-Being and Nutrition,” which goes beyond narrow SDOH pilots and treats well-being as a core organizing concept for future MA benefit policy.
CMS:
-
Defines well-being broadly (emotional health, social connection, meaning/purpose, fulfillment) and explicitly links it to person-centered care, prevention, and long-term health outcomes.
-
Invites comment on tools, measures, and constructs that could rigorously capture well-being (including complementary/integrative health, self-management, and resilience).
-
Poses targeted questions around nutrition-related benefits, including medically tailored meals, healthy food stipends, and other nutrition supports. They ask how MA plans should design, target, and evaluate such benefits, especially in populations with chronic disease.
This is one of the places where a short phrase (“well-being and nutrition”) masks a foundational design question: whether MA benefit policy should formally prioritize long-horizon, preventive and holistic interventions, and how those benefits should be standardized, valued, and evaluated within the payment architecture.
5. Data, reporting, and benefit-use information
Outside Section VIII proper, CMS’s Supplemental RFI asks for comment on simplifying or reshaping data reporting while preserving oversight, with a list that includes:
-
Network adequacy reporting.
-
Medical loss ratio (MLR) reporting.
-
“Benefit, including supplemental benefit, usage and utilization data reporting.”
-
SNP model of care reporting.
Although it is only one bullet, the request for input on benefit and supplemental benefit utilization data has deep implications. CMS is effectively signaling interest in a more systematic, possibly standardized data layer on what benefits are offered, who uses them, and with what outcomes. That information would be critical to:
-
Evaluating the real-world value of supplemental and “flex” benefits (e.g., meals, transportation, social needs benefits).
-
Designing future RA and QBP reforms that differentially reward plans for high-value benefit structures rather than for simply offering long menus of rarely used benefits.
-
Supporting CMMI experiments that tie payment to demonstrated benefit uptake and impact.
Here the tension is explicit: CMS wants richer, more actionable data on benefits but also asks for ideas to simplify and align reporting mechanisms, including automation and alignment with existing data flows. They also invite comment on which data elements are genuinely burdensome versus essential, signaling openness to pruning low-value reporting while building up high-value benefit-use data.
6. Marketing oversight, TPMOs, and AI decision tools
The RFI also includes a cluster of questions about marketing, agents/brokers, and third-party marketing organizations (TPMOs), again ranging from small technical edits to broader structural issues.
CMS seeks input on:
-
Refining the TPMO definition and segmentation (size, scope, role) to better target oversight.
-
Adjusting translation thresholds, testimonial rules, use of the Medicare card image, and outbound enrollment verification requirements.
-
How to hold “bad actors” accountable while not overly burdening compliant entities.
-
How to use data and technology—including AI—to monitor markets and power decision-support tools for beneficiaries and caregivers.
This is another area where a handful of technical RFIs hint at a larger shift: moving toward data-driven, technology-enabled oversight and decision support, with AI-enabled tools both as an opportunity (better plan choice, fraud detection) and a risk (bias, opaque steering) that CMS expects to regulate more actively.
7. Cross-cutting policy tensions
Across these sections, several strategic tensions recur:
-
Innovation vs. statutory constraint: CMS repeatedly points to CMMI and its waiver authority as the way to test models that cannot be implemented through ordinary rulemaking, acknowledging limits of the current statute.
-
Competition vs. equity/level playing field: CMS wants a more competitive MA market but is sensitive to the ways RA, QBP, and marketing practices can disadvantage smaller or safety-net-oriented plans.
-
Program integrity vs. administrative burden: From RA documentation rules to benefit and network reporting, CMS is asking whether it can rebalance what data are collected and how, to focus on high-value oversight while shedding low-yield requirements.
-
Short-term vs. long-term value: The focus on well-being, nutrition, and supplemental benefit data indicates an interest in shifting the MA paradigm toward long-horizon health investments, but CMS is clearly unsure how to operationalize this within annual bidding and payment cycles.
In short, while the RFI contains its share of small-bore questions, it is best read as an open invitation to debate the next generation architecture of MA—how the program pays plans (risk adjustment, bonuses), what it expects them to offer (benefit and well-being design), and what data and models (including CMMI pilots with waiver authority) will be needed to steer MA toward higher value and more equitable outcomes.
