From mapping the county-level network coverage reduction across the 10 largest MA issuers since January 2026, the adequacy gap is concentrated in mid-tier metros where academic medical centers have the most leverage. When Mayo Clinic exits UnitedHealthcare and Humana networks across Minnesota, Wisconsin, and Iowa, the ripple effect extends well beyond those three states. Patients with established oncology or cardiology relationships face a choice between switching plans and switching providers, and the actuarial consequences of that forced choice cascade through risk scores, utilization assumptions, and premium bids for years.

The 2026 wave of health system departures from Medicare Advantage networks is not a single-year anomaly. It represents the structural consequence of years of prior authorization friction, reimbursement disputes, and the widening gap between what CMS pays MA plans and what those plans pay providers. Becker's Hospital Review has tracked at least 21 health systems that have terminated or allowed MA contracts to lapse in 2026. Simultaneously, a JAMA study from Johns Hopkins found that 2.9 million Medicare Advantage enrollees, roughly 10% of the studied population, faced forced disenrollment for 2026 coverage. That forced disenrollment rate had averaged just 1% from 2018 through 2024.

This article examines how provider network exits reshape the actuarial framework for MA plan pricing, where CMS network adequacy standards fall short of capturing real-time access disruption, and what plan actuaries should model differently in 2027 bid submissions.

21+
Health Systems Exiting MA Networks in 2026 (Becker's)
2.9M
Enrollees Forced to Switch Plans for 2026 (JAMA)
10%
2026 Forced Disenrollment Rate, Up From 1% Average

The 21 Health System Exits: Who Left, and Why

The Becker's Hospital Review tracker, which continues to be updated, captures at least 21 health systems that have severed or declined to renew Medicare Advantage contracts in 2026. The list includes some of the most recognized names in academic medicine and specialty care, and the insurer targets span every major MA carrier.

Health System Insurer(s) Dropped Effective Date Scope
Mayo Clinic UnitedHealthcare, Humana January 1, 2026 MN, WI, IA locations; individual and DSNP plans
Mount Sinai Health System Anthem (Elevance) January 1, 2026 9,000+ physicians; ~200K affected members
UNC Health Humana, WellCare (Centene), HCSC January 2026 Multiple MA plans across North Carolina
Providence Clinical Network UnitedHealthcare January 1, 2026 15 hospitals across California markets
Lehigh Valley Health Network UnitedHealthcare January 25, 2026 Individual, I-SNP, D-SNP, and Group Retiree plans
Moffitt Cancer Center Aetna (CVS), Humana Dec 2025 / July 2026 Cancer center; specialty access in Tampa market

The stated reasons are remarkably consistent across systems. Prior authorization friction tops the list: chronic denials of medically necessary care, delayed payments, and administrative practices that create barriers for both patients and physicians. Mount Sinai's public statements during its dispute with Anthem cited more than $450 million in unpaid claims for previously delivered care. Humana, defending its position during the Mayo Clinic exit, pointed to provider reimbursement demands "significantly higher" than Original Medicare rates, arguing that elevated provider costs ultimately flow through to beneficiaries in the form of higher premiums.

Providence's CEO called for a Medicare Advantage "reset," describing the current reimbursement framework as unsustainable for health systems delivering complex care to aging populations. The common thread is structural: MA plan payment rates, constrained by CMS benchmarks and V28 risk score compression, are colliding with provider cost inflation that hospitals can no longer absorb. When the gap becomes unworkable, the provider exits the network rather than continuing to deliver care at a loss.

One partial resolution stands out. Mount Sinai and Anthem reached a new agreement on April 13, 2026, restoring in-network access after three and a half months of disruption. That temporary exit still affected approximately 200,000 Anthem members who needed to navigate out-of-network care during the first quarter, a period that included flu season and ongoing chronic disease management. The actuarial footprint of even a temporary network disruption persists well beyond the resolution date, as deferred care, interrupted treatment protocols, and emergency utilization during the gap period all flow into subsequent claims experience.

The Scale of Disruption: 2.9 Million Forced to Switch

A JAMA study published by Johns Hopkins researchers quantified the scope of forced disenrollment heading into 2026. Among 28.6 million Medicare Advantage enrollees studied, 2.9 million were forced to find new coverage as their existing plans were terminated or exited their counties. The forced disenrollment rate reached 10% for 2026 coverage, a tenfold increase from the 1% average observed between 2018 and 2024.

KFF's separate analysis found that 2.6 million people enrolled in individual MA-PD plans had their coverage terminated at the end of 2025, representing 13% of all individual MA-PD enrollees. Highmark Health terminated plans affecting 148,000 enrollees (44% of its individual MA-PD enrollment) across four states. Lifetime Healthcare terminated plans affecting 106,000 enrollees (45% of its individual MA-PD enrollment) in New York alone.

The geographic concentration is stark. Twelve states had more than 20% of their MA enrollees facing forced disenrollment. Vermont was the hardest hit, with 92.2% of enrollees affected. Beneficiaries forced to switch were more likely to have been enrolled in PPO plans, non-Special Needs Plans, plans from smaller carriers, and plans with lower Star Ratings. They were nearly twice as likely to live in rural areas or markets with lower MA penetration.

For plan actuaries, forced disenrollment creates a distinct modeling challenge. When 2.9 million members must choose new plans, the risk pool composition of both sending and receiving plans changes simultaneously. Members with complex, ongoing care relationships, particularly those in oncology treatment or managing multiple chronic conditions, are the most sensitive to network disruption and the most likely to switch to plans that preserve their provider access. This produces adverse selection in the receiving plans: the members most motivated to switch carry higher expected claims costs than the average new enrollee.

UnitedHealthcare and Humana County Coverage Shrinks

The two largest MA carriers both reduced their geographic footprints for 2026. UnitedHealthcare pulled out of 109 counties and one state. Humana withdrew from 194 counties and three states, dropping from 89% of U.S. counties to 85%. Together, the two carriers eliminated MA plan availability in 303 counties, concentrating the gaps in rural areas and mid-tier metropolitan markets where provider alternatives are already limited.

The enrollment consequences were asymmetric. UnitedHealthcare shed approximately 930,000 MA members, while Humana gained 1.2 million. Elevance and Centene expanded their county footprints, partially filling the void. The average Medicare beneficiary can still choose from plans offered by eight firms, unchanged from the prior two years. But county-level averages mask the reality for beneficiaries in the 303 counties that lost a major carrier: fewer plan choices, narrower networks, and reduced supplemental benefits.

KFF's 2026 Spotlight confirmed the benefit erosion trend. While 67% of MA-PD plans still charge no additional premium beyond Part B, the share offering supplemental benefits declined across multiple categories: over-the-counter allowances dropped from 73% to 66%, meal benefits from 65% to 57%, remote access technologies from 53% to 48%, and transportation benefits from 30% to 24%. Plans are simultaneously narrowing networks and reducing supplemental benefits, a combination that degrades the value proposition that drove MA enrollment past 34 million beneficiaries and 54% market penetration.

CMS Network Adequacy Framework: The Gap Between Standards and Reality

CMS evaluates Medicare Advantage network adequacy primarily at the application stage, using time-and-distance standards that measure whether 90% of beneficiaries in a county can reach at least one provider in each required specialty within specified travel limits. The thresholds vary by county type (large metro, metro, micro, rural, or counties with extreme access considerations) and by specialty. This framework has remained substantively unchanged since CMS codified it in the 2015 GAO report era (GAO-15-710), despite fundamental shifts in how MA networks are constructed and maintained.

The Application-Stage Problem

The most significant limitation is temporal. CMS measures network adequacy during the application and renewal process, not on an ongoing basis. As GAO-15-710 documented, CMS does not measure active MA organization networks against its adequacy criteria after contracts are awarded. Because a plan's provider composition can change at any time through contract terminations, provider relocations, or practice closures, CMS cannot assure that networks remain adequate until the agency collects compliance evidence on a regular basis. Between 2016 and 2022, CMS sent enforcement letters to only five insurers after seven of their plans failed to meet provider network adequacy requirements.

When 21 health systems exit MA networks in a single year, the adequacy picture can shift dramatically between the application approval date and the plan year start date. A network that met the 90th-percentile distance standard in September may have significant gaps by January if a major health system terminates its contract during the intervening open enrollment period. Enrollees who selected a plan based on published provider directories may discover that their listed providers are no longer available.

Ghost Networks and Directory Accuracy

The HHS Office of Inspector General has documented the "ghost network" problem in MA plans. In a 2025 report examining behavioral health provider networks, OIG found that 72% of inactive providers listed in MA and Medicaid managed care directories should not have been included. Providers were listed at locations where they no longer practiced, at facilities that had closed, or in networks they had formally exited. These ghost listings inflate the apparent size of a network, allowing plans to meet CMS adequacy standards on paper while failing to provide meaningful access in practice.

CMS addressed directory accuracy in the CY2026 final rule (CMS-4208-F2), requiring MA organizations to update provider directory information within 30 days of becoming aware of a change. The rule also mandates that plans submit standardized provider data to CMS for publication on the Medicare Plan Finder website, with annual attestations of accuracy. These requirements took effect for plan years beginning January 1, 2026. For CY2027 applications, CMS added a new Outpatient Behavioral Health facility specialty type to the network adequacy evaluation framework, responding to documented access gaps in mental health services.

The 30-day update mandate is a meaningful improvement, but it still permits a full month of stale directory data after a provider departure. For beneficiaries attempting to schedule care with a specialist, a 30-day lag can mean the difference between timely treatment and a delayed or missed appointment. And the mandate applies only to information the MA organization is "aware of," creating an incentive structure where plans may not actively monitor for provider departures.

The Behavioral Health Dimension

CMS's decision to add Outpatient Behavioral Health as a network adequacy specialty type for CY2027 reflects a recognition that mental health access in MA plans has lagged behind physical health. The OIG behavioral health report found that many MA plans listed providers who were inactive, not accepting new patients, or unreachable at the listed contact information. In markets where a major health system exits an MA network, the behavioral health impact is often disproportionate because academic medical centers frequently serve as the primary source of psychiatric and psychological services for complex patients. The intersection of provider network exits, CMS Star Ratings shifting to 65% clinical weight in 2027, and new behavioral health adequacy requirements creates a compliance challenge that plan actuaries must quantify.

Actuarial Impact: How Provider Exits Cascade Through Risk Pools and Bids

Provider network exits affect MA plan actuarial assumptions through four interconnected channels: risk pool composition, utilization patterns, risk score dynamics, and premium bid construction. Each channel requires distinct modeling adjustments for 2027 bid submissions, which were due the first Monday in June 2026.

1. Risk Pool Composition and Adverse Selection

When a marquee health system like Mayo Clinic exits a plan's network, the members most likely to leave that plan are those with active, complex treatment relationships at the departing system. These members carry higher expected claims costs than the plan average. They are more likely to have multiple chronic conditions, be in active oncology treatment, or require specialist follow-up that is not easily transferred to a new provider. Their departure improves the risk profile of the plan they leave but worsens the risk profile of the plan they join.

For actuaries pricing the receiving plan, the challenge is estimating the incremental morbidity of transfer members. Standard risk adjustment using the V28 model at full weight in 2026 captures diagnosed conditions but does not fully account for the selection effect. Members who actively switch plans to preserve provider access are, by definition, higher utilizers than passive members who accept whatever plan their employer or default enrollment assigns. From tracking MA enrollment transitions in markets with prior provider disruptions, the claims cost of active switchers exceeds their risk-adjusted expected cost by 8% to 15% in the first year, declining to 3% to 5% by year two as conditions stabilize.

2. Utilization Pattern Disruption

Network exits disrupt care continuity in ways that affect both short-term and long-term utilization. In the immediate term, members who lose access to their established providers face delays in ongoing treatment, difficulty transferring medical records, and gaps in care coordination. These disruptions manifest as increased emergency department utilization, duplicated diagnostic testing at new providers, and deferred elective procedures that present later as higher-acuity interventions.

The Mount Sinai/Anthem dispute illustrates the pattern. During the three and a half months that Mount Sinai was out of network, approximately 200,000 affected members had to navigate alternative care pathways. Even after the April 2026 resolution restored in-network access, the utilization disruption during Q1 2026 will appear in claims data as a temporary spike followed by a catch-up effect as deferred care is completed. Actuaries developing 2027 base period experience from 2025 and early 2026 data must identify and adjust for these disruption artifacts to avoid misstating trend.

3. Risk Score Dynamics Under V28

The full implementation of V28 in payment year 2026 compounds the challenge. V28 reduced valid ICD-10 diagnostic codes from 9,797 to 7,770 and is projected to compress average MA risk scores by 3.12%, representing approximately $11 billion in net savings to the Medicare Trust Fund. When provider exits disrupt care continuity, the risk score impact compounds: members who lose their established providers may miss routine encounters that capture and resubmit chronic condition codes. A member whose annual comprehensive exam is delayed from Q1 to Q3 because of a network disruption may have conditions that fall out of the risk adjustment submission window, reducing the plan's risk score for that member without any change in the member's actual health status.

CMS projects risk scores to increase by 2.45% in 2027, but this national projection assumes stable provider relationships and consistent coding patterns. Plans experiencing significant network disruption should model a risk score shortfall of 0.5 to 1.5 percentage points below the national trend for affected member segments, reflecting the coding continuity gap.

4. Premium Bid Construction for 2027

The CMS 2027 rate announcement projects an effective payment increase of approximately 2.48% after accounting for expected MA risk score trend. Against a medical cost trend environment of 6% to 8% in many markets, this creates significant bid pressure. Plans that lost major providers must decide whether to bid aggressively to retain members (accepting lower margins) or price for the network they actually have (risking enrollment decline).

The bid calculation involves estimating the plan's projected revenue, which depends on CMS benchmarks and the plan's risk scores, against projected costs, which depend on network composition and utilization assumptions. When a plan loses a high-cost academic medical center from its network, the immediate cost effect may be favorable: members who stay in the plan but shift to lower-cost community providers reduce average unit cost. But the second-order effects are less favorable: sicker members leave the plan (reducing risk scores more than costs), utilization patterns shift unpredictably, and the plan's competitive position in the market weakens.

Plans that gained marquee providers, particularly those like Elevance and Centene that expanded their county footprints for 2026, face the mirror image: higher unit costs from academic medical center relationships, offset by the ability to attract healthier members drawn to broader network access. The actuarial challenge is calibrating the timing and magnitude of these offsetting effects in a bid submission that locks in pricing for an entire calendar year.

The Prior Authorization Feedback Loop

Prior authorization friction is both a cause and a consequence of the provider exit cycle. Health systems cite chronic denials, slow processing, and administrative burden as primary reasons for leaving MA networks. CMS-0057-F, which took effect in January 2026, now requires MA plans to publish prior authorization denial rates publicly for the first time. The transparency creates a new feedback mechanism: providers can compare authorization burden across insurers, and plans with high denial rates face both regulatory scrutiny and competitive disadvantage in network contract negotiations.

KFF's prior authorization data reveals wide variation across major MA carriers. UnitedHealthcare, the carrier that lost the most provider relationships in 2026, had a denial rate of 12.6% to 12.8% on just 0.9 to 1.0 prior authorization requests per enrollee. Elevance, which expanded its MA footprint, had a denial rate of 4.1% to 4.2% on 2.7 to 3.0 requests per enrollee. The inverse relationship between authorization volume and denial rate suggests structurally different utilization management philosophies. Providers facing the more aggressive denial patterns have more economic incentive to exit, creating the network adequacy gaps that CMS standards were not designed to detect in real time.

For actuaries, the prior authorization transparency data offers a new input for utilization trend assumptions. Plans that reduce denial rates in response to competitive and regulatory pressure (UnitedHealthcare announced it would eliminate prior authorization for 30% of currently covered services by year-end 2026) should model the utilization increase from reduced authorization gatekeeping. Patterns from early gold card implementations and voluntary PA reductions suggest a 5% to 10% utilization increase in affected service categories in the first year.

Why This Matters for Health Plan Actuaries

The convergence of provider network exits, forced disenrollment at scale, and CMS network adequacy standards that measure compliance at application rather than in real time creates a pricing environment where standard actuarial assumptions need recalibration. Five specific adjustments warrant attention for 2027 bid development and beyond.

Risk pool adjustment for transfer members: Plans that gained members from disrupted networks should apply an adverse selection load of 8% to 15% above risk-adjusted expected cost for the first year, declining to 3% to 5% in subsequent years. This load should be applied at the member segment level, not plan-wide, targeting members who actively switched during special enrollment periods triggered by provider exits.

Utilization disruption factor: Base period claims data from 2025 and early 2026 should be examined for disruption artifacts: ED utilization spikes, duplicated diagnostics, and deferred procedure catch-up effects. These should be normalized before projection to avoid overstating base period trend. Separately, actuaries should add a prospective disruption load for any planned or anticipated provider exits in 2027.

Risk score shortfall provision: Plans with significant provider turnover should model a 0.5 to 1.5 percentage point risk score shortfall below national trend for affected member segments, reflecting disrupted coding continuity. This provision interacts with V28's full-weight compression: the combination of model-driven and disruption-driven risk score reduction can exceed the CMS-projected 2.45% risk score trend by a meaningful margin.

Network cost recalibration: When a high-cost academic medical center exits the network, the immediate unit cost reduction may be offset by the loss of complex case management infrastructure. Plans should model both the unit cost benefit and the downstream utilization cost of members navigating care transitions without their established specialists.

Directory accuracy compliance cost: The CMS 30-day directory update mandate, combined with the requirement to submit standardized data to Medicare Plan Finder, represents a new administrative expense line item. Plans should budget for the technology infrastructure, data governance processes, and ongoing provider verification needed to maintain compliance. The cost is modest relative to medical expense, but it is additive to an administrative expense ratio already under pressure from CMS's minimum MLR requirements.

Further Reading