Medicare Advantage Plan Exits Force 3 Million to Switch in 2026

A JAMA research letter published February 18, 2026 by Johns Hopkins researchers put the number on the exit wave: roughly 2.9 million Medicare Advantage enrollees, about 10% of the non-employer HMO and PPO population, had to find new coverage for 2026 after their plan left their county. That rate averaged just over 1% a year from 2018 to 2024. It reached 6.9% in 2025. A tenfold move in two years.

Key Takeaways

• 2.9 million enrollees forced to switch, a 10% rate against a 2018 to 2024 average just above 1% and 6.9% in 2025.
• MA enrollment is projected at 34 million for 2026, down from 34.9 million, taking MA's share of Medicare from 50% to roughly 48%, its first decline in nearly two decades.
• V28 cut valid diagnostic codes from 9,797 to 7,770 and CMS projected average risk scores falling 3.12%, worth roughly $11 billion in reduced payments.
• One Humana contract covering about 45% of its MA members fell from 4.5 to 3.5 stars, removing the 5% quality bonus from that enrollment in a single rating cycle.
• Total plan value added fell more than 7% year over year on Milliman's MACVAT, with medical deductibles rising 72% from $33 to $57.

Who Left, and Where

The three largest carriers all shrank their footprints. UnitedHealthcare dropped one state and 109 counties, displacing roughly 180,000 enrollees while remaining the largest carrier by enrollment. Humana cut the most, exiting 194 counties and three states, moving from 89% to 85% of US counties and from 48 states to 46. Aetna will offer plans in 43 states and 2,159 counties, down from 44 and 2,259, discontinuing about 90 plans across 34 states, most of them PPO products.

Smaller carriers left outright. Clear Spring Health closes its entire MA business effective June 1, 2026 across Illinois, Georgia and Colorado, and those exits fall hardest in rural counties where choice was already thin.

The aggregate effect is a reversal. CMS projects total MA enrollment at 34 million for 2026, down from 34.9 million, with MA's share of Medicare beneficiaries falling from 50% to approximately 48%.

Providers moved in the same direction. At least 21 health systems dropped one or more MA contracts for 2026, including Mayo Clinic going out-of-network with most UnitedHealthcare and Humana plans, on prior authorization denial rates and slow reimbursement.

The Revenue Side Moved Before the Footprint Did

The JAMA study's risk factors read as a list of plan characteristics. They are better read as a list of revenue mechanisms.

Risk adjustment. The CMS-HCC model completed its phase-in to Version 28 for payment year 2026, at 33% in PY 2024, 67% in PY 2025 and 100% now. V28 cut valid diagnostic codes from 9,797 under V24 to 7,770 while restructuring categories from 86 to 115, dropping diagnoses CMS assessed as weak cost predictors. CMS projected average risk scores declining 3.12%, roughly $11 billion. For carriers concentrated in chronic populations the effect exceeded that, with reported declines of 4 to 5% on specific sub-populations.

Quality bonuses. Contracts at four stars or above receive a 5% bonus on the county benchmark. The 2026 average contract rating is 3.98, just under the threshold, and only about 40% of contracts qualify. Humana's example is the clearest: one contract covering roughly 45% of its MA members fell from 4.5 to 3.5 stars, removing bonus eligibility from that base in one cycle.

Now the exit pattern resolves. PPOs were dropped more than HMOs because out-of-network access raises the loss ratio with no offsetting revenue. Non-SNP plans were dropped more than D-SNPs because SNP populations code into HCCs that survived V28. Sub-four-star plans and smaller carriers were dropped because the first lost the 5% bonus and the second lacks the scale to negotiate provider rates against it.

Each exit is a county where benchmark plus risk-adjusted revenue plus bonus stopped covering trend. MedPAC's March 2026 report still estimates MA overpayments at $76 billion for 2026, 14% above what the same beneficiaries would cost in traditional Medicare, which is the tension: the program remains overpaid in aggregate while individual county-plan combinations are not viable.

Cutting Benefits Is the Other Lever, and It Feeds the Problem

Where carriers stayed, they repriced, and Milliman's MACVAT measures how far.

Total value added fell more than 7% from 2025 to 2026, the largest annual drop in recent history. Part C benefit value dropped about $17 per member per month and Part D about $4. Average medical maximum out-of-pocket rose from $5,100 to $5,440. Medical deductibles jumped 72%, from $33 to $57. Part D deductibles moved from roughly $230 to $375, with 83% of members now facing one against 23% in 2024, and zero-premium MA-PD plans fell by 231, a 9.5% decrease.

The headline premium moved the other way. CMS estimates average monthly premium falling from $16.40 to $14.00. Both numbers are real and they describe the same decision: a shift from premium-funded to cost-sharing-funded plan design. The financial risk moves from the plan to the member, which changes utilization behavior and therefore the claim distribution the plan is pricing against.

Members responded immediately. Milliman observed carriers with declining value added losing 4% to 33% of enrollment, while one carrier that improved its position grew 22%.

That is where the two levers collide. The members most able to leave a thinner plan are the ones with the fewest care needs and the least attachment to a network. Cutting benefits to protect margin removes the enrollees whose costs the benchmark covers and retains the ones whose diagnoses V28 no longer pays for at the old weight. The margin problem that drove the benefit cut is made worse by the selection the benefit cut produces, and the risk score correction that started it does not reverse.