2.9 Million Members Moved in 2026. The 2027 Cycle Is Already Building.

In February 2026, researchers at Johns Hopkins Bloomberg School of Public Health and Georgetown published a research letter in JAMA that gave the Medicare industry a number it had been circling for a year: roughly 2.9 million Medicare Advantage members were forced to find new coverage for 2026 because their plan stopped operating where they lived.

That is about 1 in 10 enrollees in the study sample. For context, the average annual forced disenrollment rate from 2018 through 2024 was just over 1%. It rose to 6.9% in 2025. It hit 10% in 2026. A tenfold move in two years, in a program that had spent two decades being boringly stable.

What actually drove it

None of this is cyclical. Three things are squeezing carriers at once.

Utilization has stayed elevated. The care seniors deferred in 2020 and 2021 worked its way back into the system and never fully normalized. Medical loss ratios across MA have run hot ever since.

Reimbursement has not kept pace. CMS finalized a 2.48% average payment increase for 2027, which comes to 4.98% once risk score trend is included. That is a step down from the 5.06% carriers got for 2026, and it arrived after a lobbying campaign to lift it off the 0.09% CMS floated in the advance notice. Carriers planned their 2027 bids in that gap.

Star ratings are doing structural damage. Plans rated 4 stars and above get their benchmark raised by 5 percentage points, or 10 in double-bonus counties. Drop to 3.5 and that goes away entirely. KFF puts total quality bonus spending above $13B for 2026, so at plan scale this is a real revenue cliff, and it forces a decision: absorb the loss or leave the market.

Rural books took roughly double the hit

The study found the disruption was not evenly distributed. Members in rural areas, and in markets with lower MA penetration, were disrupted at roughly twice the rate of everyone else. PPO enrollees, members of smaller carriers, and members of lower-rated plans were all disproportionately affected. Smaller carriers accounted for about half of all disruptions; UnitedHealthcare alone was close to 14%, with Aetna and Elevance each near 8%. In 7 states more than 40% of MA enrollees had to switch, with Vermont at 92%. Idaho, Wyoming, North Dakota, South Dakota, Maryland and New Hampshire all cleared the 40% line.

The economics behind that are unforgiving in four directions at once. Per-member costs run higher because there are fewer providers, longer travel distances and higher acuity. CMS benchmark rates are lower in many rural counties. Risk pools are small enough that a handful of expensive members moves the whole medical loss ratio. And there is almost no room to negotiate when there are only two hospitals in a hundred miles.

Thin margins in good years become negative margins in a squeeze. When a carrier trims what Humana's CFO described as the lower tail of profitability, rural counties are what the lower tail looks like.

Those are also exactly the books that are hardest to replace and easiest to lose, which is the subject of the third post in this series.

What is already lined up for 2027

Humana confirmed on its July 29 earnings call that plan exits for 2027 will affect approximately 600,000 members, about 8% of its 7.2 million MA book. The company expects to recapture just over 40% of them into other Humana plans, in line with what it managed after the 2025 exits.

Run that out. Roughly 360,000 members will not land in another Humana plan. They will shop, or they will go to traditional Medicare, or they will do nothing and get auto-assigned somewhere they did not choose. Letters land in September.

Humana is the one that has put a number on it. UnitedHealthcare exited plans covering more than 600,000 members in the 2026 cycle and is reportedly weighing a preliminary 2027 list covering 34 counties across 12 states, though the company has said that list is not final. Aetna and Elevance are working the same math. Becker's running tally of health systems dropping or narrowing MA contracts in 2026 stands at 25 and is still being updated, which is the provider side of the value chain saying the economics do not work either.

Why this is a book problem, not a market problem

Every one of those displaced members is on somebody's book of business right now. Somebody wrote them, services them, and gets paid a renewal on them.

When the letter arrives, the member calls someone. If the agency that wrote them has not called first, the member is fielding pitches from whoever did. And under 2027 CMS caps, replacing that member costs $725 in commission against $363 to keep them, before any lead spend at all.

The retention window is real but narrow. It opens when carriers send their CMS notification and termination letters in September and closes when AEP ends December 7.

The data is public and mostly unusable

Here is what frustrates us about all of this: CMS publishes everything you need to see it coming.

Plan crosswalks show which contracts are terminating, which are consolidating, and where CMS directs affected members. Service area files show which counties a plan is dropping. Star ratings show which plans are about to lose their quality bonus. Landscape files show what is available in each county, with premium, MOOP and benefits attached.

It is all free, and it is all published months before AEP opens.

It is also spread across 9 separate CMS and HUD datasets, delivered as tab-delimited files, keyed inconsistently between sources, with column names written for a CMS analyst rather than an agent. Reading them correctly and joining them without silently mismatching a plan is a data engineering job. Most agencies do not have one, and should not need one.

What we built

ShieldLync is our answer to that gap. You send plan IDs and counties, no PHI required (if you want a member by member list, you can also upload names) and no integration to build. We run your book against the CMS disruption data and return a ranked list of at-risk members, replacement recommendations with real benefit-level comparisons, and call sheets your agents can work from.

It runs before AEP because that is when it is useful. A perfect list delivered October 20 is a report. A good list delivered in September is a retention program.

The 2027 exits are already decided. The bids are in, the letters are printed, and the only variable left is who reads the data first.