When Congress allowed pandemic-era Medicaid continuous coverage to expire in March 2023, states began the largest eligibility review in the program's history, and over the following eighteen months disenrolled more than 25 million people, per KFF's unwinding tracker, which compiled state reports through the process's end in mid-2024. The headline number understates the pattern: in state after state, a majority of disenrollments occurred for procedural reasons — returned mail, missed deadlines, unprocessed paperwork — rather than a finding that the person no longer qualified. People of color and people with limited English proficiency were disenrolled at higher rates, and several states' own equity audits confirmed the gap.
What was the unwinding?
From March 2020, a congressional condition on extra federal funding barred states from removing anyone from Medicaid, and enrollment grew by roughly 23 million. The deal's end launched unwinding: every enrollee's eligibility had to be redetermined, a task states had not performed at scale in three years, executed by agencies with chronic understaffing and legacy data systems. States proceeded at different speeds under federal oversight, with the Centers for Medicare and Medicaid Services requiring corrective action plans for the worst performers — a list that at various points included large states with disproportionate procedural disenrollments.
Why did paperwork drive the losses?
Eligibility redetermination depends on reaching people at addresses and income circumstances that had changed since 2020. Renewal packets went to stale addresses; income checks hit wage databases with errors; call centers had hours-long waits; and the burden fell on enrollees to prove they qualified for a program designed for exactly people with unstable circumstances. CMS reporting rules pushed states to break out procedural disenrollments, and the numbers were stark: nationally, roughly 70 percent of unwinding disenrollments were procedural, per KFF's aggregation. Public-health researchers connected the mechanism to churn — the well-documented pattern that coverage interruptions cause care interruptions, medication gaps, and delayed diagnoses, with re-enrollment often following only after a health crisis.
Who lost coverage?
State-level disaggregation, where published, showed Black and Latino enrollees disenrolled at higher rates than white enrollees — data reported in state equity dashboards and analyses of states like Virginia and Pennsylvania, while several states declined to publish race data at all, a gap KFF flagged throughout. Children were a distinct story: the share of kids uninsured rose, per early indications from national survey data, and Congress responded with a permanent continuous-eligibility option for children that states began adopting. An unknown share of adults disenrolled from Medicaid transitioned to marketplace coverage — federal estimates during the process put successful transitions well under half of those eligible, with the rest becoming uninsured or re-enrolling later.
What does the coverage map look like now?
The unwinding interacted with the country's deepest health-policy divide: the ten holdout states that never expanded Medicaid under the Affordable Care Act, where adults below the poverty line have no pathway at all — a coverage gap of several million people concentrated in the South. Post-unwinding enrollment settled below the pandemic peak but above 2019 levels, and the experience hardened a policy lesson documented in program research: eligibility design is retention design. States that adopted multi-year renewals, ex parte automated renewals through data matches, and pre-populated forms retained more people, and those mechanisms — not outreach campaigns — tracked with lower procedural loss.
What should readers watch?
Medicaid enrollment moves with policy cycles again: work-requirement waivers, immigration-related eligibility rules, and federal financing debates all carry coverage consequences at unwinding scale. The program's own data infrastructure — state dashboards, CMS enrollment reports, KFF tracking — now exists and proved its value, and the unwinding's documented harms gave Congress, for the first time, a real-time national dataset on what procedural churn costs. Whether that dataset informs the next round of eligibility change, or is ignored in favor of budget arithmetic, is the live question.
For more context, read How America measures hunger: the survey behind the food-budget numbers.
For more context, read bead program broadband funding.
For more context, read veterans treatment court explained.