US Coverage Changes During Medicaid Unwinding in 2023.

Importance: The unwinding of Medicaid's continuous coverage provision, beginning in April 2023, disenrolled more than 25 million people. Effects on overall insurance coverage-including reenrollment in Medicaid-and affordability of care remain unclear. Objective: To estimate variation in coverage losses during unwinding of the Medicaid continuous coverage provision by state and demographic groups and evaluate trends in new Medicaid applications before, during, and after the COVID-19 public health emergency. Design, Setting, and Participants: In this cross-sectional study, using administrative data from the US Centers for Medicare & Medicaid Services (CMS), state-level Medicaid terminations as a share of pre-unwinding Medicaid enrollment were estimated during the first year of unwinding and new application rates were examined. CMS administrative enrollment data were compared with quarterly estimates from the National Health Insurance Survey (2019-2023), examining demographic changes in self-reported Medicaid, uninsurance, and cost-related barriers to care among low-income individuals in the second half of 2023 using an interrupted time series. A nationally representative sample of US residents younger than 65 years applying to and enrolled in Medicaid in federal survey data was included. Subgroup analyses were conducted by health status, race and ethnicity, education, pregnancy status, and age. Survey data were collected from January 2019 to December 2023, administrative data were collected from January 2019 to April 2024, and data were analyzed from November 2024 to July 2025. Exposures: The unwinding of the Medicaid continuous coverage provision, under which people could maintain program enrollment even if their eligibility status changed. Main Outcomes and Measures: Enrollment in Medicaid (administratively reported and self-reported), percentage uninsured, cost-related delays in care, and new applications to Medicaid in states with high and low procedural termination rates. Results: As a share of pre-unwinding Medicaid enrollment, the median (range) state termination rate during unwinding in 2023 was 26.6% (8.3-55.3). Net enrollment decreases, including new and returning enrollees, were smaller (median [range] decrease, 13.9% [-15.0 to 30.2]). New applications fell 29.8% (from a mean [SD] of 27 914.20 [3669.97] to 19 589.61 [2622.08]) during the COVID-19 pandemic but increased 29.0% (from a mean [SD] of 19 589.61 [2622.08] to 25 273.11 [3650.18]) during unwinding. Gaps between administratively measured and survey-measured enrollment grew during the COVID-19 pandemic but narrowed during unwinding. During unwinding, survey-reported Medicaid enrollment decreased among women (-2.74 percentage points; 95% CI, -4.82 to -0.66), healthier individuals (-2.65 percentage points; 95% CI, -4.84 to -0.46), young adults (-3.37 percentage points; 95% CI, -6.32 to -0.42), recently pregnant adults (-8.19 percentage points; 95% CI, -16.35 to -0.03), and White individuals (-2.85 percentage points; 95% CI, -5.36 to -0.34). Changes in uninsurance overall were modest but rose significantly among healthier individuals. Cost-related barriers to care rose among White individuals, college-educated individuals, and healthier individuals. Conclusions and Relevance: In this cross-sectional study, millions lost Medicaid during unwinding (with wide state variation), but survey data indicated smaller coverage reductions. Re-enrollment after termination and unawareness of COVID-19 pandemic coverage partially explain this discrepancy. Coverage losses were concentrated among certain demographic groups, with some evidence of reduced care affordability. These results may inform expectations about recent Medicaid proposals that would increase administrative barriers to coverage.