Medicaid expansion states have far lower uninsured rates than non-expansion states, but the mortality gap between the two groups mostly disappears once you control for region

Summary

Yes, but only on one of the two outcomes that matter, and the size of the gap depends on what you hold fixed. The 41 states (plus DC) that adopted the ACA's Medicaid expansion have a materially lower uninsured rate than the 10 states that have not -- 6.9% vs 12.8% of the population, a gap that is statistically strong (p=0.0035) and survives controlling for region. On premature death (years of life lost before age 75, a proxy for overall mortality; figures below use only the state-level rows of health.chr_premature_death, excluding that table's county rows and its one national rollup row which share the same geo_level column), the raw gap is smaller and only marginally significant (p=0.054), and it is not robust: dropping any one of four Southern non-expansion states (Alabama, Mississippi, South Carolina, or Tennessee) from the comparison pushes it back above the conventional significance threshold. When region (the Census-defined South, 16 states) is added as a control, the non-expansion coefficient on premature death collapses from a nominal +1,419 years-lost/100k to a statistically insignificant +336 (p=0.58), while the South-region coefficient itself is large and highly significant (+2,375, p<0.0001). In plain terms: non-expansion states clearly look different on insurance coverage -- that gap is expansion's own footprint, not a Southern-states artifact. They do not look meaningfully different on premature mortality once you account for the fact that most non-expansion states are also Southern states with a long-standing regional mortality disadvantage that predates the ACA.

Method

Classification. Non-expansion states (per KFF's tracker, current status as of 2026): Alabama, Florida, Georgia, Kansas, Mississippi, South Carolina, Tennessee, Texas, Wisconsin, Wyoming (10 states). All other states plus DC (41 units) are classified as expansion. This corpus has no Medicaid-expansion-status column of its own (search_catalog returned no match for 'expansion'), so the classification was built from KFF's published list rather than an AskAmerica table -- flagged here as an external input, not a warehouse-derived fact.

Outcomes tested. (1) Uninsured rate: census.acs_health_insurance, 2023 ACS 5-year estimates (the latest year loaded; the source publishes roughly 2 years behind, so 2024-2026 are not available yet), population-weighted across each group's states. (2) Premature death rate (years of potential life lost before age 75, per 100,000, age-adjusted): health.chr_premature_death, 2025 County Health Rankings release. IMPORTANT DATA-QUALITY NOTE: this table's geo_level column mixes 'state', 'county', and a single 'nation' rollup row in one field -- an unfiltered aggregate over it silently double-counts (a rollup_contamination diagnostic fired on an exploratory count query earlier in this session that summed all three levels together without a filter). Every figure in this report uses an explicit WHERE geo_level='state' filter, so the 9,885 / 8,466 YPLL figures and everything derived from them are unaffected by that contamination risk -- flagging it here because the same table could easily be mis-aggregated by a future query that omits the filter.

Tests run. Welch's two-sample t-test for each outcome by group; a sensitivity_analysis (leave-one-state-out refit) on the premature-death gap to check whether any single state was carrying the result; and an OLS regression of each outcome on non-expansion status plus a South-census-region indicator (16 states: DE, DC, FL, GA, MD, NC, SC, VA, WV, AL, KY, MS, TN, AR, LA, OK, TX) to test whether the raw group gap survives once region is held fixed -- region is a plausible confound because 7 of the 10 non-expansion states (AL, FL, GA, MS, SC, TN, TX) are Southern states with a long, well-documented regional mortality disadvantage independent of Medicaid policy. This is a single cross-sectional snapshot of the current year's data only; no other analytical method was run in this session.

Finding 1 — Insurance coverage: the gap is large and holds up

Non-expansion states have a population-weighted uninsured rate of 12.8% versus 6.9% in expansion states -- roughly 5.9 percentage points, or about 85% relatively higher. A Welch t-test across the 51 units gives t=3.64, p=0.0035. Controlling for South-region membership in an OLS regression, the non-expansion coefficient stays large and significant: +3.23 percentage points (p=0.0013), while the South coefficient itself is not significant (p=0.13) once expansion status is in the model. This is directionally consistent with published tracking (KFF; Commonwealth Fund) showing expansion states cut their uninsured rate roughly twice as much as non-expansion states between 2013 and 2022 (30% to 15% vs 30% to 24%), though that is a separate figure from a different source, not one recomputed here.

Finding 2 — Premature mortality: the gap is fragile and is mostly a regional artifact

The raw group means (state-level rows only, per the geo_level filter described above) are 9,885 years-of-life-lost/100k in non-expansion states vs 8,466 in expansion states (+1,419, or about 17% relatively higher), with a Welch t-test p=0.054 -- borderline. A leave-one-out sensitivity check (refitting a bivariate regression of premature death on non-expansion status, dropping one state at a time) shows the result is not robust: the coefficient's significance flips above p=0.05 when Alabama, Mississippi, South Carolina, or Tennessee is individually removed, and the coefficient itself ranges from 1,036 to 1,690 across the 51 leave-one-out refits -- a wide range for a result this close to the threshold. Once a South-census-region indicator is added to the regression, the picture changes further: the non-expansion coefficient falls to +336 years-lost/100k and becomes statistically insignificant (p=0.58), while the South coefficient is +2,375 and highly significant (p<0.0001, overall model R²=0.37). In other words, the apparent mortality difference between expansion and non-expansion states in this cross-section is much better explained by which states are in the South than by which states expanded Medicaid -- an important distinction the raw two-group comparison alone would miss.

This finding is narrow and purely descriptive, and it is not a challenge to separately published peer-reviewed research (a 2012 NEJM study and a 2021 Lancet Public Health national cohort study, both linked below) that finds real mortality reductions attributable to Medicaid expansion using each study's own longitudinal data over time and its own statistical design -- neither of those designs was run in this session; only the cross-sectional regressions described above were run here. In a same-year cross-section, non-expansion status and Southern region are correlated enough (7 of 10 non-expansion states are Southern) that the raw mortality gap computed here cannot be attributed to expansion status without controlling for region -- and once region is controlled in the regression actually run above, the expansion-status effect is not distinguishable from zero.

Bottom line

Medicaid expansion states look meaningfully, robustly different from non-expansion states on insurance coverage today -- that gap is real and not explained away by region. They do not look meaningfully different on premature mortality once region is accounted for in this cross-section; the modest raw mortality gap that exists is fragile (sensitive to a handful of Southern non-expansion states) and is better explained by the South's long-standing regional health disadvantage than by Medicaid expansion status alone. Separately published research using each study's own longitudinal data over time (cited above, not run here) has found real mortality effects from expansion that a single-year snapshot like this one cannot capture.

Every query behind this report

10 warehouse calls ran in this session, in order. Each is reproducible against the same snapshot.

SELECT geo_level, MAX(release_year) as max_year, COUNT(*) FROM health.chr_premature_death GROUP BY geo_level

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 'non_expansion' ELSE 'expansion' END AS status
  FROM geo.state_ref
)
SELECT e.status,
  COUNT(*) AS n_states,
  SUM(a.no_health_insurance) AS total_uninsured,
  SUM(a.total_population) AS total_pop,
  100.0*SUM(a.no_health_insurance)/SUM(a.total_population) AS uninsured_pct
FROM census.acs_health_insurance a
JOIN geo.state_ref g ON g.state_fips = a.state
JOIN exp e ON e.state_abbr = g.state_abbr
WHERE a.year = '2023' AND a.geography = 'state'
GROUP BY e.status

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 'non_expansion' ELSE 'expansion' END AS status
  FROM geo.state_ref
)
SELECT e.status, c.state_abbr, c.premature_death_rate
FROM health.chr_premature_death c
JOIN exp e ON e.state_abbr = c.state_abbr
WHERE c.geo_level = 'state' AND c.release_year = 2025
ORDER BY e.status, c.state_abbr

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 'non_expansion' ELSE 'expansion' END AS status
  FROM geo.state_ref
)
SELECT e.status, g.state_abbr, 100.0*a.no_health_insurance/a.total_population AS uninsured_pct
FROM census.acs_health_insurance a
JOIN geo.state_ref g ON g.state_fips = a.state
JOIN exp e ON e.state_abbr = g.state_abbr
WHERE a.year = '2023' AND a.geography = 'state'
ORDER BY e.status, g.state_abbr

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 'non_expansion' ELSE 'expansion' END AS status
  FROM geo.state_ref
)
SELECT e.status, 100.0*a.no_health_insurance/a.total_population AS uninsured_pct
FROM census.acs_health_insurance a
JOIN geo.state_ref g ON g.state_fips = a.state
JOIN exp e ON e.state_abbr = g.state_abbr
WHERE a.year = '2023' AND a.geography = 'state'

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 'non_expansion' ELSE 'expansion' END AS status
  FROM geo.state_ref
)
SELECT e.status, c.premature_death_rate
FROM health.chr_premature_death c
JOIN exp e ON e.state_abbr = c.state_abbr
WHERE c.geo_level = 'state' AND c.release_year = 2025

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 1 ELSE 0 END AS non_expansion
  FROM geo.state_ref
)
SELECT e.state_abbr, e.non_expansion, c.premature_death_rate
FROM health.chr_premature_death c
JOIN exp e ON e.state_abbr = c.state_abbr
WHERE c.geo_level = 'state' AND c.release_year = 2025

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 1 ELSE 0 END AS non_expansion,
    CASE WHEN state_abbr IN ('DE','DC','FL','GA','MD','NC','SC','VA','WV','AL','KY','MS','TN','AR','LA','OK','TX') THEN 1 ELSE 0 END AS south
  FROM geo.state_ref
)
SELECT e.non_expansion, e.south, c.premature_death_rate
FROM health.chr_premature_death c
JOIN exp e ON e.state_abbr = c.state_abbr
WHERE c.geo_level='state' AND c.release_year=2025

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 1 ELSE 0 END AS non_expansion,
    CASE WHEN state_abbr IN ('DE','DC','FL','GA','MD','NC','SC','VA','WV','AL','KY','MS','TN','AR','LA','OK','TX') THEN 1 ELSE 0 END AS south
  FROM geo.state_ref
)
SELECT e.non_expansion, e.south, c.premature_death_rate
FROM health.chr_premature_death c
JOIN exp e ON e.state_abbr = c.state_abbr
WHERE c.geo_level='state' AND c.release_year=2025

WITH exp AS (
  SELECT state_abbr,
    CASE WHEN state_abbr IN ('AL','FL','GA','KS','MS','SC','TN','TX','WI','WY') THEN 1 ELSE 0 END AS non_expansion,
    CASE WHEN state_abbr IN ('DE','DC','FL','GA','MD','NC','SC','VA','WV','AL','KY','MS','TN','AR','LA','OK','TX') THEN 1 ELSE 0 END AS south
  FROM geo.state_ref
)
SELECT e.non_expansion, e.south, 100.0*a.no_health_insurance/a.total_population AS uninsured_pct
FROM census.acs_health_insurance a
JOIN geo.state_ref g ON g.state_fips=a.state
JOIN exp e ON e.state_abbr=g.state_abbr
WHERE a.year='2023' AND a.geography='state'