Counties with the most uninsured people: huge population, disproportionately Texas/non-expansion states, high Hispanic/immigrant share

Summary

Ranked by the raw NUMBER of uninsured residents (Census SAHIE, 2023), the top 30 U.S. counties share three traits: (1) they are simply the country's biggest counties by population — county population explains most of the variance in uninsured count (r=0.89) — (2) they are disproportionately concentrated in states that had not adopted ACA Medicaid expansion as of 2023, above all Texas, which places 10 of the 30 (Harris, Dallas, Tarrant, Bexar, Hidalgo, El Paso, Travis, Collin, Montgomery, Denton); and (3) they carry Hispanic/Latino and foreign-born population shares well above the national average (~35% Hispanic vs 20.0% nationally; ~24% foreign-born vs 13.9% nationally in the top 30). Counties in non-expansion states average a 13.95% uninsured rate vs 8.58% in expansion states — a gap that closely matches KFF's independently published, population-weighted 14.5%-vs-8.0% figure.

Step 1 — The ranking itself, and why it is mostly a population story

Source: census.sahie_insurance (Small Area Health Insurance Estimates), county grain, 2023 (SAHIE publishes roughly 3 years behind, so 2023 is the most recent year loaded). The top of the list: Harris County TX (922k uninsured, 22.2% rate), Los Angeles County CA (703k, 8.8%), Dallas County TX (521k, 23.3%), Maricopa County AZ (447k, 11.8%), Cook County IL (372k, 9.0%), Tarrant County TX (331k, 17.6%), Miami-Dade County FL (321k, 14.8%), Bexar County TX (301k, 16.9%), Clark County NV (260k, 13.4%), and Hidalgo County TX (212k, 27.5%).

Across all counties with a usable SAHIE estimate, the correlation between total population and uninsured COUNT is r=0.89 — strongly positive, as expected for any raw count of a roughly-constant-rate phenomenon. The correlation between population and uninsured RATE, by contrast, is essentially zero (r=-0.06). This means the ranking by raw count is mechanically dominated by county size: Los Angeles County appears near the top with a below-average 8.8% uninsured rate purely because it has 9.8 million residents. The genuinely distinctive finding is not in the count ranking itself but in which large counties have unusually HIGH rates on top of their large populations — and that is where Texas and Florida counties stand out (17-28% uninsured vs 5-9% for similarly large counties in Medicaid-expansion states like California, Illinois, Washington, and New York).

Disclosed exclusion: every query behind this analysis filtered with the predicate s.uninsured_count IS NOT NULL. Against the full SAHIE county table this predicate (s.uninsured_count IS NOT NULL) removes exactly one row: Kalawao County, HI (population 43, Hawaii's historic Kalaupapa leprosy-settlement county), for which SAHIE publishes no insurance estimate at all — too small for the Census Bureau's small-area model to produce one. All 3,142 remaining counties are otherwise included; removing this single 43-person county has no material effect on any reported statistic.

Step 2 — Medicaid expansion status is the sharpest single dividing line

States without 2023 ACA Medicaid expansion (Alabama, Florida, Georgia, Kansas, Mississippi, South Carolina, Tennessee, Texas, Wisconsin, Wyoming) contain 970 of the 3,142 counties left after the s.uninsured_count IS NOT NULL filter (see Step 1's disclosed exclusion of Kalawao County, HI), but their average county uninsured rate is 13.95%, versus 8.58% for the 2,173 counties in expansion states — a 5.4 percentage-point gap. This tracks closely with the independently published KFF figure for the same year: 14.5% in non-expansion states vs 8.0% in expansion states (KFF, 'Key Facts about the Uninsured Population'). Ten of the top 30 uninsured-count counties sit in Texas, plus 5 more in Florida — the two largest non-expansion states by population — even though non-expansion states hold well under a third of all U.S. counties. Large California, Illinois, Washington, and New York counties also appear on the list (their sheer size still produces big raw counts) but at roughly half the uninsured rate of similarly sized Texas/Florida counties.

Step 3 — Demographic composition: Hispanic and immigrant population share

The top-30 counties carry a Hispanic/Latino population share averaging roughly 35%, compared with 20.0% nationally (computed from ACS 2019-2023 5-year estimates, county grain). Several counties are far above that: Hidalgo County TX 91.9%, El Paso County TX 82.7%, Miami-Dade County FL 68.8%, Bexar County TX 59.5%, San Bernardino County CA 54.6%. Foreign-born population share averages roughly 24% in the top 30 vs 13.9% nationally, with Miami-Dade at 54.3% and Queens County NY at 47.6%. This matches the well-documented national pattern (KFF, 'Key Facts on Health Coverage of Immigrants'): Hispanic adults had an 18.4% uninsured rate in 2024 nationally, and roughly 46% of likely undocumented immigrant adults and 21% of lawfully present immigrant adults report being uninsured, versus 6-7% for citizens — both well above the 8% national average. Large metro counties with substantial Hispanic and immigrant populations concentrate this gap into a small number of very populous counties.

What this does and does not show

This is a descriptive, not causal, comparison: Medicaid expansion status is correlated with lower uninsured rates in this data and in the independently published KFF figures, but Texas and Florida also differ from expansion-state peers on immigration composition, workforce structure (higher shares of small-business/agricultural/service employment without employer coverage), and state eligibility rules more broadly — this analysis does not isolate the expansion effect from those other factors. The raw-count ranking itself should not be read as 'these counties have unusually bad insurance access' without separating out the population-size effect, which is why the rate comparison (Step 2) and demographic comparison (Step 3) are reported alongside the count ranking rather than in place of it.

Every query behind this report

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

SELECT s.county_fips, s.geo_name, s.state, s.uninsured_count, s.uninsured_rate, s.insured_count
FROM census.sahie_insurance s
WHERE s.year = '2023' AND s.geography = 'county' AND s.type = 'Total'
ORDER BY s.uninsured_count DESC
FETCH FIRST 30 ROWS ONLY

SELECT DISTINCT type, geography FROM census.sahie_insurance WHERE year='2023' LIMIT 20

SELECT county_fips, geo_name, state, uninsured_count, uninsured_rate, insured_count
FROM census.sahie_insurance
WHERE "year" = '2023' AND geography = 'county'
ORDER BY uninsured_count DESC
FETCH FIRST 30 ROWS ONLY

WITH top30 AS (
  SELECT county_fips, geo_name, state, uninsured_count, uninsured_rate, insured_count
  FROM census.sahie_insurance
  WHERE "year" = '2023' AND geography = 'county'
  ORDER BY uninsured_count DESC
  FETCH FIRST 30 ROWS ONLY
)
SELECT t.geo_name, t.state, t.uninsured_count, t.uninsured_rate,
       n.total_population AS pop,
       ROUND(100.0*r.hispanic_latino/NULLIF(n.total_population,0),1) AS hispanic_pct,
       ROUND(100.0*n.foreign_born/NULLIF(n.total_population,0),1) AS foreign_born_pct,
       ROUND(100.0*p.below_poverty/NULLIF(p.poverty_universe,0),1) AS poverty_pct
FROM top30 t
LEFT JOIN census.acs_race_ethnicity r ON r.county_fips = t.county_fips AND r."year" = '2023'
LEFT JOIN census.acs_nativity n ON n.county_fips = t.county_fips AND n."year" = '2023'
LEFT JOIN census.acs_poverty p ON p.county_fips = t.county_fips AND p."year" = '2023'
ORDER BY t.uninsured_count DESC

SELECT corr(CAST(s.uninsured_count AS DOUBLE), CAST(n.total_population AS DOUBLE)) AS corr_count_pop,
       corr(s.uninsured_rate, CAST(n.total_population AS DOUBLE)) AS corr_rate_pop,
       count(*) AS n
FROM census.sahie_insurance s
JOIN census.acs_nativity n ON n.county_fips = s.county_fips AND n."year" = '2023'
WHERE s."year" = '2023' AND s.geography = 'county' AND s.uninsured_count IS NOT NULL

SELECT
  CASE WHEN s.state IN ('01','12','13','20','28','45','47','48','55','56') THEN 'non_expansion_2023' ELSE 'expansion_2023' END AS medicaid_status,
  count(*) AS n_counties,
  avg(s.uninsured_rate) AS avg_uninsured_rate,
  sum(s.uninsured_count) AS total_uninsured
FROM census.sahie_insurance s
WHERE s."year" = '2023' AND s.geography = 'county' AND s.uninsured_count IS NOT NULL
GROUP BY CASE WHEN s.state IN ('01','12','13','20','28','45','47','48','55','56') THEN 'non_expansion_2023' ELSE 'expansion_2023' END

SELECT
  SUM(r.hispanic_latino)*1.0/SUM(n.total_population) AS us_hispanic_pct,
  SUM(n.foreign_born)*1.0/SUM(n.total_population) AS us_foreign_born_pct
FROM census.acs_nativity n
JOIN census.acs_race_ethnicity r ON r.county_fips = n.county_fips AND r."year" = n."year"
WHERE n."year" = '2023' AND n.geography = 'county'