Rural hospital closures push people ~20 miles farther for care, and raise inpatient mortality risk 6-9% for time-sensitive conditions

Summary

When a rural hospital closes, the nearest remaining hospital is typically about 3-8 miles farther away for the loss of the "nearest" facility (one study found median distance rising from 13.1 to 16.4 miles), and roughly 20 miles farther for the specific services (like inpatient or emergency care) the closed hospital used to provide, per a systematic review of the closure literature. Less common services, like inpatient substance-abuse treatment, can require traveling ~40 miles farther. On mortality: the best-designed single study (Gujral & Basu, using California hospital discharge data 1995-2011) found that rural hospital closures raised adjusted inpatient mortality for time-sensitive conditions (heart attack, stroke, sepsis, asthma/COPD combined) by 0.46-0.78 percentage points, a 5.9%-8.7% relative increase — with no measurable effect from urban closures, and the worst effects (11-13% relative increase) falling on Medicaid patients and racial minorities. A separate 2024 NBER study of ER closures specifically found no overall effect on county-level all-cause mortality, but did find increased drug-related and heart-attack deaths in some subgroups and regions. The honest picture: distance increases are well-established and consistent across studies; the mortality effect is real but condition-specific and contested in aggregate — it shows up clearly for time-sensitive emergencies, not for mortality broadly.

How many rural hospitals have actually closed

The AskAmerica corpus (CMS Provider of Services termination file, health.cms_pos_termination_history) records CMS certification terminations, but that field mixes true closures with routine change-of-ownership re-certifications (termination code 04) and other administrative actions, so it overstates true closures on its own. The authoritative tracker is the UNC Cecil G. Sheps Center for Health Services Research, which manually verifies each closure against the criterion that the hospital stopped providing acute inpatient care (mergers, conversions to Critical Access Hospital, or conversions to the new Rural Emergency Hospital designation don't count).

As of this report: 197 rural hospital closures/conversions since January 2005 (109 complete closures + 88 converted-to-outpatient-only), of which 154 have occurred since 2010 (86 complete + 68 converted). The pace accelerated sharply from the mid-2010s: a low of 2-3 per year around 2010-2011 rose to a peak of 17 in both 2015 and 2019, dipped during the pandemic (federal relief funds propped up many rural hospitals in 2020-2021), and has run 5-7 per year since 2022. The chart above shows the year-by-year count reconstructed from the Sheps Center's public closure list.

How much farther people have to travel

Multiple independent studies converge on the same rough magnitude:

• Nearest hospital, overall: among Medicare beneficiaries who lost their nearest hospital to closure, median travel distance rose from 13.1 to 16.4 miles (Health Affairs Scholar, 2025 surgical-access study).
• Service-specific distance: a systematic review (Mills et al., Journal of Rural Health, 2024) found the average increase in distance to the specific services a closed hospital had provided was about 20 miles for common services (e.g., inpatient care) and about 40 miles for less-common services (e.g., substance-use treatment).
• Regional detail: in the rural South, mean distance to the nearest and second-nearest hospital were 15.9 and 26.4 miles respectively, and counties with two closures showed measurably reduced use of preventive care among older adults.
• Ambulance/EMS times: closures added roughly 11-16 additional minutes to transport patients to the nearest emergency facility, and increased EMS activation/response time by about 7 minutes in one study.

Whether more people die

The mortality literature is smaller and more mixed than the distance literature, and the honest answer depends on which mortality outcome is asked about:

• Time-sensitive inpatient conditions (the clearest positive finding): Gujral & Basu (NBER Working Paper 26182, California hospital discharge data 1995-2011) found rural hospital closures raised adjusted inpatient mortality for stroke, AMI (heart attack), sepsis, and asthma/COPD combined by 0.46-0.78 percentage points — a 5.9% to 8.7% relative increase — while urban closures showed no measurable effect. Broken out by condition: stroke mortality rose ~3.1% and AMI mortality rose ~4.5%; effects were substantially worse for Medicaid patients (+11.3%) and racial/ethnic minorities (+12.6%), and asthma/COPD mortality actually fell (likely a compositional effect of fewer, sicker patients being admitted locally versus transferred).
• County-level all-cause mortality (the null/contested finding): a 2024 NBER working paper (Chatterji, Ho & Wu, w32189) using a difference-in-differences design on 2005-2018 rural ER closures found no significant effect on overall county-level all-cause mortality across the full sample. It did find closure was associated with increased drug/alcohol/suicide deaths among younger white females and increased AMI mortality specifically in the South and West — i.e., the effect concentrates in subgroups and specific causes of death rather than showing up as a broad population-level mortality rise. A widely cited 2015 Health Affairs study of nearly 200 closures (rural and urban) similarly found no significant change in community-wide hospitalization rates or mortality.

Putting these together: hospital closures reliably increase travel distance and time to care. They reliably worsen outcomes for people who arrive at a hospital with a time-sensitive emergency (measured as inpatient mortality conditional on admission). They do not show up as a clear, broad increase in a whole county's overall death rate in the available studies — likely because the affected population (people who would have used the closed hospital for an emergency) is a small share of a county's total deaths, and because some patients are diverted to farther-but-still-capable hospitals rather than going without care entirely.

Data and methodology notes

This corpus (AskAmerica/askengine) carries CMS facility termination records (health.cms_pos_termination_history, health.cms_pos_facilities) and USDA rural-urban classifications (geo.rural_urban_continuum), which can identify hospital closures in nonmetro counties, but that termination field is not a clean 'closure' flag — it also captures change-of-ownership re-certifications and other non-closure administrative events, so a raw count from it (which we ran and discarded, see below) overstates true closures relative to the Sheps Center's manually verified list. Nor does the corpus carry patient-level travel-distance or mortality outcomes tied to specific closure events; those numbers in this report come entirely from the peer-reviewed and government literature cited above, not from a corpus query. We deliberately did not use the corpus's raw termination count (1,078+882+92+... rows by termination code, several thousand nonmetro hospital terminations since ~1990) as a headline figure because it conflates true closures with ownership changes; the Sheps Center figure (154 since 2010) is the number a rural-health researcher would actually cite.

Every query behind this report

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

SELECT date_part('year', t.termination_date) AS closure_year, count(*) AS closures
FROM health.cms_pos_termination_history t
JOIN health.cms_pos_facilities f ON t.provider_number = f.provider_number
JOIN geo.rural_urban_continuum r ON t.county_fips = r.county_fips
WHERE t.is_terminated = true AND t.provider_category_code = '01'
  AND r.metro_nonmetro = 'Nonmetro'
  AND t.termination_date >= DATE '2010-01-01'
GROUP BY date_part('year', t.termination_date)
ORDER BY closure_year

SELECT termination_code, count(*) AS n
FROM health.cms_pos_termination_history t
JOIN geo.rural_urban_continuum r ON t.county_fips = r.county_fips
WHERE t.is_terminated = true AND t.provider_category_code = '01'
  AND r.metro_nonmetro = 'Nonmetro'
GROUP BY termination_code
ORDER BY n DESC