Rural hospital closures add roughly 20 miles to the drive for care — and modestly raise mortality only for time-sensitive conditions in closure-era studies with the statistical power to detect it

Summary

Rural hospital closures reliably push people farther from care: the U.S. Government Accountability Office found that residents of the 64 rural communities that lost a hospital between 2013 and 2017 had to travel a median of about 20 miles farther for common inpatient and emergency care (roughly 3.3-3.4 miles before closure to 23.9-24.2 miles after), and up to 39 miles farther for less common services such as substance-abuse treatment. Whether closures increase deaths is less settled: the best-known peer-reviewed estimate (Gujral & Basu, NBER, using 1995-2011 California closures) reports a real but modest rise — inpatient mortality for time-sensitive conditions (heart attack, stroke, sepsis, COPD) rose about 0.46 percentage points, a 5.9% relative increase — concentrated in rural closures with no comparable effect from urban ones. A more recent national Medicare study covering a different period (Hoffman et al., 32 closures 2014-2018) reports no significant increase in 30-day mortality, because hospitalizations and average length of stay rose instead — evidence that most displaced patients simply got care somewhere else, sometimes at a higher-quality facility. Every number below is read directly from the cited published studies; no statistical model was re-run on primary microdata in this session. The honest synthesis: the distance increase is large and consistent across every study that measured it; the mortality effect is real on average in the most-cited estimate but small, era-specific, and not found in every rigorous study — it shows up most clearly for time-critical conditions and for the most isolated communities and disadvantaged populations, and can be masked in aggregate data because most patients successfully reroute.

How much farther do people have to go?

The clearest, most authoritative number is from the U.S. Government Accountability Office's 2020 study (GAO-21-93) of the 64 rural hospitals that closed from 2013 to 2017. It compared the median distance affected residents had to travel to the nearest remaining provider before and after each closure:

• Inpatient care: median distance rose from 3.4 to 23.9 miles
• Emergency department care: median distance rose from 3.3 to 24.2 miles
• Less-common services (e.g., substance-abuse treatment): the median additional distance was 39.1 miles
• Physician supply in affected counties also fell, from 71.2 to 59.7 physicians per 100,000 residents (2012 to 2018)

Other studies corroborate the ~20-mile order of magnitude and add texture:

• The Center for American Progress found the average distance from a financially fragile ("low-margin") rural hospital to the next-closest emergency department was already 22 miles, with 55% of these hospitals more than 20 miles from the next ED and one in ten more than 35 miles away — meaning closure of these hospitals specifically would push many communities into much longer drives.
• A 2026 Texas study (Zheng, Horel & Blackburn, Inquiry) used road-network mapping of hospital closures from 2012-2023 and found that among census tracts whose nearest ED changed after a closure, drive time to the nearest ED rose by an average of 13 minutes, and rural/small-town tracts had 2.7-3.4 times higher odds of experiencing a travel-time increase at all than metropolitan tracts (baseline drive time in rural tracts already averaged 23 minutes to the nearest ED versus 10 minutes in metro tracts).
• EMS-specific data (Nikpay et al. 2021, national) found total EMS ground miles driven rose 16% and miles per trip rose 19% in counties that experienced a rural closure, with no change in the number of trips — meaning ambulances are covering the same call volume over much longer routes. A separate national study (Miller et al. 2020) found EMS transport time rose 2.6 minutes and total activation time rose 7.2 minutes post-closure, though with highly variable effects across sites.

Do more people die?

Here the literature is genuinely mixed, and a scoping review (Mullens et al., Journal of Rural Health, 2024, screening 5,054 citations down to 20 original empirical studies) explicitly flags "significant heterogeneity" in mortality findings as a key evidence gap.

Evidence for a real mortality increase: The most-cited estimate is Gujral & Basu (NBER Working Paper 26182, 2019). The authors compare California hospital service areas with and without a closure between 1995 and 2011, before and after each closure event, netting out area- and time-fixed differences. For rural closures, they report inpatient mortality for four time-sensitive conditions (AMI, stroke, sepsis, COPD/asthma) rising 0.46 percentage points — a 5.9% relative increase — while urban closures showed no mortality effect at all in their published results. This is consistent with the mechanism that time-sensitive emergencies are exactly where the added travel distance documented above should matter most; other work summarized in the scoping review reports similar increases concentrated among minoritized populations specifically.

Evidence against a detectable increase: Hoffman et al. (Health Services Research, 2024/2025) analyzed 100% Medicare claims for 32 closures from 2014-2018 — a more recent period, after Medicaid expansion reshaped which hospitals were closing — and report no significant change in 30-day mortality, no significant change in average travel distance to the hospital ultimately used, and only a small increase in readmissions concentrated among the most geographically isolated closures. The authors' own interpretation: closures triggered a substitution effect — hospitalization rates and average length of stay both rose after closure in their data, suggesting displaced patients sought care at other, sometimes higher-quality, hospitals rather than going without care or facing a higher chance of death. Other studies reviewed in the scoping review similarly report no effect on hospitalization or mortality rates, and some evidence that consolidation can even improve outcomes when the alternative hospital patients end up at is materially better.

Reconciling the two: the effect looks real, small, and conditional rather than universal. It is most detectable (a) in time-sensitive, life-threatening conditions where minutes matter, (b) among the most isolated communities where no reasonable alternative hospital exists nearby, and (c) among disadvantaged or minoritized populations less able to reroute to a better-resourced facility. It is least detectable in aggregate all-cause mortality studies of recent closures, because most Medicare beneficiaries successfully substitute to another hospital — often, on average, one that is no worse and sometimes measurably better.

Data and methodology notes

All figures in this report are drawn from the primary or near-primary published sources cited below, read directly via web_fetch — not from search-engine summaries alone, with one exception noted below. No AskAmerica regression or causal-inference tool was invoked against primary microdata for this question; the mortality and distance estimates above are the published studies' own reported results, taken as given rather than reproduced here.

One figure that surfaced only in a search-engine synthesis ("a one-minute increase in EMS response time raises mortality by 8-17%," and "average distance from a closed rural hospital to the next-closest hospital is 12 miles") could not be traced to a specific primary paper in the time available and is deliberately omitted from the dashboard rather than presented as verified.

This connector's own health schema carries data that could extend this analysis to more recent (2018-2026) closures — health.cms_pos_termination_history (CMS Provider of Services termination records) and health.hospital_with_geography (facility-level location plus USDA rural-urban classification) — but a query against cms_pos_termination_history returned a backend error (S3 key not found) at the time of this analysis, so no original AskAmerica-computed closure count or distance figure could be added; this report is a literature synthesis, not a warehouse computation, and that gap should be logged as a data-availability defect for that table.