# When distance to the hospital starts to matter

*Marius Comper · 9 September 2026 · [Interactive edition](https://mariuscomper.uk/aproape-de-spital/en/) · [Ediția în română](https://mariuscomper.uk/aproape-de-spital/)*

53.6% of Romanians live in a locality that hosts its own hospital with continuous-stay beds. Only 1.1% — about 244,000 people, across 87 localities — live more than 30 straight-line kilometres from the nearest one. For each of Romania's 3,181 localities, we compared 2014-2022 mortality (adjusted for age and for urban/rural status) with the real distance to the nearest functioning hospital. The result isn't a line that climbs from day one: mortality stays nearly flat out to 15 kilometres, rises gently between 15 and 30, and only becomes clearly higher past 30 kilometres — a threshold, not a slope, concentrated almost entirely in the Danube Delta and a handful of border communes.

- **0.99 → 1.11**: adjusted mortality ratio, from hospital-hosting localities to those beyond 30 km.
- **1.1%** of the population lives more than 30 km from the nearest functioning hospital.
- **15 km**: the distance below which adjusted mortality barely differs from hospital-hosting localities.
- **59%**: how much higher age-specific mortality is in rural areas than urban ones, at age 40-44 — more than the entire distance effect.

## A threshold, not a slope

Splitting localities into seven groups by distance to the nearest hospital and pooling the ratio within each group gives: 0.99 for hospital-hosting localities, 0.96 at 0-5 km, 0.99 at 5-10 km, 1.01 at 10-15 km, 1.03 at 15-20 km, 1.04 at 20-30 km, 1.11 beyond 30 km. The direct correlation between distance and an individual locality's ratio is weak (r=0.15), but the pattern across groups, where small-locality noise cancels out through pooling, is clear: almost nothing changes out to 15 kilometres, then it climbs. Only 87 localities, with 244,255 residents (2023), are more than 30 kilometres from a hospital — almost all in the Danube Delta or in border communes in Satu Mare, Constanța and Mehedinți.

## The urban-rural gap is bigger than the whole gap above

National age-specific rates, computed separately for urban and rural areas, show a bigger gap than the entire distance effect: +59% at ages 0-4 and 40-44, +20% at 65-69, +11% at 85 and over. This gap is already removed from the calculation above through standardization — the 0.99 to 1.11 figures show what's left AFTER the "it's a town" advantage is taken out.

## A hospital in town is not insurance

Several towns with their own hospital have adjusted mortality 25-55% above expectations, with a confidence interval that excludes 1.00. Five are former mining towns — Petrila (1.54) and Lupeni (1.39) in the Jiu Valley, Cavnic (1.39) in Maramureș, Rovinari (1.35) in Oltenia's lignite basin, Nucet (1.28) in Bihor — where industrial decline appears to matter more than proximity to a hospital. The other five (Segarcea, Isaccea, Chișineu-Criș, Băilești, Ineu) are small agricultural or border towns, without as clear a shared explanation.

## Where distance actually shows

At the other end: Livada (Satu Mare, 22.1 km, SMR 1.52), C.A. Rosetti (Tulcea, 60.4 km, SMR 1.49), Babadag (Tulcea, 30.2 km, SMR 1.47), Acâș (Satu Mare, 26.0 km, SMR 1.46) — localities more than 20 km from a hospital, with statistically robust adjusted mortality above expectations, clustered in the Danube Delta and a handful of border communes.

## A question left open

A cluster of localities in Vâlcea and Teleorman, 20-25 km from a hospital, has a very low and statistically "robust" adjusted ratio — but implausibly low next to the country's typical rate. We haven't published it as a list: the most likely explanation is denominator inflation (population registered as resident there, but long gone to work abroad), but we couldn't verify this directly. It remains an open question, not a finding.

## Method, briefly

Deaths (POP206D, 2014-2022) are allocated to the person's last domiciliu, not to the locality where the death occurred. The expected rate comes from an indirect standardization across 18 age groups and urban/rural status (POP206A, POP107D). Distance is a straight line (haversine) between centroids, to the nearest locality with a continuous-stay hospital (SAN102C, 2024). The full sources and their limits are in the interactive edition's methodology section.