The map
Recorded deaths in your commune or town
Search a commune, town or municipality. The card shows the standardized mortality ratio (SMR): deaths observed in 2014–2022 divided by deaths expected from age and urban/rural classification, using registered-domicile population. A value of 1.00 equals the statistical reference. The second button shows distance between geographic centres towards a unit with beds in 2024, without hospital addresses or travel times.
Units with fewer than 300 people registered by domicile, or fewer than 10 expected deaths over nine years, stay in the "insufficient sample" class: at small populations, a handful more or fewer deaths in one year swings the rate wildly without demonstrating a change in risk. On small screens, pinch to zoom or use the corner buttons.
Comparing distance groups
How mortality differs between distance groups
The 3,177 administrative units with at least 300 people registered by domicile, ten expected deaths and known distance are split into seven groups. Each ratio divides the group’s total observed deaths by its total expected deaths, rather than averaging individual ratios. Distances are between geographic centres towards units reporting beds in 2024. Four of the 3,181 units on the map are excluded from this comparison.
The ratio is 0.99 in units with beds in 2024, followed by 0.96, 0.99 and 1.01 in groups up to 15 km between centres; it reaches 1.03 at 15–20 km, 1.04 at 20–30 km and 1.11 beyond 30 km. The last group’s ratio is about 10% above the 10–15 km group, using unrounded values. The correlation between distance and SMR in units without beds is weak (r=0.15). These groups do not demonstrate that a person’s risk rises after a particular kilometre.
The 87 units beyond 30 km contain 244,255 people registered by domicile on 1 January 2023, out of an analysed total of 21,952,692 (1.1%). This does not estimate how many people actually live over 30 km from a hospital address. The units cluster in the Danube Delta and a few border areas.
What the comparison adjusts
Different rates in rural and urban areas
Reference rates are calculated separately by age group for rural and urban areas. Each unit is therefore compared with the reference for its own classification. The denominator is registered-domicile population; these rates do not separately measure healthcare access.
| Age group | Urban (‰) | Rural (‰) | Difference |
|---|---|---|---|
| 0-4 | 1.16 | 1.84 | +59% |
| 40-44 | 1.70 | 2.71 | +59% |
| 65-69 | 22.07 | 26.39 | +20% |
| 85 and over | 168.87 | 186.71 | +11% |
The table describes differences between recorded rates without attributing them to a cause. Standardization accounts for age and rural or urban classification; it does not remove differences in income, occupation, migration or effective access to services.
Reported beds and higher mortality
Towns with beds in 2024
These ten towns report continuous-stay beds in 2024 and have standardized mortality in 2014–2022 that is 25–55% above the calculated reference. The confidence interval excludes 1.00. Petrila, Lupeni, Cavnic, Rovinari and Nucet have a mining history, but this analysis does not measure the effect of industrial decline on deaths.
- 1Hunedoara1.542,933 deaths observed vs. 1,906 expected
- 2Maramureș1.39598 deaths observed vs. 431 expected
- 3Hunedoara1.393,004 deaths observed vs. 2,169 expected
- 4Dolj1.361,052 deaths observed vs. 776 expected
- 5Gorj1.35656 deaths observed vs. 487 expected
- 6Tulcea1.29597 deaths observed vs. 462 expected
- 7Arad1.281,052 deaths observed vs. 821 expected
- 8Bihor1.28257 deaths observed vs. 201 expected
- 9Dolj1.282,336 deaths observed vs. 1,826 expected
- 10Arad1.271,211 deaths observed vs. 949 expected
Distance is zero by construction because the administrative unit reports beds in 2024. This does not place every person beside a hospital or establish that the same hospital operated throughout 2014–2022. Each case has a 95% confidence interval excluding 1.00. Click a name for its profile.
Distances between centres
More distant units with higher mortality
These ten units have over 20 km between their geographic centre and the centre of the nearest unit with beds in 2024, and a mortality ratio above 1.00 with a confidence interval excluding 1.00. The list describes the recorded association; it does not establish the cause of deaths.
- 1Satu Mare1.5222.1 km · 738 deaths vs. 486 expected
- 2Tulcea1.4960.4 km · 171 deaths vs. 115 expected
- 3Tulcea1.4730.2 km · 1,112 deaths vs. 758 expected
- 4Satu Mare1.4626.0 km · 391 deaths vs. 267 expected
- 5Satu Mare1.4630.3 km · 389 deaths vs. 267 expected
- 6Satu Mare1.4325.6 km · 184 deaths vs. 128 expected
- 7Tulcea1.3957.2 km · 544 deaths vs. 392 expected
- 8Tulcea1.3935.4 km · 371 deaths vs. 268 expected
- 9Constanța1.3739.6 km · 271 deaths vs. 198 expected
- 10Arad1.3323.9 km · 1,070 deaths vs. 804 expected
Each has a 95% confidence interval that excludes 1.00 and at least 10 expected deaths over the nine years. Click a name and the map moves there.
What we left out
A question left open
A group of units in Vâlcea and Teleorman, at 20–25 km between centres in this model, has a very low mortality ratio with a confidence interval excluding 1.00. We do not present it as a list of safer places. One unverified hypothesis is that its denominator includes people whose domicile is registered there but who actually live elsewhere. These data cannot establish the explanation.
Method and sources
Where the numbers come from
Deaths by administrative unit and year come from INS TEMPO POP206D for 2014–2022. They are allocated by the deceased person’s last domicile or usual residence, rather than the place of death. Reference rates use POP206A deaths and POP107D population in five-year age groups, separately for urban and rural areas.
POP107D measures population registered by domicile on 1 January: Romanian citizens with a registered main address in Romania. The INS definition disregards usual residence and the duration or reason for absence. Profiles and population shares use totals on 1 January 2023; expected deaths use annual populations in 2014–2022. These are not counts of people actually living there. The difference from the domicile/usual-residence basis of the deaths can affect SMR, including through migration; its size is not estimated here.
SMR divides observed deaths by the sum of expected deaths: annual registered-domicile population in each age group multiplied by the national rate for its urban or rural classification. Standardization adjusts for age and urban/rural classification, without controlling income, occupation, roads or ambulance provision. Ratios are withheld below 300 registered people or ten expected deaths. Rankings require a 95% Byar confidence interval excluding 1.00; statistical precision does not correct the population or distance limitations.
SAN102C records continuous-stay hospital beds, public and private combined, in 2024. The 289 administrative units with at least one bed have zero distance by definition. For other units, the haversine calculation links the geographic centre of their polygon to the nearest centre of a unit with beds. It uses neither hospital addresses nor homes, roads, emergency access or travel time. The boundaries of the 3,181 units and their centres come from geo-spatial.org, based on ANCPI, version 26 March 2025, CC BY-SA 4.0.
The analysis retains the extraction of 9 September 2026 and the period 2014–2022; it does not present the latest available series. The 2024 network postdates the deaths. Using it as a reference for earlier years assumes stability that has not been demonstrated: we have not reconstructed annual hospital openings, closures or capacity. Network changes can alter group assignments.
This is an ecological comparison between administrative units describing an association. It identifies neither a causal effect, an individual risk nor deaths that could have been avoided. It cannot establish whether a particular person is over 30 km from care. Definitions and interpretation were editorially checked on 17 October 2026; review by 17 October 2027 or when data change. Marius Comper is responsible for publication.