Romania · hospital stays by patient’s home county · 2019–2025
Where people from each Romanian county end up in hospital
In 2025, about 946,000 hospital stays, almost one in four, took place outside the patient’s registered home county. For Giurgiu, Călărași and Ialomița it is more than half.
Hospital beds are spread unequally between counties: from 3.2 to 10.5 per 1,000 residents. The number of hospital stays, relative to the population and its age, differs far less.
- in the county’s hospitals
- in another county
Tap a county to open its details. Bar length: hospital stays of people registered in the county, per 1,000 residents, adjusted for the age of the population. The bar is split by the share of stays in another county, which is not age-adjusted. For Bucharest and Ilfov, beds are counted together (10.5 per 1,000), on the assumption that the two share the same hospitals.
Beds differ widely between counties, hospital stays far less#
Giurgiu has 3.2 hospital beds per 1,000 residents, 44% of the national rate. Bucharest and Ilfov, taken together, have 10.5. The ratio between the extremes is 3.3 to 1.
Hospital stays look different. The measure used here compares the stays of people registered in a county with the number there would be if each age group of the resident population were hospitalised as often as in the whole country. The county with the lowest ratio (Arad) is at 83% and the one with the highest (Hunedoara) at 127%. In 30 of the 41 units the ratio is between 90% and 110%. With the population by registered domicile as denominator the range is almost the same: 84% to 126%.
For Giurgiu the ratio is 93%, and 65.5% of the stays took place in another county. Hunedoara is at the other end: 9.2 beds per 1,000 and a ratio of 127%. These data do not show why.
The two observations stand side by side, and neither explains the other: beds differ far more than stays do, and a large share of stays take place outside the home county. In a calculation made afterwards, outside the analysis plan, the stay ratio and the share of stays in another county are not correlated across counties (coefficient 0.05).
For which care people end up in another county#
The state’s tables count stays by case group twice: by the patient’s registered home county and by the hospital’s county. Where people from a county have more stays than its hospitals reported, the difference is the minimum number of stays that took place elsewhere. The true number can be higher, because the county’s hospitals also treat patients from outside.
For open-heart surgery, hospitals in 28 counties reported no case in 2025. For spinal fusion, 16 counties. For knee replacement, eight. “No case reported” refers to these case groups and does not prove that the service is absent altogether.
For births, strokes and other hip and femur operations the computed minimum is small: 13%, 9% and 9%. The minimum does not say how many stayed in the county.
Rare operations are done in few centres, so some of this travel is expected. The table below and each county’s details show for which case groups it happens.
Case groups: the minimum outside the county
| Case group | Stays in the country | Minimum outside the county | Counties whose hospitals reported no case |
|---|---|---|---|
| Spinal fusions | 8,072 | 62% | 16 |
| Open-heart surgery | 5,872 | 61% | 28 |
| Craniotomies | 8,135 | 48% | 5 |
| Liver and pancreas surgery | 9,981 | 46% | 0 |
| Acute leukaemia | 9,290 | 44% | 1 |
| Major breast-cancer surgery | 6,904 | 43% | 0 |
| Knee replacements | 11,323 | 42% | 8 |
| Retina procedures | 14,008 | 37% | 1 |
| Major chest surgery | 9,796 | 34% | 2 |
| Newborns under 1,500 g | 1,599 | 33% | 1 |
| Tonsil and adenoid operations | 10,230 | 33% | 0 |
| Severe burns or burns with skin graft | 1,290 | 31% | 5 |
| Lymphoma and non-acute leukaemia | 40,884 | 30% | 0 |
| Hip replacements | 15,695 | 28% | 1 |
| Major bowel and rectal surgery | 30,556 | 26% | 0 |
| Tracheostomy or ventilation over 95 hours | 25,946 | 21% | 0 |
| Births | 127,142 | 13% | 0 |
| Major amputations for circulatory disease | 5,170 | 13% | 0 |
| Stroke | 85,318 | 9% | 0 |
| Other hip and femur operations | 12,430 | 9% | 0 |
Hospitals that receive patients from other counties#
Bucharest hospitals had about 448,000 stays of patients registered in another county: 57% of their stays and 47% of all out-of-county stays in the country.
Half of all out-of-county stays took place in 37 of the 545 hospitals that reported. The list starts with the institutes and university hospitals of Bucharest, Cluj-Napoca, Craiova, Timișoara and Iași.
Hospitals with the most stays of patients from another county, 2025
| Hospital | Stays | From another county | Share | From another county, 2019 |
|---|---|---|---|---|
| Institutul Clinic Fundeni București | 47,926 | 33,630 | 70% | 37,034 |
| Spitalul Clinic de Urgență Militar Central | 48,017 | 29,185 | 61% | 32,942 |
| Spitalul Universitar de Urgență București | 50,384 | 26,079 | 52% | 25,087 |
| Spitalul Județean Clinic de Urgență Cluj-Napoca | 56,544 | 20,769 | 37% | 21,872 |
| Spitalul Clinic Colentina București | 30,724 | 17,663 | 58% | 18,268 |
| Spitalul Universitar de Urgență “Elias” București | 31,592 | 17,050 | 54% | 17,122 |
| Spitalul Județean Clinic de Urgență Craiova | 54,962 | 15,043 | 27% | 11,699 |
| Spitalul Clinic de Urgență București | 28,023 | 14,356 | 51% | 15,756 |
| Spitalul Județean Clinic de Urgență “Pius Brînzeu” Timișoara | 45,993 | 14,078 | 31% | 14,762 |
| Institutul Oncologic “Prof. Dr. Al. Trestioreanu” București | 17,853 | 13,768 | 77% | 17,175 |
| Spitalul Clinic de Urgență pentru Copii “M. S. Curie” București | 21,374 | 13,572 | 64% | 12,669 |
| Spitalul Județean Clinic de Urgență “Sf. Spiridon” Iași | 40,476 | 13,313 | 33% | 14,776 |
| Spitalul Clinic de Urgență “Sf. Ioan” București | 26,218 | 12,724 | 48% | 13,995 |
| Spitalul Clinic de Urgență “Bagdasar-Arseni” București | 20,718 | 12,584 | 61% | 14,131 |
| Institutul Oncologic “Prof. Dr. I. Chiricuță” Cluj-Napoca | 18,387 | 12,393 | 67% | 15,923 |
| Spitalul Clinic de Urgență pentru Copii “Gr. Alexandrescu” București | 20,233 | 12,162 | 60% | 12,470 |
| Institutul de Endocrinologie “C. I. Parhon” București | 18,226 | 11,526 | 63% | 11,968 |
| Spitalul Clinic de Psihiatrie “Dr. Alexandru Obregia” București | 23,795 | 11,364 | 48% | 12,252 |
| Spitalul General CF 2 București | 17,792 | 11,154 | 63% | 12,175 |
| Spitalul Municipal Clinic de Urgență Timișoara | 35,466 | 10,994 | 31% | 12,602 |
| Spitalul Clinic de Urgență “Prof. Dr. Agrippa Ionescu” | 16,973 | 10,993 | 65% | 10,655 |
| Spitalul de Cardiologie Covasna | 14,388 | 10,969 | 76% | 11,687 |
| Sanatoriul Techirghiol | 14,562 | 10,760 | 74% | 10,224 |
| Spitalul Județean Clinic de Urgență Târgu Mureș | 35,765 | 9,950 | 28% | 11,384 |
| Spitalul Clinic de Recuperare Iași | 18,400 | 9,662 | 52% | 8,592 |
Since 2019, stays in the home county fell; stays in another county did not#
In 2019 hospitals reported 4,150,361 stays. In 2025, 3,870,911. Almost all of the fall happened in the home county: from 3.22 to 2.92 million stays, 9% fewer. Stays in another county rose from about 926,000 to about 946,000.
The share of stays in another county therefore rose from 22.3% to 24.4%, and it rose in 38 of 42 counties. Without the stays of people registered in Ilfov, 90% of which are outside the county, the share is 22.8% (20.9% in 2019).
Among hospitals reported in both years, the largest increases in stays of out-of-county patients are at Delta Health and Medicover, private hospitals in Bucharest, and at the Craiova County Emergency Hospital. Sanador, also private, is in the top ten as well. Monza Ares appears in the reports only after 2019. The Oradea County Emergency Hospital has almost twice as many stays as in 2019, a sign that the reporting unit changed, so it was not compared.
The largest decreases are at the Central Military Hospital, the two oncology institutes, in Bucharest and Cluj-Napoca, and the Fundeni Clinical Institute.
- in the home county
- in another county
The analysis plan written before the bed data were opened#
The starting hypothesis was different: that people are hospitalised more often where there are more beds. It is a regularity described in other countries. Four hypotheses and their thresholds were written down before the data on beds, age and mortality were opened. Until then only the crude hospitalisation rates by county and the case counts by group had been seen.
In the form set in advance, three of the four failed. Across 41 units, the rank correlation between beds and stays is 0.269 (95% confidence interval −0.07 to 0.55), below the 0.40 threshold. The interval is wide and includes the threshold.
The result depends on how the units are drawn. With Bucharest and Ilfov separate, the coefficient is 0.34. Without the seven units whose hospitals have over 10% more stays than the people registered there, a variant also written in advance, it is 0.40, right at the threshold (interval 0.03 to 0.68). With the population by registered domicile as denominator it drops to 0.11.
Medical stays are not more closely associated with beds than surgical stays are (0.12 against 0.37). For a group of conditions that can often be treated without a hospital (high blood pressure, diabetes, back pain and others), the estimated correlation is close to zero (−0.04), with an equally wide interval.
One hypothesis passed: in a two-variable model, the coefficient for beds is larger than the one for age-adjusted mortality, used as a rough measure of the need for care. The model describes 16% of the differences between counties, and the difference between the two coefficients is itself uncertain.
The result does not show that beds do not matter. It shows that, at county level, for inpatient stays only and with 41 points, the estimates do not reach the thresholds set and remain uncertain.
The four hypotheses, as written in advance
- failedAge-adjusted stays rise with the number of beds (rank correlation of at least 0.40)0.269; interval −0.07 … 0.55
- passedIn a model with beds and mortality, the coefficient for beds is larger and its interval excludes zerobeds 0.40, mortality 0.27; the model describes 16% of the variation
- failedMedical stays vary more between counties and are more closely associated with beds than surgical staysvariation 13.1% against 11.3%; correlation 0.12 against 0.37
- failedStays for a group of conditions often treatable without hospital are associated with beds (at least 0.40); those in a comparison group are notgroup treatable without hospital −0.04; comparison group 0.23
What these data do not show#
Stays, not people. A patient admitted three times appears three times. A patient transferred between two hospitals appears twice.
The address on the ID card, not where the person lives. A student registered in Sălaj and admitted in Cluj-Napoca appears as a stay in another county. Romania has 21.67 million people registered as domiciled in the country and 19.05 million residents. The rates here divide stays counted by registered domicile by the resident population; the two do not overlap exactly.
No day cases. The tables cover inpatient stays only. A county where many cases moved to day care appears with fewer stays.
Nobody can see where each patient went. The state publishes how many stays the people of a county had and how many of a hospital’s stays come from its own county. The destination by county is not published.
A case group is not an operation. Some groups depend on how a hospital reports. Coronary interventions for heart attack are close to zero for people registered in Dolj, Cluj and Constanța. The group could not be interpreted and was left out.
The journey does not say why. A patient ends up in another county because the service is missing at home, because another hospital is closer to their village, because they were referred or because they chose to. The tables do not tell these apart and say nothing about quality of care.
Reconstructed numbers. Stays in another county are computed from percentages published with two decimals. The national total can differ from the true one by a few dozen stays, which is why it is rounded to the thousand.
Method and sources#
Stays by the patient’s registered home county and by the hospital’s county come from the hospital morbidity tables published by the National Institute for Health Services Management (INMSS) for 2018–2025. For 2018, 2021, 2022 and 2023 the annual table starts on 1 April or 1 July and was completed with the table for the first months. The annual total equals the one in the same institution’s quarterly tables exactly for 2019, 2020, 2024 and 2025 (3,870,911 in 2025); for the completed years it differs by at most 35 stays.
Stays that remained in the county are computed from the quarterly tables: the number of records of the hospitals in each county multiplied by the percentage of cases from the same county. The sum over counties reproduces the published national percentage (75.55%).
Concordance check: hospital births, counted by the mother’s registered home county, were compared with live births by county from the National Institute of Statistics (by the mother’s domicile or usual residence). Nationally, hospitals reported 0.87 births per live birth in the statistics. The correlation across counties is 0.996, and in 39 of 42 counties the ratio is within 10% of the national one. The check shows that the two sources divide the counties the same way; it does not validate each case.
The age adjustment uses five groups, as INMSS publishes them for the whole country: 0–4, 5–17, 18–44, 45–65 and over 65. The population was split at 45–64 and 65 and over; with a 45–65 and 66 and over split the main correlation changes from 0.269 to 0.272. The population is the resident population on 1 July 2025. Beds are inpatient hospital beds, public and private, in 2025. Age-adjusted mortality uses deaths by county from official statistics and the resident population.
The analysis plan, dated 10 November 2026, the deviations from it and the full results are in the project files. The plan was not lodged with an independent registry. The page was not read by a physician or a public-health specialist before publication.
- INMSS, drg.ro: hospital morbidity indicators by diagnosis-related group, by patient’s home county and by hospital county, 2018–2025 (in Romanian)
- INMSS, drg.ro: quarterly hospital activity indicators (share of cases from the same county, number of records), 2019–2025 (in Romanian)
- National Institute of Statistics, TEMPO: POP106A (resident population by age and county), POP108D (population by domicile), POP201A (live births), POP206A and POP206B (deaths)
- National Institute of Statistics, TEMPO: SAN102B (beds in health units, by county)