The document presents itself as a state of play of the quality of the FINESS directory and as an analysis grid that can be used to evaluate its new version, FINESS+ (p. 2). It indicates which information it considers robust and which requires caution, with a usage recommendation for each (p. 32). It reminds that this data is used by applications, by the CREAI and in the EHPA survey (p. 4). It finally describes how to access the data (annex 1, p. 35-36).
"
1️⃣ SUMMARY
FINESS is the national directory of health, social and medico-social establishments, managed by the Digital Health Agency and accessible at finess.esante.gouv.fr and data.gouv (p. 3). According to the authors, its quality is regularly called into question without having been subject to a systematic evaluation from the perspective of its use in research (p. 2, 6). The document assesses to what extent this data allows for the analysis of the medico-social offer and its transformations between 2004 and 2023, based on the case of establishments for elderly people and for people with disabilities (p. 2). It explores a 2023 "augmented" file of 47 604 structures and the historical data from 2004-2022 (p. 7-15), compares FINESS to the social action directory, to the CollSol-EHPAD survey, to a database from ARS Bretagne and to the ES-Handicap survey (p. 15-23), then illustrates possible treatments (p. 23-31). It concludes with a reliability grid and five conditions of use (p. 32).
2️⃣ WHAT TO REMEMBER
1️⃣ A reliability that varies according to the information (finding). Counting of structures, legal status and care regime are considered robust; the reception capacities are fragile (not reported in 26.4% of cases for the 1st discipline); the opening dates are questionable (774 EHPAD "opened" in 1901) (p. 2, 9, 25, 32).
2️⃣ Breaks in series to be identified (finding). The reduction of 9,691 registered establishments between 2013 and 2014 is first presented as linked "more plausibly" to a change in scope, with a cleaning of the database remaining possible (p. 14-15), then attributed to a change in scope in conclusion (p. 32).
3️⃣ A method presented as transferable (proposal from the authors). Confront the directory with independent sources on reasoned samples: according to the authors, this approach "can" be transposed to other directories, and their grid "could" be applied to FINESS+ (p. 33).
3️⃣ TO ACT
1️⃣ Apply the five conditions of use (recommendations made by the document). Document the treatments, harmonise the nomenclatures before any comparison over time, prioritise the most reliable indicators, identify breaks in series, cross-reference with other sources when precision is an issue (p. 32).
2️⃣ Choose your variables with the reliability grid (existing tool cited by the document). For example: for capacities, cross-reference the sources and prioritise the median; for geolocation, limit spatial analyses to the period after 2012 (p. 32).
3️⃣ Access the data (existing tools cited by the document). Files on data.gouv.fr and extraction on finess.esante.gouv.fr, limited to 3,500 responses per search (annex 1, p. 35-36).
🔄 Update — external source
The page data.gouv.fr of the dataset used by the authors indicates that a new version of FINESS was deployed on 20 July 2026 and that the generation of these flows has stopped; it points to two new datasets (structures, activities) and specifies that the history remains accessible (data.gouv.fr, consulted on 7 October 2026). The document announced FINESS+ "for July 2026" (p. 3, 33); the links in annex 1 should be verified before use.
Suggestion Practices in Health: before taking a figure from FINESS (territorial diagnosis, key figures), note the vintage and the variable used, and refer to the grid on p. 32.
4️⃣ 3 PRACTICAL QUESTIONS
1️⃣ Can we rely on the number of places indicated in FINESS? The document considers it fragile: not reported in 26.4% of cases (1st discipline), discrepancies between sources. It recommends cross-checking, favouring the median and indicating uncertainty (p. 9, 20, 32).
2️⃣ Has the number of EHPADs decreased? According to the data used, the number of active EHPADs "has seen a growth between 2002 and 2014", followed by a relative stagnation; the ratio for 10,000 inhabitants aged 65 and over has decreased from 4.6 in 2004 to 3.6 in 2023 (p. 23-24).
3️⃣ Are the geographical coordinates usable for a map? They have existed since 2007 but are only correctly reported from 2012; the document recommends restricting spatial analyses to this period (p. 12, 32). Example: the ESAT (p. 31).
5️⃣ TO GO FURTHER
Blog Health Practices: « One in three daily caregivers expresses a need for rest, unmet for half of them » (2 October 2026), according to a publication from DREES. Indirect thematic link (Health Practices analysis): it concerns the use of respite services, including temporary accommodation. Link open and verified on 7 October 2026. Research limit: pages 1 to 4 of the blog and search engines explored; no article dedicated to FINESS or the data of the medico-social offer has been identified; the 473 pages of the blog have not been browsed one by one.
Resource external : Digital health agency, dataset « FINESS - FINESS-STR Structures » on data.gouv.fr — flow of the new version of FINESS (2026), updated daily, with monthly and annual snapshots and a documented schema on GitHub. Complementary: this is the version that the document proposes to evaluate with its grid (p. 33). Page verified on 7 October 2026; exact creation date of the dataset not verified.
6️⃣ HASHTAGS
#healthpractices #MédicoSocial #FINESS #AdministrativeData #OpenData #EHPAD #Disability #TerritorialDiagnosis
=========================================================
To go further, order a research or documentary watch
This article : An editorial production based on
augmented experience by artificial intelligence - The English version of the site is an automatic translation: the French version is authoritative - Click on the image 