Titre : | Deciphering Hospital Networks. Using Graph Theory Methods |
Auteurs : | Yuanfei Anny Huang |
Type de document : | Mémoire |
Année de publication : | 2019 |
Description : | 46p. / ann., tabl;, carte |
Langues: | Anglais |
Classement : | MPH19/ (Master EHESP International master of public health - MPH) |
Mots-clés : | Réseau ; Hôpital ; Transfert sanitaire ; France ; Analyse donnée |
Résumé : | Objective: For many patients, an acute hospital stay is followed by being transferred to a rehabilitation hospital. For the healthcare system, the flow of patient transfers represent an important movement and use of resources within the network. In order to examine the organisational determinants of these patient transfers within France, a previous study, FEHAP, was undertaken in 2014, using unweighted exponential random graph modelling (ERGM) techniques. This project aims to extend the original analysis by using the newer weighted ERGM technique, which allows for the consideration of the number of patients transferred along each pathway between hospitals. In doing so, this project will also assess the usefulness of the weighted ERGM technique as applied to analysing inter-hospital patient transfers. Methods: The original data from the FEHAP study was reconstructed into regional networks with and without hospital self-transfer loops. These comprised of a total of 54,889 inter-hospital patient transfers, across the regions of Bretagne, Lorraine and Rhône-Alpes, in the year 2012. These networks were analysed firstly using unweighted ERGM techniques, and then using weighted ERGM techniques. Results: Results obtained through weighted ERGM techniques corroborated with those using unweighted ERGM. Firstly, they showed that the structure of each network was not random. Particularly for Bretagne and Lorraine, the department variable shows the strongest effect for predicting patient transfer, with all of the other variables (legal status, travel time, MCO beds, SSR beds, no MCO beds, no SSR beds, median MCO length of stay) also being statistically significant. As well, the effect of the legal status variable was stronger for networks with loops compared to those without loops, and there is also a relationship between legal status and whether the hospital only offered acute or rehabilitation care. However, five of the weighted ERGM models for Rhône-Alpes, the largest network, failed to converge. Another limitation was that the weighted ERGM algorithm was not able to provide any goodness of fit information. Conclusion: Weighted ERGM appears to be a promising technique for analysing patient transfer data. However, further developments may need to occur before it is used for network simulations. Despite this drawback, this project was able to extend the findings from the 2014 FEHAP study. In particular, it confirmed that the geographic department of a hospital is an important predictor for patient transfers. It also demonstrated that the legal status of a hospital is statistically significant as a predictor, contrary to the original FEHAP study findings, where it was only significant for Rhône-Alpes. However, this effect is diminished for the networks without loops. Finally, this project also demonstrates that a range of mechanisms, possibly including competition, may explain the relationship between the length of stay at acute hospitals and the likelihood of transfer. (R.A.) |
Diplôme : | Master MPH of public health |
Plan de classement simplifié : | Master of Public Health - master international de Santé Public (MPH) |
En ligne : | http://documentation.ehesp.fr/memoires/2019/mph//Yuanfei Anny HUANG.pdf |
Exemplaires (1)
Code-barres | Cote | Support | Localisation | Section | Disponibilité |
---|---|---|---|---|---|
098063 | MPH19/0007 | Mémoire | Rennes | Magasin | Empruntable Disponible |
Documents numériques (1)
MPH/yuanfei URL |