Antimicrobial resistant infections : contribution of the AP-HP Clinical Data Warehouse ; La résistance dans les infections bactériennes : apport de l'entrepôt de données de santé de l'AP-HP

The objective of this thesis was to evaluate the contribution of hospital data for the study of antibiotic-resistant bacterial infections, through the example of bacteremia and the clinical data warehouse of the Assistance Publique - Hôpitaux de Paris.A clinico-microbiological database of more than 30,000 patients hospitalized with bacteremia in 14 hospitals with acute care activity between 2016 and 2019 was structured (the BactHub database). An exploratory work evaluated the contribution of a Natural Language Processing program for the identification of primary sites of bacteremia in medical reports, in comparison with ICD-10 coding.The main characteristics of the patients, first episodes and bacteria are described, according to the community-onset or nosocomial origin of the bacteremia. More than half of the patients were male and >60 years old. Mortality rates were significant, ranging from 14-19% in the hospital to 20-26% at D90. Antibiotic resistance rates were significant: 3GC-R K. pneumoniae, 21-37%; 3GC-R E. coli 13-17%; MRSA 11-14%. The comparison of our results with the literature supports the use of the database for research.Then, a study underlines the major impact of the identification of a 3GC-R Klebsiella spp. (OR 4.7), or a 3GC-R E. coli (OR 2.5), on the risk of recurrence of bacteremia at 1 year, in community-onset bacteremia. In contrast, identification of MRSA was not related to recurrence.Finally, key points and challenges related to the study of the link between individual antibiotic exposure and the occurrence of antibiotic-resistant community-acquired bacterial infections are presented. To improve and standardize future studies, a proposal is discussed. In conclusion, enriching the BactHub database with data from the SNDS would greatly increase its potential for research. ; L'objectif de cette thèse était d'évaluer l'apport des données hospitalières pour l'étude des infections bactériennes résistantes aux antibiotiques, à travers l'exemple des bactériémies et de l'Entrepôt de Données de ...