[Evaluation of the mortality related to nosocomial infections. An approach by a homogeneous group of patients, from bacteriology and data of hospitalization (1989-1993)].
We cross matched the informations from the medical records of the PMSI (French Program of Medicalisation of the Information System) with those from the bacteriology data base. The hospitalisation summaries provide the length of hospital stay, the mortality and the pathology defined by DRG (Diagnosis Related Group). Bacteriological data allow the diagnosis of nosocomial infections (NI) when using an appropriate methodology ("doubles" and early samples are not taken in account). Then it is possible to estimate the mortality when there are NI and No. 4499 stays NI were compared with 140,463 stays without NI. Mortality is 14.4% in NI group (17.5% in medecine and 8.6% in surgery) against 2.1% in the group without NI (2.5% in medecine and 0.8% in surgery). Among the 58 DRG's studied (more than 19 cases with NI), mortality and relative risk varie a lot. Results are estimations because they are many bias.