EVALUATION OF NOSOCOMIAL INFECTIONS FROM BACTERIOLOGICAL DATA AND DRG

We cross matched the informations from the medical records of the PMSI (French Program of Medicalisation of the Information System) with tho se from the bacteriology data base. The medical record provides the le ngth of hospital stay, the mortality and the pathology characterized b y the DRG. Bacteriological data allow the diagnosis of nosocomial infe ctions (NI) when using an appropriate methodology (''doubles'' and ear ly samples are not taken in account). Then it is possible to estimate the increase of the average length of hospital stay when there is NI. The 2572 records with NI have a duration of 22.5 days against 9.7 for the 67056 records without NI. Medical DRG are compared to chirurgical DRG. For 33 DRG with a suffisant size, the length is multiplicated by a factor which varies between 1,1 to 3,6 (mean=2,3). We study also the frequence of deaths and the date of the first bacteriological sample which allows to estimate the beginning of the NI. We propose a correct ion based on the sample date because natural ''longer'' stays may have more risk to acquire NI.