[Hospital deaths attributable to nosocomial infections: surveillance in a university hospital]

Abstract

Objective: A system for monitoring deaths was set up at the Besançon University Hospital, in the Franche-Comté region of France, in January 2002. This article describes the principal results after 1 year of activity.

Materials and methods: The main outcome was the descriptive surveillance of deaths due to nosocomial infections (NI), by means of a two-step process, in the identification of cases. The first step was the declaration of information concerning the possible existence of an NI at the time of death and the assignment of a prognostic score to each patient upon admission. The second step involved comparison of opinions concerning cases "selected" on the basis of clinical records. This system was completed by a methodological analysis of deaths considered to be partially or totally due to NIs, with the aim to suggest preventive actions.

Results: The overall rate of mortality in our hospital was 2.44 per 100 patients admitted. The overall proportion of deaths linked to hospital-acquired infections was 15.3%. We analysed eight of the 20 deaths possibly due to NI (MacCabe score of 1 or 0) in detail. In three of these eight cases, we showed that the cause of death was not NI. Methodical analysis of patient records showed that corrective actions could have been taken in three of these five cases.

Conclusion: Our study suggests that it is possible to monitor deaths due to nosocomial infections continuously in a university hospital and favours re-implementing death analysis committees acknowledging the "useful error" concept which is culturally difficult to accept.