Surveillance of nosocomial infections by computer analysis of positive culture rates.

We describe a surveillance method that identifies excessive rates of positive cultures based on patient location, culture site, and organism identification. During a 6-month period, this surveillance method, coupled with epidemiological investigations, identified 19 small clusters of cross-infections and three small outbreaks of intravenous catheter-related bacteremias. These infections were associated with apparent breakdowns in proper technique. Concurrent, standard surveillance activities identified only three of these problems. These results demonstrate that computer analysis of positive culture rates is a sensitive and time-efficient method for detecting potentially preventable nosocomial infections.