Prevalence surveys of nosocomial infections using a random sampling method in Marseille hospitals

In 1992, we conducted two prevalence surveys on hospital-acquired infections (HAIs) in a group of eight university affiliated hospitals with a total of 4462 acute care beds. Most of the intensive care units (ICUs) already had a prospective surveillance scheme. The need for HAI rates that could be estimated easily and economically led us to develop a prevalence survey by using a stratified sampling method. The units were distributed into four groups: ICU; clean surgery; other surgery; and medical. In each group a sample of units was randomly selected using different sampling fractions. The first survey was conducted in May and involved 1220 patients. The second survey in November included 1389 patients. The HAI rates in May and November were 8.6% (95% CI 7.4-9.8%) and 7.1% (95% CI 6-8.2%), respectively. This study allowed us to direct resources towards high risk units and clean surgical procedures. Based on the results, prospective surveillance was implemented in units of the clean surgery group.