Day-specific incidence of nosocomial infection estimated from a prevalence survey

The authors present a method for estimating hospital day-specific incidence of nosocomial infection from a prevalence survey. At Boston City Hospital this incidence was found to rise from zero on the first hospital day to become maximal during the fourth through the seventh weeks of hospital stay, and then to fall gradually. The risk in the second week of hospitalization was over four times that in the first week. The pattern of an initial rise and subsequent fall in day-specific incidence with increasing hospital stays was found to be more marked in patients with discharge diagnoses associated with greater susceptibility to infection and was similar among patients with both short and long total durations of hospital stay. As day-specific incidence varies by hospital day, and proportion of admissions remaining in hospital to any specified day varies by hospital population, both these sources of variation must be taken into account to interpret comparisons among hospital populations.