Differential risk of nosocomial infection.

Our ability to interpret apparent differences in findings among surveys of nosocomial infection is limited by differences in the methods used, and differences among the study populations themselves. Methodologic areas in which there are substantial differences include definitions of rates of infection, criteria for infection and methods of case-finding. The studies themselves differ with respect to the characteristics of hospital populations, their underlying diseases, clinical procedures, patterns of lengths of hospital stay and efforts at prevention of infection. There are three separate concerns for constructing and interpreting comparisons: the choice of comparison or reference subjects, the control of confounding variables and the adjustment for variations in hospital stay. In order for studies to be credible, they must be accompanied by convincing evidence that the confounding effects of variables extraneous to the study have been identified and appropriately controlled in the analysis. It has been found that the day-specific incidence of nosocomial infection rises from near zero on the first hospital day to maximal during the fourth through the seventh weeks of hospital stay. Thus, the risk of nosocomial infection for a specific patient also depends on the hospital day. It may be possible to make better use of limited infection control resources by concentrating efforts to prevent infections on the patients who are at greater risk, during the time when the day-specific incidence is highest.