Methodologic issues in hospital epidemiology. I. Rates, case-finding, and interpretation

Surveys of nosocomial infection that have been published in the last two decades present great diversity in both purpose and methods. These differences limit our ability to draw generalizations. Difficulties in understanding and comparing past studies derive from diverse sources. Methodologic areas where there are substantial differences included definitions of rates of infection, criteria for infection, and methods of case-finding. Studies also differ with respect to characteristics of hospital populations, their underlying diseases and their patterns of lengths of stay, clinical procedures, and efforts at prevention of infection. All of these features may change over time and vary with geographic setting. Meaningful comparisons can be made only if uniform definitions of rates are adopted, standardized methods of data collection are employed, and procedures are used to minimize problems in interpretation. These problems in interpretation arise from the confounding effects of multiple uncontrolled variables, the limited applicability of intervention trials, unmeasured perturbation in the steady-states of study populations, and subsequent difficulties in identifying causal factors and measuring their quantitative effects.