RESUMO
The objective of this study was to develop a method for analyzing differences in the performance of hospitals with respect to outcome by separating patient factors from procedural factors. The setting included a prospective follow-up of a sample of 5571 patients undergoing all types of surgical procedures in general surgery departments of 11 hospitals (20 surgical departments) across Israel. Of these, 769 underwent surgery involving the opening of the bowel, and they are the subjects of this report. Our method consisted of a prospective follow-up by a nurse epidemiologist, including detailed clinical data from the day of admission to hospital discharge. Analysis was directed at identifying reasons for the observed variability in wound infections among departments. Observed rates were compared with "expected" rates calculated from a logistic model pooled over departments. An attempt was made to separate patient-inherent characteristics, such as age, sex, and diagnosis, from procedural factors, depicting the patient's experience during his hospitalization. Results indicated that the marked interdepartmental differences in the observed infection rates were not accounted for by differences in the "case mix" among departments. Procedural risk factors in this data set played the main role in explaining the observed variability among surgical departments. We conclude that the simple method presented here used the data pooled over departments to define the main risk determinants for infection in this data set. It separated intrinsic patient factors from procedural characteristics, and could be used in studies where the main interest is to compare institutions, and point at reasons behind the differences in outcomes.