IDENTIFYING PATIENTS AT HIGH RISK OF SURGICAL WOUND INFECTION

TLDR

The study develops a simple multivariate risk index to predict surgical wound infection likelihood using data from 58,498 1970 surgical patients. Using stepwise logistic regression on 10 risk factors, the authors built a model based on four factors and validated it on an independent cohort of 59,352 patients. The simplified index identifies half of patients who account for 90% of infections and predicts risk twice as accurately as traditional wound contamination classification, potentially improving surveillance efficiency.

Abstract

To predict the likelihood that a patient will develop a surgical wound infection from several risk factors, the authors used information collected on 58,498 patients undergoing operations in 1970 to develop a simple multivariate risk index. Analyzing 10 risk factors with stepwise multiple logistic regression tech niques, they developed a model combining information on four of the risk factors to predict a patient's probability of getting a surgical wound infection. Then, with information collected on another sample of 59,352 surgical patients admitted in 1975–1976, the validity of this index as a predictor of surgical wound infection risk was verified. With the simplified index, a subgroup, consisting of half the surgical patients, can be identified in whom 90% of the surgical wound infections will develop. By the inclusion of factors measuring therisk due to the patient's susceptibility as well as that due to the level of wound contamination, the simplified index predicts surgical wound infection risk about twice as well as the traditional classification of wound contamination (Goodman-Kruskal G=0.67 vs. 0.36, p<0.0001). Use of this new index might substantially increase the efficiency of routine surgical wound infection surveillance and control.