A MULTIVARIATE-ANALYSIS OF FACTORS CONTRIBUTING TO LEAKAGE OF INTESTINAL ANASTOMOSES

BACKGROUND: The study of anastomotic leaks is critically important to surgeons because morbidity and mortality increase many fold in the aft ermath of an anastomotic disruption. Previous studies that have attemp ted to identify significant factors contributing to leakage of intesti nal anastomoses used animal models or have analyzed retrospective data using univariate analysis. Our objective was to identify factors cont ributing to leakage of intestinal anastomoses. STUDY DESIGN: We conduc ted a retrospective, multivariate analysis of 764 patients who underwe nt 813 intestinal anastomoses. RESULTS: The overall rate of anastomoti c leakage was 3.4 percent. No difference was found in rates of leakage among different techniques of anastomosis or among different anastomo tic locations. Colonic anastomoses leaked no more frequently than anas tomoses of the small intestine. Proximal fecal diversion did not decre ase the frequency of leaks. Multivariate analysis identified six signi ficant predictive variables: a serum albumin level of less than 3.0 g/ L, use of corticosteroids, peritonitis, bowel obstruction, chronic obs tructive pulmonary disease, and perioperative transfusion of more than 2 U packed red blood cells. The in-hospital mortality rate in patient s with and without leaks was 39.3 percent and 7 percent, respectively. Multivariate analysis showed that anastomotic leaks: were an independ ent predictor of mortality. CONCLUSIONS: Factors predictive of anastom otic leaks include chronic obstructive pulmonary disease, peritonitis, bowel obstruction, malnutrition, use of corticosteroids, and perioper ative blood transfusion.