Risk factors and outcomes in postpancreaticoduodenectomy pancreaticocutaneous fistula

Abstract

A significant fraction of patients undergoing pancreaticoduodenectomy develop a postoperative pancreaticocutaneous fistula. To identify risk factors for this complication and to delineate its impact on patient outcomes, we conducted a retrospective review of 1891 patients undergoing pancreaticoduodenectomy between 1981 and 2002. Overall, 216 patients (11.4%) developed a postoperative pancreaticocutaneous fistula. In univariate analysis, gender, coronary disease, diabetes mellitus, operative times, blood loss, radical lymphadenectomy, gland texture, and specimen pathology correlated with fistula rates. In a multivariate model, however, only gland texture and coronary disease were statistically predictive. A soft gland was associated with a 22.6% fistula rate, a 20.4-fold increase in fistula risk over those patients with a medium or firm gland (95% confidence interval, 4.7-90.9). No patient with a firm gland developed a fistula. Although 30-day postoperative mortality was not different between those patients with and those without fistula (1.4% versus 1.5%), the mean length of stay was longer (26.0 days versus 13.2 days) and the rates of certain complications were increased in those patients with fistula. In this single-institution experience, pancreaticocutaneous fistula was most strongly predicted by pancreatic texture. Choice of anastomotic technique did not correlate with fistula rates. Pancreaticocutaneous fistula increases postoperative length of stay and morbidity but was not directly associated with increased postoperative mortality.