Predicting successful catheter drainage in patients with severe pancreatic fistula after pancreatoduodenectomy: a multicenter cohort study

Abstract

Introduction: Catheter drainage appears to be superior to relaparotomy in the management of postoperative pancreatic fistula, but is not successful in all patients. The aim of this study is create a prognostic model for successful catheter drainage in patients with pancreatic fistula after pancreatoduodenectomy. Methods: This is multicenter cohort study on patients undergoing catheter drainage as first intervention for pancreatic fistula after pancreatoduodenectomy (January 2005 to September 2013) in 9 Dutch Pancreatic Cancer Group centers. Possible prognostic factors for successful catheter drainage (i.e. survival without relaparotomy) were evaluated using logistic regression and selected using the Akaike Information Criterion. The model was internally validated and definitive predictors were combined in a nomogram. Results: Of 2196 patients undergoing pancreatoduodenectomy, catheter drainage was the first intervention for pancreatic fistula in 227 patients. Primary catheter drainage was successful in 175/227 patients (77%). Multivariable logistic regression revealed the following negative prognostic factors for success: male sex (odds ratio [OR] 0.46, 95% confidence interval [CI] 0.21-1.00, P=0.049), higher age (for every 5 years over 50; OR0.69 95%CI0.57-0.84; P< 0.001) and respiratory failure in 24h before catheter drainage (OR0.10, 95%CI0.03-0.33, P< 0.001). A prognostic model incorporating these factors yielded an c-statistic of 0.76. The prognostic nomogram demonstrated a success range from 98-14%. Conclusions: Male sex, higher age and respiratory failure are negative prognostic factors for successful catheter drainage in patients with pancreatic fistula after pancreatoduodenectomy. Future prospective trials should determine whether early detection and management of pancreatic fistula can prevent respiratory failure and ultimately improve clinical outcome.