Presenter: Jonathan Pastrana Del Valle MD | Beth Israel Deaconess Medical Center Background: Distal pancreatectomy is widely employed in the treatment of benign and malignant conditions of the body and tail of the pancreas. Postoperative pancreatic fistula is a frequent and potentially morbid complication after this operation, occurring in up to 20% of distal pancreatectomy patients. Effective methods for predicting and preventing postoperative pancreatic fistula are still sought. While effective risk assessment models have been developed for postoperative pancreatic fistula after pancreaticoduodenectomy, to our knowledge, a similar model for distal pancreatectomy does not exist. The aim of this study was to investigate factors associated with postoperative pancreatic fistula following distal pancreatectomy and to create a risk model to predict this complication using data from the ACS-NSQIP database. Methods: Patients who underwent distal pancreatectomy from 2014 to 2017 were selected from the ACS-NSQIP pancreatectomy targeted database. Clinically relevant (grade B or C) fistulas were identified using the 2016 ISGPS definitions. We utilized chi square tests on categorical variables and logistic regression on continuous variables to assess associations of preoperative and intraoperative variables with postoperative fistula. Variables with a p<0.10 on univariable analysis were then included in a multivariable logistic regression model, and a nomogram was constructed from this model. The predictive ability of the nomogram was assessed using the concordance statistic (c-statistic), and its calibration was tested using the Hosmer-Lemeshow test. Results: A total of 7,088 patients who underwent distal pancreatectomy from 2014 to 2017 were identified in the ACS-NSIQP database. Clinically relevant postoperative pancreatic fistula were identified in 1,174 (16.6%) of the patients. Univariable analysis identified male sex, higher BMI, smoking history, congestive heart failure, disseminated cancer, low pre-operative albumin, chemotherapy within 90 days of surgery, radiation within 90 days of surgery, intra-operative drain placement, vascular resection, and histology were associated with pancreatic fistula (all p<0.05). Gland texture, pancreatic duct size, and presence of chronic pancreatitis were not significantly associated with fistula. Variables independently associated with pancreatic fistula in the multivariable model were male sex, higher BMI, smoking history, congestive heart failure, higher preoperative platelet count, intraoperative drain placement, vascular resection, and histology (all p<0.01). The nomogram constructed from this analysis is shown in panel A. The concordance statistic of the nomogram is 0.625 (ROC curve shown in panel B). The nomogram appeared well calibrated (panel C) with a Hosmer-Lemeshow p=0.723 consistent with a good fit. Conclusion: Male sex, higher BMI, smoking history, congestive heart failure, higher pre-operative platelet count, intra-operative drain placement, vascular resection, and histology are independently associated with risk of developing postoperative pancreatic fistula after distal pancreatectomy. A nomogram based on these risk factors yielded a c- statistic of 0.625 suggesting that other variables not accounted for may contribute to the risk of pancreatic fistula and/or there is some inherent unpredictability in the development of this complication. The nomogram does provide a well calibrated prediction of risk based on easily determined pre-operative and intra-operative variables, and thus may be useful in post-operative management. Further study to externally validate this nomogram is warranted.