The PrEDICT-DGE score as a simple preoperative screening tool identifies patients at increased risk for delayed gastric emptying after pancreaticoduodenectomy

Background: Morbidity after Pancreaticoduodenectomy (PD) has remained unchanged over the past decade. Delayed Gastric Emptying (DGE) is a major contributor with significant impact on healthcare-costs, quality of life and, for malignancies, even survival. We sought to develop a scoring system to aid in easy preoperative identification of patients at risk for DGE. Methods: The ACS-NSQIP dataset from 2014 to 2018 was queried for patients undergoing PD with Whipple or pylorus preserving reconstruction. 15,154 patients were analyzed using multivariable logistic regression to identify risk factors for DGE, which were incorporated into a prediction model. Subgroup analysis of patients without SSI or fistula (primary DGE) was performed. Results: We identified 9 factors independently associated with DGE to compile the PrEDICT-DGE score: Procedures (Concurrent adhesiolysis, feeding jejunostomy, vascular reconstruction with vein graft), Elderly (Age>70), Ductal stent (Lack of biliary stent), Invagination (Pancreatic reconstruction technique), COPD, Tobacco use, Disease, systemic (ASA>2), Gender (Male) and Erythrocytes (preoperative RBC-transfusion). PrEDICT-DGE scoring strongly correlated with actual DGE rates (R-2 = 0.95) and predicted patients at low, intermediate, and high risk. Subgroup analysis of patients with primary DGE, retained all predictive factors, except for age>70 (p = 0.07) and ASA(p = 0.30). Conclusion: PrEDICT-DGE scoring accurately identifies patients at high risk for DGE and can help guide perioperative management.