Development and validation of an open pancreatic necrosectomy risk score in acute pancreatitis
Recommended Citation
Potti C, Chau L, Ferguson R, Rakitin I, Obeid N, Stanton C. Development and validation of an open pancreatic necrosectomy risk score in acute pancreatitis. Surg Endosc 2023; 37:S535-S536.
Document Type
Conference Proceeding
Publication Date
6-7-2023
Publication Title
Surg Endosc
Abstract
Introduction: There are a paucity of literature describing risk factors for requirement of open pancreatic necrosectomy. This study aims to develop a risk model predictive of progression to open necrosectomy amongst patients with acute pancreatitis in a tertiary center. Methods: Adult patients admitted with acute pancreatitis from 7/1/ 2013 to 7/1/2022 were included. Variables of interest were selected using backward stepwise selection with criteria for entry P = 0.1 and exit P = 0.2. Variables available for selection include patient demographics, cause of pancreatitis, comorbidities, prior 30-day readmission, number of computed tomography (CT) imaging, and serum laboratory values within 72 h of admission. Logistic regression models and corresponding nomogram were fitted based on selected variables to predict requirement for open pancreatic necrosectomy during the same admission. Performance of the model was assessed by computing the area under the receiver operating characteristic curve (AUROC) after tenfold stratified cross-validation. 95% confidence intervals were calculated with 200 bootstrap replications. Results: 3493 admissions with 3022 patients admitted for pancreatitis were included. Most common etiologies of pancreatitis included alcohol (61.6%) and gallstones (29.2%). 3% of the cohort progressed to open pancreatic necrosectomy with 1% requiring repeat operative intervention. The model identified 8 clinical factors predictive of progression to open pancreatic necrosectomy during the same admission: male sex, race, etiology, ICU admission, organ failure on admission, number of prior CTs, presence of pancreatic necrosis on CT, and prior 30-day readmission. The model AUROC was 0.855 (95% C.I. = 0.79-0.92). Conclusion: We demonstrate a risk score using 8 clinical factors that predict progression to open pancreatic necrosectomy during the same admission among patients admitted with acute pancreatitis.
Volume
37
First Page
S535
Last Page
S536
