Severe Acute Pancreatitis Prediction: A Model Derived From a Prospective Registry Cohort

作者：Juan Carlos Barrera Gutierrez, Ian Greenburg, Jimmy Shah, Priyanka Acharya, Mingyang Cui, Elaina Vivian, Brad Sellers, Prashant Kedia, Paul R. Tarnasky · 发表于：Cureus · 年份：2023 · DOI：10.7759/cureus.46809 · 被引用次数：7 · 研究领域：Pancreatitis Pathology and Treatment、Pancreatic and Hepatic Oncology Research、Appendicitis Diagnosis and Management

Background Severe acute pancreatitis (SAP) has a mortality rate as high as 40%. Early identification of SAP is required to appropriately triage and direct initial therapies. The purpose of this study was to develop a prognostic model that identifies patients at risk for developing SAP of patients managed according to a guideline-based standardized early medical management (EMM) protocol. Methods This single-center study included all patients diagnosed with acute pancreatitis (AP) and managed with the EMM protocol Methodist Acute Pancreatitis Protocol (MAPP) between April 2017 and September 2022. Classification and regression tree (CART®; Professional Extended Edition, version 8.0; Salford Systems, San Diego, CA), univariate, and logistic regression analyses were performed to develop a scoring system for AP severity prediction. The accuracy of the scoring system was measured by the area under the receiver operating characteristic curve. Results A total of 516 patients with mild (n=436) or moderately severe and severe (n=80) AP were analyzed. CART analysis identified the cutoff values: creatinine (CR) (1.15 mg/dL), white blood cells (WBC) (10.5 × 109/L), procalcitonin (PCT) (0.155 ng/mL), and systemic inflammatory response system (SIRS). The prediction model was built with a multivariable logistic regression analysis, which identified CR, WBC, PCT, and SIRS as the main predictors of severity. When CR and only one other predictor value (WBC, PCT, or SIRS) met thresholds, then th...