Comparison of APACHEII, Ranson, BISAP and CTSI scores in early prediction of the severity of acute pancreatitis based on large sample database

Objective To clarify the accuracy of APACHEⅡ, Ranson, BISAP and CTSI scoring systems for predicting the progression of mild acute pancreatitis (MAP) to moderate acute pancreatitis (MSAP) and severe acute pancreatitis (SAP), and death risk of patients with acute pancreatitis (AP). Methods All data from 2080 consecutive adult patients who were admitted within 3 days of disease onset were selected from AP database between 2014 and 2017. The severity was classified according to the revised Atlanta classification systems. Patients who died during hospitalization or discharged automatically were defined as patients at risk of death. The predictive accuracies for MSAP, SAP and death risk were compared using receiver operating characteristic (ROC) curves. Results The 2080 patients with AP were divided into MAP (n=857, 41.2%), MSAP (n=892, 42.9%), and SAP (n=331, 15.9%) according to the revised Atlanta classification system. ROC curve analysis showed APACHEⅡ score, Ranson score, BISAP score and the CT severity index (CTSI) had no predictive value for MSAP, but have predictive value for SAP and death risk. APACHEⅡ score had the highest accuracy in predicting SAP with area under the curve (AUC) values of 0.785 and 0.746 on the 1st and 2nd day after admission, respectively, and the APACHEⅡ score on admission day 1 had the highest accuracy in predicting death risk (AUC=0.845). Conclusions Various scoring systems had predictive value only for SAP and death risk, and APACHEⅡ score had the highese accuracy in predicting SAP and death risk. Key words: Pancreatitis, acute necrotizing; Mortality risk; Prediction; Database