Timely identification of patients with acute pancreatitis who are likely to have a severe disease course is critical. Based on that, many scoring systems have been developed throughout the years. Although many of them are currently in use, none of them has been proven to be ideal. In this study, we aimed to compare the discriminatory power of relatively newer risk scores with the historical ones for predicting in-hospital major adverse events, 30-day mortality and 30-day readmission rate.
Patients who had been admitted due to acute pancreatitis were retrospectively investigated. Five risk scoring systems including HAPS, Ranson, BISAP, Glasgow, and JSS were calculated using the data of the first 24h of admission. Predictive accuracy of each scoring system was calculated using the area under the receiver-operating curve method.
Overall 690 patients were included in the study. In-hospital major adverse events were observed in 139 (20.1%) patients of whom, 19 (2.5%) died during hospitalization. 30-day all-cause mortality and 30-day readmission were observed in 22 (3.2%) and 27 (3.9%) patients respectively. Negative predictive value of each score was markedly higher compared to positive predictive values. Among all, JSS scoring system showed the highest AUC values across all end-points (0.80 for in-hospital major adverse events; 0.94 for in-hospital mortality; 0.91 for 30-day mortality). However, all five scoring systems failed to predict 30-day readmission.
JSS was the best classifier among all five risk scoring systems particularly owing to its high sensitivity and negative predictive value