Mo1334 Validation and Comparison of the New Severity Classification Systems With Old Atlanta Classification for Severity of Acute Pancreatitis

Abstract

Background Two new classification systems for the severity of acute pancreatitis (AP) have been proposed recently, the determinant based classification (DBC) and revised Atlanta classification (RAC). We aimed to validate and compare these classification systems with original Atlanta classification (OAC). Aims To validate and compare the DBC and RAC with original Atlanta classification (OAC) Methods 469 adult patients with AP admitted to a tertiary care center from January 2009-June 2013 were included in the study. The new classification systems were validated and compared in terms of outcomes (need for interventions, total hospital and intensive care unit (ICU) stay and mortality). Results The mean age of patients was 39.9±13.4 years (331 males) with the commonest etiology being alcohol (161, 34.3%) followed by gall stones (125, 26.6%). There were 119 (25.4%) patients with mild and 250 (74.6%) patients with severe AP as per OAC. Pancreatic necrosis was present in 66.1% and infected pancreatic necrosis in 23.1% patients. 126 (26.9%) patients underwent interventions (endoscopic n= 49, 10.4%, radiological n=95, 20.2% and surgical n=47, 10%). 93 (19.8%) patients died. As per DBC, 97(20.7%), 172 (36.7%), 152 (32.4%), and 48(10.2%) patients were determined to have mild, moderate, severe, and critical AP, respectively. As per RAC, 119 (25.4%), 160 (34.1%), and 190 patients (40.3%) were determined to have mild, moderately severe, and severe AP, respectively. Higher grades of severity were associated with worse outcomes in DBC, RAC and OAC. Predictive accuracies were evaluated using area under the receiver operator characteristics curve (AUROC) and Somer's D co-efficient. The DBC, RAC and OAC were comparable in predicting the need for interventions (AUROC 0.53, 0.55, 0.54, p=0.36) and length of hospital stay (Somer's D, 0.27, 0.26, 0.23, p=0.41). However, both DBC and RAC had comparable but better accuracy than OAC in predicting need for ICU admission (AUROC 0.73 for both vs. 0.62 for OAC, P<0.001), length of ICU stay (Somer's D, 0.35 for both vs. 0.24 for OAC, p<0.001) and mortality (AUROC 0.78 for both vs. 0.61 for OAC, p<0.001). Conclusion Determinant based classification and revised Atlanta classification categorize patients into subgroups that reflect clinical outcomes. Both have comparable and higher predictive accuracy than old Atlanta classification for need for ICU admission, length of ICU stay and mortality.