Objectives: Currently, there is no scoring system for predicting severity in acute pancreatitis in children. Our intent was to evaluate the performance of existing scoring systems in children, to develop a system for children, and to examine the etiology of acute pancreatitis in children.
Methods: A chart review of children with acute pancreatitis was conducted at six centers, three serving as criterion centers and three as validation centers. Ranson and Glasgow scores were calculated for each admission. Additional clinical data were collected, and parameters correlating with severity were incorporated into a new scoring system. Performance characteristics were calculated for each system.
Results: A total of 301 admissions were reviewed, 202 in the criterion group and 99 in the validation group. Eight parameters were included in a new scoring system for children. The parameters were as follows: age (<7 yr), weight (<23 kg), admission WBC (>18,500), admission LDH (>2,000), 48-h trough Ca2+ (<8.3 mg/dl), 48-h trough albumin (<2.6 g/dl), 48-h fluid sequestration (>75 ml/ kg/48 h), and 48-h rise in BUN (>5 mg/dl). When the cut-off for predicting a severe outcome was set at 3 criteria, the new system had better sensitivity versus Ranson and Glasgow scores (70% vs 30% and 35%, respectively) and a better negative predictive value (91% vs 85% and 85%). The specificity (79% vs 94% and 94%) and positive predictive value (45% vs 57% and 61%) fell slightly.
Conclusion: The new scoring system performs better in this group than do existing systems.