C-reactive protein accurately predicts severity of acute pancreatitis in children

Abstract

Introduction: Predicting severity of acute pancreatitis enables optimization of care, reducing morbidity and length of stay. Modified adult scoring systems have not been able to adequately predict severity in children.

Methods: This was a retrospective study of children presenting with a first episode of acute pancreatitis from 2002 to 2020 in a single tertiary paediatric surgical centre. Serum markers including CRP at 48 h of admission were analysed. Promising biomarkers underwent ROC (Receiver Operating Curve) analysis, and these were compared to the modified Glasgow Pancreas Score. An AUC (Area Under Curve) > 0.90 was taken as an excellent predictor of severity.

Results: Data of 59 children were analysed, median age 13 years. 22 patients (37%) had a severe episode. ROC analysis demonstrated CRP as the best predictor of severity giving an AUC of 0.92. Optimum cut off value for CRP was 107.5 mg/L (p < 0.0001) producing sensitivity of 91%, specificity of 84%. This was superior to the modified Glasgow Pancreas score, which produced a sensitivity of 36% and specificity of 100%.

Conclusion: We have shown that a CRP value of > 108 mg/L within 48 h of admission can be used to predict severity of acute pancreatitis in children with greater accuracy than current scoring systems.

Type of study: Diagnostic test.

Level of evidence: Level I.

Keywords: CRP; Children; Pancreatitis; Scoring systems.