Comparison of the Balthazar score of acute pancreatitis between computed tomography and ultrasound in children: pitfalls of ultrasound in diagnosing and evaluating pancreatitis

Purpose To demonstrate the utility of ultrasound for predicting the outcome of pancreatitis in pediatric patients using the Balthazar score. Methods Twenty-four children diagnosed with pancreatitis, who underwent computed tomography (CT) followed by ultrasound within 24 h, were included. The Balthazar score was calculated using both modalities based on the appearance of the pancreas (score 0-4). The association between the Balthazar scores with CT and ultrasound was evaluated using single linear regression analysis. To evaluate the sonographic accuracy, the ultrasound severity index and findings were compared with those of CT. Presence or absence of abnormal pancreatic parenchymal change on ultrasound was compared to the presence or absence of pancreatic necrosis on CT. The CT and sonographic findings were evaluated in eight intra-abdominal segments. Results The Balthazar scores yielded by CT (2.2 +/- 1.1, range 0-4) and ultrasound (2.0 +/- 1.1, range 0-4) showed a significantly strong correlation (r = 0.918, p < 0.001). The accuracy of ultrasound in determining the CT Balthazar scores was 91.7% (95% confidence interval 73.0-99.0%, 22/24 patients). In the two cases with pancreatic necrosis on CT, only one case was detected as abnormal pancreatic parenchymal change. Sonographic diagnostic accuracy in the pancreatic head was lower than that in the body and tail (accuracy of the pancreatic head, body, and tail enlargement/edema = 83.3%/75%, 100%/100%, and 100%/100%, respectively). Conclusions The total CT and ultrasound severity indices were significantly correlated. Ultrasound is a useful modality for evaluating not only the initial pancreatic condition but also the severity of pediatric pancreatitis.