The development and application of pediatric complicated appendicitis prediction model

Acute appendicitis in children refers to the acute inflammation of the appendix, which accounts for 20% ∼ 30% of cases of acute abdomen in pediatric surgery.

This study aimed to establish a decision tree model of complicated appendicitis in children using appendiceal ultrasound combined with an inflammatory index and evaluated its clinical efficacy in pediatric patients.

A total of 395 children admitted to the Emergency Department of the Shanghai Children's Hospital from January 2018 to December 2021 and diagnosed with appendicitis by postoperative pathology were retrospectively analyzed. According to the postoperative pathology, the children were divided into a complicated and non-complicated appendicitis group, respectively. Routine laboratory inflammatory indicators, including white blood cell count, N(%), neutrophil (Neu) count, Neu/lymphocyte ratio (NLR), C-reactive protein (CRP), and procalcitonin were collected from the two groups. Collecting data on ultrasound examination of the appendix includes whether the appendix diameter is thickened, whether the echogenicity of the mesenteric rim surrounding the appendix is enhanced, whether there is rich blood supply in the appendix, and whether there are fecaliths in the appendix lumen. The risk factors for complicated appendicitis were screened out by univariate and multivariate logistic regression analyses, the binary logistic regression prediction and decision tree models were established, respectively, and the receiver operating characteristic (ROC) curve was used to verify the accuracy of the two prediction models.

Binary logistic regression analysis showed that CRP, NLR, the presence of an appendicolith, and peripheral retina echo enhancement were independent risk factors for complicated appendicitis in children (P< 0.05). The decision tree model had an overall accuracy of 79%, an area under the ROC curve (AUC) of 0.809 (95% confidence interval [CI] 0.780-0.865), and sensitivity and specificity of 71.3% and 77.7%, respectively. The logistic regression model had an overall accuracy of 74.9%, an AUC value of 0.823 (95% CI, 0.765-0.853), a sensitivity value of 80.3%, and a specificity of 71.8%.

This predictive model, based on ultrasound of the appendix combined with inflammatory markers, provides a useful method to assist pediatric emergency physicians in diagnosing childhood appendicitis. The decision tree model reflected the interaction of various indexes, and the model was simple, intuitive, and effective.