# Discriminatory Value and Validation of a Risk Prediction Model Based on Serum Cytokines in Pediatric Acute Appendicitis: A Single-Center Experience of 483 Cases

## TL;DR

This study developed a model using cytokine levels and age to predict complicated appendicitis in children, improving early diagnosis and outcomes.

## Contribution

A novel risk prediction model for complicated appendicitis in children using serum cytokines and age was developed and validated.

## Key findings

- Serum levels of IL-6, IL-8, and IL-10 were highly predictive of complicated appendicitis.
- The model achieved 70% sensitivity and 73.9% specificity with an AUC of 0.7949.
- External validation showed strong accuracy with an AUC of 0.8567.

## Abstract

Objectives: Pediatric acute appendicitis (AA) is one of the most prevalent acute abdominal conditions in pediatric surgery. Children with complicated acute appendicitis (CA) may need timely surgical decisions and have a worse prognosis. In this study, we explored the risk factors and developed a predictive model for complicated AA in children. Methods: A retrospective analysis was conducted on patients data from those hospitalized for acute appendicitis, confirmed by post-surgery pathological results, at Children’s Hospital of Chongqing Medical University between September 2022 and October 2023. Lasso regression was performed to identify risk factors, and multivariate logistic regression analysis was used for model establishment. Results: Serum levels of IFN-γ, IL-5, IL-6, IL-8, and IL-10 before surgery were useful in classifying acute appendicitis in children. IL-6, IL-8, and IL-10, on their own, had high predictive values for CA in children. Independent risk factors for CA were age, IL-10, and IFN-γ. A multifactorial logistic regression prediction model was established, demonstrating good predictive efficacy. Its predictive sensitivity was 70.0%, specificity 73.9%, with an AUC of 0.7949. Furthermore, the results of the external validation indicated that the model’s accuracy was good, with an AUC of 0.8567. Conclusions: Early identification of CA is imperative for timely clinical decision-making. Prediction models based on age, IL-10, and IFN-γ may be reliable and accurate in predicting the incidence of CA, which may lead to better clinical outcomes for children with AA.