Risk factors and predictive models for secondary hemorrhage in pediatric coblation tonsillectomy

Abstract

Objective

To identify risk factors for secondary post-tonsillectomy hemorrhage (SPTH) in pediatric patients undergoing Coblation procedures and develop a predictive model to support perioperative management.

Methods

A retrospective analysis was performed on 15,729 children who underwent low-temperature Coblation tonsillectomy at the Children's Hospital, Zhejiang University School of Medicine, between June 2019 and June 2024. Univariate and multivariate analysis were employed to identify factors associated with the occurrence of SPTH. A prediction model was developed based on the identified independent risk factors. To enhance the model's predictive performance and calibration, Weighted Logistic Regression was applied to construct a Nomogram model. The model was validated and evaluated using bootstrap resampling, the consistency index (C-index), Receiver Operating Characteristic (ROC) curve, and Brier Score.

Results

Among the 15,729 patients, 9834 (62.52 %) were boys, and 5895 (37.48 %) were girls.The mean age was 6.55 ± 2.38 years, and the average length of hospitalization was 3.94 ± 1.51 days. Among the 15,729 children, 235 (1.5 %) experienced SPTH, and 42 (0.3 %) required additional general anesthesia for surgical hemostasis. Univariate analysis identified that gender, age, surgical indication, degree of tonsillar embedding, Body Mass Index (BMI), postoperative diet, and surgeon experience were significantly associated with SPTH (P < 0.05).Multivariate logistic regression analysis revealed that severe tonsillar embedding, junior surgeon, obesity, poor postoperative diet, and age ≥12 years were independent risk factors for SPTH. A Nomogram model was developed based on these independent risk factors. The model was internally validated using bootstrap resampling (1000 iterations). The results showed that the C-index was 0.817, indicating good calibration and stability; ROC curve analysis revealed an Area Under the Curve(AUC)of 0.816, demonstrating strong discriminatory ability; and the Brier Score was 0.0155, indicating minimal error between predicted probabilities and actual outcomes.

Conclusion

Severe tonsillar embedding, junior surgeon, obesity, poor postoperative diet, and age ≥12 years are independent risk factors for Coblation SPTH in children. The Nomogram model can quickly and efficiently calculate the bleeding rate of SPTH in children, providing valuable guidance for clinical practice.