A predictive model for long-term coronary artery lesion risk in Kawasaki disease.

INTRODUCTION

A major complication of Kawasaki disease (KD) is coronary artery lesions (CALs), which can lead to myocardial ischemia, myocardial infarction, and even mortality. Therefore, identifying risk factors for CALs is critically important. The purpose of this study was to develop a nomogram model to predict long-term risk of CALs one year later.

MATERIALS AND

METHODS

This retrospective study analyzed clinical data, laboratory test results, echocardiographic findings, and follow-up information from 2,481 pediatric patients diagnosed with KD who were admitted to the Children's Hospital of Soochow University between July 2016 and December 2022. Multivariate logistic regression was used to identify factors associated with long-term CAL risk and construct a nomogram. The model's performance was evaluated using receiver operating characteristic (ROC) curves, areas under the curve (AUC), calibration curves, and decision curve analysis (DCA).

RESULTS

Logistic regression analysis revealed that male sex, prolonged hospitalization, prolonged fever, decreased hemoglobin (Hb), decreased hematocrit (HCT), and hyponatremia were significant predictors of long-term CAL risk in KD patients. In the training dataset, the model achieved an AUC of 0.801, with a sensitivity of 82.5% and a specificity of 65.5%. In the validation dataset, the AUC was 0.796, with a sensitivity of 66.7% and a specificity of 83.5%. The calibration curve was aligned with the predicted curve. Additionally, DCA revealed a high net benefit of the model.

CONCLUSION

The nomogram prediction model exhibited high accuracy and can help physicians identify KD patients who may have long-term CALs.