A predictive model for left ventricular hypertrophy in hypertensive children

As current electrocardiographic (ECG) criteria are of low diagnostic value compared with echocardiography (ECHO) for LVH, establishing a more effective ECG predictive model is necessary.

Objective

To develop and validate model to improve the diagnostic capability of ECG for LVH in pediatric primary hypertension.

Methods

A retrospective study of 502 hypertensive children were recruited in the study between January 2019 and December 2024. The cohort were randomly divided into train (n=402) and test sets (n=100) with a proportion of 8:2. LVH was diagnosed using ECHO criteria. A total of 22 ECG parameters were evaluated. A predictive nomogram was developed using least absolute shrinkage and selection operator (LASSO) and multivariate logistic regression.

Results

LVH was identified in 117 (29.1%) of the training set and 29 (29.0%) of the test set. Body mass index (BMI), R _I + S _V4 , and S _D + S _V4 were identified as independent predictors of LVH. The nomogram model showed good performance, with an area under the curve (AUC) of 0.822 in the training set and 0.803 in the test set. Calibration curves and Hosmer-Lemeshow test indicated good agreement between predicted and actual probabilities. DCA demonstrated clinical usefulness. The model outperformed previous models, as confirmed by NRI and IDI.

Conclusions

The nomogram model incorporating BMI, R _I + S _V4 , and S _D + S _V4 significantly improves the ECG diagnosis of LVH in pediatric primary hypertension. The model offers a reliable tool for the early detection of LVH in hypertensive children.