Alternative to Body Surface Area as a Solution to Correct Systematic Bias in Pediatric Echocardiography z Scores

Background: Z scores are the method of choice to report dimensions in pediatric echocardiography. Z scores based on body surface area (BSA) have been shown to cause systematic biases in overweight and obese children. Using aortic valve (AoV) diameters as a paradigm, the aims of this study were to assess the magnitude of z score underestimation in children with increased body mass index z score (BMI-z) and to determine if a predicting model with height and weight as independent predictors would minimise this bias. Methods: In this multicentre, retrospective, cross-sectional study, 15,006 normal echocardiograms in healthy children 1-18 years old were analyzed. Residual associations with body size were assessed for previously published z score. BSA-based and alternate prediction models based on height and weight were developed and validated in separate training and validation samples. Results: Existing BSA-based z scores incompletely adjusted for weight, BSA, and BMI-z and led to an underestimation of > 0.8 z score units in subjects with higher BMI-z compared with lean subjects. BSA-based models led to overestimation of predicted AoV diameters with increasing weight or BMI-z. Models using height and weight as independent predictors improved adjustment with body size, including in children with higher BMI-z. Conclusions: BSA-based models result in underestimation of z scores in patients with high BMI-z. Prediction models using height and weight as independent predictors minimise residual associations with body size and generate well fitted predicted values that could apply to all children, including those with low or high BMI-z.