Scaling Echocardiographic Cardiac Dimensions to Body Size: A Bayesian Analysis in Healthy Men and Women.

Background: Proper scaling of cardiac dimensions is of paramount importance in making correct decisions in clinical cardiology. The usual normalization of cardiac dimensions to overall body size assumes an isometric relationship. We sought to investigate these relationships to obtain the best allometric coefficient (AC) for scaling.

Methods: Ninety-seven healthy volunteers were included. The dimensions to be scaled were the left atrial volume, the end-diastolic and end-systolic left ventricular volumes, and the diameter of the tricuspid annulus. A Bayesian statistical analysis was applied with isometric coefficients as priors.

Results: The linear correlations between cardiac dimensions and body size were modest, ranging from 0.12 (-0.10-0.32) for the left atrial volume and height to 0.70 (0.58-0.80) for the end-diastolic volume and height. The ACs varied across the different cardiac dimensions and body size measurements. For the best linear relationships, the isometric coefficients were outside the 95% highest density interval of the posterior distribution for the left atrial volume-weight (AC: 0.7; 0.4-0.9) and end-diastolic volume-height (AC: 2.3; 1.7-2.9), whereas they were different from 1 for the left atrial volume-weight, end-diastolic volume, and diameter of the tricuspid annulus-body surface area (AC: 0.6; 0.3-0.8). Not scaling the cardiac dimensions to their corresponding ACs can lead to important errors in size estimations of cardiac structure.

Conclusions: The ACs found in this study are somewhat different from the corresponding isometric coefficients and often different from 1. This finding should be considered when normalizing cardiac structures to body size when making clinical decisions.