Comparison of left ventricular ejection fraction in congenital heart disease by visual versus algorithmic determination.

Abstract Two-dimensional (2D) echocardiographic visual estimation of left ventricular (LV) ejection fraction (EF) in patients with acquired heart disease yields results that are comparable to those obtained by recommended algorithms. However, there is no information concerning its accuracy in congenital heart disease. This study compares applicability, accuracy, and reproducibility of LVEF by visual estimation with that by currently used algorithms (cylinder hemiellipsoid, ellipsoid biplane, and biplane method of disks) in 92 consecutive patients with congenital heart disease but unrepaired complete atrioventricular septal defect, univentricular heart, and hypoplastic left ventricle, using 3D echocardiography as the reference method. Visual estimation of LVEF could be applied in all cases. Because of technically inadequate 2D echocardiographic images for volume measurement, analysis for comparison could be performed in 71 patients (77%), aged 1 day to 47 years. The correlation between 3D echocardiographic and visual estimation was 0.91 (SEE 3.3%), cylinder hemiellipsoid, 0.86 (SEE 3.9%), ellipsoid biplane 0.87 (SEE 3.9%), and biplane method of disks 0.93 (SEE 3.2%). Intraobserver variability was similar for all 2D echo methods. Interobserver variability was greater for visual estimation. In conclusion, visual estimation of LVEF is applicable to most patients with congenital heart disease, yielding results that are comparable to those obtained by currently used 2D echocardiographic algorithms.