Role of the vectorcardiogram-derived spatial QRS-T angle in diagnosing left ventricular hypertrophy Sumche Man, MD, Chinar Rahmattulla, BSc, Arie C. Maan, PhD, Eduard Holman, MD, Jeroen J. Bax, MD, Ernst E. van der Wall, MD, Martin J. Schalij, MD, Cees A. Swenne, PhD ⁎ Department of Cardiology, Leiden University Medical Center, Leiden, The Netherlands Received 14 June 2011 Abstract Introduction: Current criteria for electrocardiographic (ECG) diagnosis of left ventricular hypertrophy (LVH) have a low diagnostic accuracy. Addition of demographic, anthropomorphic, and additional ECG variables may improve accuracy. As hypertrophy affects action potential morphology and intraventricular conduction, QRS prolongation and T-wave morphology may occur and become manifest in the vectorcardiographic variables spatial QRS-T angle (SA) and spatial ventricular gradient. In this study, we attempted to improve the diagnostic accuracy for LVH by using a combination of demographic, anthropomorphic, ECG, and vectorcardiographic variables. Methods: The study group (n = 196) was divided in 4 subgroups with, on one hand, echocardiographically diagnosed LVH or a normal echocardiogram and, on the other hand, with any of the conventional ECG signs for LVH or with normal ECGs. Each subgroup was randomly split into halves, yielding 2 equally-sized (n = 98) data sets A and B. Age, sex, height, weight, body mass index, body surface area (BSA), frontal QRS axis, QRS duration, QT duration, maximal QRS vector magnitude, SA, and ventricular gradient magnitude and orientation were univariate studied by receiver operating characteristic analysis and were used to build a stepwise linear discriminant model using P b .05 as entry and P N .10 as removal criterion. The discriminant model was built in set A (model A) and tested on set B. Stability checks were done by building a discriminant model on set B and testing on set A and by cross-validation analysis in the complete study group. Results: The discriminant model equation was D = 5.130 × BSA - 0.014 × SA - 8.74, wherein D greater than or equal to 0 predicts a normal echocardiogram and D less than 0 predicts LVH. The diagnostic accuracy (79%) was better than the diagnostic accuracy of conventional ECG criteria for LVH (57%). Conclusion: The combination of BSA and SA yields a diagnostic accuracy of LVH that is superior to that of the conventional ECG criteria. © 2012 Elsevier Inc. All rights reserved. Keywords: Vectorcardiogram; Spatial QRS-T angle; Left ventricular hypertrophy Introduction Electrocardiographic diagnosis of left ventricular hyper- trophy (LVH) has challenged researchers for decades. 1,2 Several criteria to diagnose LVH using the limb leads/frontal plane 3-8 and the chest leads/transversal plane 8-12 are listed in the recommendation for standardization and interpretation of the electrocardiogram (ECG). 1 All criteria have in common that they require R and/or S amplitudes to reach a given threshold, and in addition, combinations with non-voltage criteria (eg, left axis deviation) are also mentioned. 6 All proposed criteria were validated either by autopsy or by echocardiography. Evidently, none of the proposed criteria had sufficient diagnostic accuracy, as in the course of time several different diagnostic rules remained to be presented. 1,2 In a recent review, 2 the sensitivity and specificity of commonly-used ECG LVH criteria (eg, Sokolow-Lyon criterion) varied from 0% to 68% and 53% to 100%, respectively. Therefore, as recommended in the ECG LVH criteria guideline, 1 further studies are needed to define a better LVH criterion, among others by inclusion of demographic, anthropomorphic (eg, age and weight), and ECG variables (eg, QRS duration). 1 Hypertrophy is associated with alterations in the action potential morphology. 13-20 In spatial ECG (vectorcardiography, Available online at www.sciencedirect.com Journal of Electrocardiology 45 (2012) 154 – 160 www.jecgonline.com ⁎ Corresponding author. Cardiology Department, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, The Netherlands. E-mail address: c.a.swenne@lumc.nl 0022-0736/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jelectrocard.2011.10.001