Limitations in the aVR algorithm should not lead to a redefinition of ventricular tachycardia

We have prepared the following response to the letter submitted by Drs Vereckei and Miller.1In conclusion, we agree with Drs Vereckei and Miller that current algorithms poorly differentiate VTs from pre-excited tachycardias, and that an electrocardiographic diagnosis of VT has to be seen from that perspective. However, we cannot agree that this limitation should be used as a reason to accept an aVR algorithm misdiagnosis. Pre-excited tachycardias differ from VTs in terms of mechanism, prognosis and treatment. An antidromic circus movement tachycardia can be easily stopped with a simple carotid sinus massage or adenosine administration,9 and later treated with straightforward ablation; this unfortunately is not true of most VTs. Our study compared the ability of five electrocardiographic methods to differentiate ventricular tachycardia (VT) from other broad QRS tachycardias. We used the methodologies described in the original five publications, and a uniform definition of VT/non-VT (SVT). We made it clear that the VT group comprised patients that had only VT.2 In the original papers describing the Brugada and Griffith's algorithms and the lead II criterion, pre-excited tachycardias were not grouped with VTs. In those studies, a diagnosis of VT was only deemed correct when the VT had been confirmed otherwise.3–5 It is surprising that Vereckei et al.6 made no mention of their use of different (and questionable) methodology in the Methods section of their paper. It was surprising to learn that in the Vereckei et al. study, a misdiagnosis made when using the aVR algorithm (e.g. a pre-excited atrial flutter diagnosed to be VT) was ultimately categorized as a correct diagnosis. Such an approach would incorrectly increase the apparent diagnostic power of the algorithm. We strongly disagree with that methodology. A false-positive result should be considered an incorrect diagnosis event, which should therefore lower the specificity and diagnostic accuracy of an algorithm. The approach of Vereckei et al. raises questions as to how a correct answer was categorized, i.e. when a pre-excited tachycardia was diagnosed as an SVT. Vereckei and Miller justified their methodology by stating that because the aVR algorithm cannot differentiate well between pre-excited tachycardias and VTs then the former must be considered as VTs. With respect, we find that unacceptable. To use such ‘logic’ could result in the reclassification of other difficult to diagnose broad QRS tachycardias. Idiopathic fascicular VTs with rapid QRS onset are often misdiagnosed as SVTs when using algorithms, and SVTs in heart failure patients with organic left bundle branch block have many VT-like features and are often misdiagnosed as VTs.7 Should we reclassify the former as SVTs and the latter as VTs? We wonder if the authors' opinion/methodology is biased due to their own algorithm being weak at identifying pre-excited tachycardias (in our cohort, the Pava criterion correctly diagnosed 56% of pre-excited tachycardias, while the aVR algorithm only 22%). Of note, our results showing moderate specificity and accuracy of the aVR algorithm are almost identical to the results of the only other independent, ubiased study.8 Conflict of interest: none declared.