Aims: To compare the sensitivity (SN), specificity (SP), and diagnostic accuracy (ACC) for ventricular tachycardia (VT) diagnosis of five electrocardiographic methods for wide QRS-complex tachycardia (WCT) differentiation, specifically the Brugada, Bayesian, Griffith, and aVR algorithms, and the lead II R-wave-peak-time (RWPT) criterion.
Methods and results: We retrospectively analysed 260 WCTs from 204 patients with proven diagnoses. The SN, SP, ACC, and likelihood ratios (LRs) were determined for the five methods. Of the 260 tracings, there were 159 VTs and 101 supraventricular tachycardias. All five methods were found to have a similar ACC although the RWPT had a lower ACC than the Brugada algorithm (68.8 vs. 77.5%, P = 0.04). The RWPT had lower (60%) SN than the Brugada (89.0%), Griffith (94.2%), and Bayesian (89%) algorithms (P < 0.001). The Griffith algorithm showed lower (39.8%) SP than the RWPT (82.7%), Brugada (59.2%), and Bayesian (52.0%) algorithms (P< 0.05). The positive LRs for a VT diagnosis for the RWPT criterion and the Brugada, Bayesian, aVR, and Griffith algorithms were 3.46, 2.18, 1.86, 1.67, and 1.56, respectively.
Conclusion: The present study is the first independent 'head-to-head' comparison of several WCT differentiation methods. We found that all five algorithms/criteria had rather moderate ACC, and that the newer methods were not more accurate than the classic Brugada algorithm. However, the algorithms/criteria differed significantly in terms of SN, SP, and LR, suggesting that the value of a diagnosis may differ depending on the method used.