Validation of a new EKG algorithm for the diagnosis of wide QRS tachycardias in adults with repaired tetralogy of fallot

Introduction

Tetralogy of Fallot (TOF) represents the largest group of adults with repaired cyanotic congenital heart disease, and it is estimated that 32% of these patients will present clinical arrhythmias. Several electrocardiographic algorithms have been described for differentiating between ventricular tachycardia (VT) and supraventricular tachycardia (SVT) with wide QRS complexes, but their usefulness in this population remains unclear.

At our center, a simplified algorithm (Figure 1) has been developed, showing high sensitivity and specificity (98% and 90%, respectively), with an overall efficiency of 95.7%.

Objectives

The aim of our study was to validate this algorithm in an new patient cohort.

Methods

This was a descriptive, longitudinal, ambispective study that sequentially included patients with repaired TOF who underwent electrophysiological study (EPS) between January 2019 and March 2025.

EKGs were blindly analyzed by two expert electrophysiologists.

Classical algorithms—BRI, Wellens, Pava, Vereckei, Brugada, limb lead algorithm, and the simplified UHLP algorithm—were applied for VT diagnosis and their diagnostic performance compared, using EPS results as the gold standard.

Interobserver agreement was also assessed for all algorithms.

Finally, Z tests were performed for sensitivity, specificity, and efficiency to compare classical algorithms with UHLP during the validation phase, and to compare the development and validation phases of the UHLP algorithm.

Results

A total of 50 EKGs from 29 patients were included (mean age 47.62 ± 8.49 years; 20.68% female).

Diagnoses: 50% VT, 36% atrial flutter, and the remainder atrial, AV nodal reentrant, or orthodromic tachycardias.

56% showed right bundle branch block morphology.

There was a non-significant trend toward a wider QRS duration in VT vs. SVT (180 ms vs. 160 ms; p = 0.15), and VT showed a significantly shorter cycle length (296 ms vs. 439 ms; p = 0.038).

Our algorithm achieved the best sensitivity and specificity (100% and 76%, respectively), equal to those of the Pava algorithm, both yielding an efficiency of 88% in our sample (Table 1), with an interobserver agreement of 0.798.

Conclusions

Validation of the algorithm demonstrates high sensitivity, specificity, and efficiency, outperforming the classical algorithms and matching the performance of Pava’s algorithm.

However, our algorithm was specifically designed for patients with baseline conduction disturbances and congenital heart disease.

It also shows good interobserver agreement, confirming that it is simple and reproducible.Figure 1Table 1