Reader- and instrument-dependent variability in the electrocardiographic assessment of arrhythmogenic right...

Despite the use of standardized definitions, widely varying prevalence estimates of electrocardiographic (ECG) features related to arrhythmogenic right ventricular dysplasia/cardiomyopathy (ARVD/C) have been reported in different cohorts. This study was aimed at examining the variability in the ECG interpretation resulting from the same reader, different readers, and using different ECG-resolutions.

Blinded to other clinical data, 2 readers examined quantitative and qualitative ECG features of 20 (10 ARVD/C) randomly selected individuals. ECGs were recorded at standard-speed (SS) and double-speed-double-amplitude (DS) settings. The SS ECGs were scanned, magnified 4×, and evaluated using electronic calipers (EL). One reader repeated all measurements. For both readers, the intraclass correlation coefficient (ICC) for the measurement of QRS duration was good between conventional and electronic evaluation [DS vs EL: Reader 1--0.64 (0.52-0.73); Reader 2--0.67 (0.55-0.76)][SS vs EL: Reader 1--0.60 (0.47-0.70); Reader 2--0.60 (0.47-0.70)]. Using the same resolution, the intrareader ICC was good for SS [0.70 (0.59-0.78)], DS [0.85 (0.80-0.90)], and EL [0.70 (0.69-0.83)] resolutions, but deteriorated for interreader comparisons [0.50 (0.36-0.62), 0.75 (0.66-0.82), and 0.75 (0.66-0.82), respectively]. For qualitative parameters, the intra- and interreader agreement was inconsistent for all but 2 parameters. Both readers were in perfect agreement while interpreting right precordial T-wave inversion [κ= 1] and right bundle branch block morphology (RBBB) [κ= 0.83 (0.5-1.0)] even when using SS resolution.

Right precordial t-wave inversion and RBBB are the only ECG parameters that can be detected consistently even using the conventionally used ECG-resolution. The substantial variability in evaluation of other parameters is not improved even with the use of higher resolutions.