Background: Left ventricular hypertrophy (LVH) is a marker of cardiac end-organ damage and a risk factor for cardiovascular morbidity and mortality. Although electrocardiogram (ECG) is the most common tool for LVH assessment in contemporary clinical trials and cohort studies, the repeatability of ECG-LVH criteria has not been sufficiently examined. Objectives: Characterize the repeatability and minimal detectable change of ECG-LVH criteria. Methods: A total of 63 participants (mean age 50 years; 31 females) underwent two visits one week apart. At each visit, two digital ECGs were obtained following a standardized protocol. The ECG data were processed centrally to automatically obtain waveform measurements needed to calculate Cornell voltage (CV) LVH (SV3 + RaVL >2800 μV for men and >2000 μV for women), Cornell voltage product (CVP) LVH ((RaVL + SV3) X QRS duration ≥244 mV sec; for both Cornell criteria, 0.8 mV was added to the voltage sum for women), and Sokolow-Lyon (SL) LVH (SV1 +RV5/V6 ≥3500 μV). In addition to using these criteria as dichotomous LVH variables, we also used the waveforms measurements composing them as continuous variables, referred to here as CV-index, CVP-index, and SL-index. We used random-effects, mixed models to parse the variance of the variables into their between-participant, between-visit, and within-visit components, and calculated the intra-class correlation coefficient (ICC), minimal detectable change (95% confidence), and Kappas. Results: Between-participant variation accounted for 93% to 97% of the total variation in CV-index, CVP-index, and log of SL-index. The ICCs (95% confidence intervals) were 0.97 (0.96, 0.98) for CV-index, 0.97 (0.95, 0.98) for CVP-index, and 0.93 (0.90, 0.96) for log of SL-index. Minimal detectable change between repeat measures of CV-index, CVP-index, and log of SL-index were ≥236.7 μV, ≥26.7 mV, and ≥0.09 μV, respectively. The within-visit Kappa (95% confidence limits) was 0.73 (0.38, 1.00) for SL LVH and 1 for the other two LVH criteria. The between-visit Kappa was 1 for CV LVH, 0.66 (0.04, 1.00) for CVP LVH, and 0.40 (-0.03, 0.83) and 0.64 (0.26, 1.00) for the SL LVH first and second measurements of each visit, respectively. When defining ECG-LVH as presence of ≥1 ECG-LVH criteria, the within visit Kappa became 0.78 (0.49, 1.00) for the first visit and 1 for the second visit; the between visit Kappa values became 0.50 (0.12, 0.88) and 0.70 (0.38, 1.00) for the first and second measurements of each visit. Conclusion: CV, CVP, SL indices as continuous variables have excellent repeatability. The dichotomous ECG-LVH criteria have excellent within visit agreement, but between visit agreement ranges from fair to poor, which improved by combining multiple criteria. These results alleviate concerns about the repeatability the ECG LVH use in clinical trials and epidemiologic studies.
Read full abstract