Electrocardiographic predictors of ventricular arrhythmic events in ICD recipients: signal beyond the left ventricular ejection fraction threshold

Selection for implantable cardioverter-defibrillator (ICD) therapy based on left ventricular ejection fraction (LVEF) may miss at-risk patients with LVEF > 35%. We tested whether a simple electrocardiogram (ECG) burden adds discriminatory power for appropriate ICD therapy beyond clinical factors and LVEF. Single-center retrospective cohort of 236 ICD recipients. Six routinely reported ECG abnormalities (heart rate > 75 bpm; QRS duration > 110 ms; obtuse QRS–T angle; delayed transition zone; prolonged QTc; prolonged Tp–Te) were prespecified, summed to form an ECG burden (0–6), and categorized as 0–1, 2–3, or ≥ 4 abnormalities. Logistic regression estimated odds ratios (OR) with 95% CIs for appropriate ICD therapy. Multivariable models included LVEF category (≤ 35% vs > 35%), ischemic vs non-ischemic cardiomyopathy, ICD indication (secondary vs primary prevention), and ECG burden category. Discrimination of the clinical + LVEF model with and without ECG burden was assessed using the AUC (C-statistic) and the DeLong test. Mean age was 62.1 ± 16.2 years; 71% were male; 61% had ischemic cardiomyopathy; 72% underwent primary-prevention ICD implantation; and mean follow-up was 7.6 ± 2.9 years. Appropriate ICD therapy was delivered to 26/236 (11.0%) patients. A high ECG burden (≥ 4 abnormalities) was associated with appropriate ICD therapy after adjustment (OR 3.6, 95% CI 1.8–7.2; p < 0.001). Adding ECG burden to a clinical + LVEF model modestly improved discrimination in this cohort, with the AUC (C-statistic) from 0.67 to 0.72 (DeLong p = 0.01). Kaplan–Meier curves showed graded separation across ECG-burden categories (log-rank p < 0.001). Among patients with LVEF > 35%, therapy was observed more frequently in the high-burden group than in the low-burden group. In this single-center retrospective cohort of ICD recipients, a higher ECG abnormality burden was associated with a higher risk of appropriate ICD therapy and modestly improved discrimination when added to a clinical + LVEF model. These findings are hypothesis-generating and warrant prospective, multicenter external validation using standardized ECG measurements before any clinical application.