Predictors of Appropriate Implantable Cardioverter-Defibrillator Intervention in Adults with Congenital Heart Disease

Background: Evidence to guide risk stratification after ICD implantation in adults with congenital heart disease (ACHD) is limited. We sought predictors of appropriate ICD intervention in this population. Methods: Consecutive ACHD patients implanted with an ICD (2009-2023) were analysed retrospectively. Time to first appropriate ICD intervention was modelled from a single, consistently defined start of observation. Because only 24 events were available, candidate predictors were pre-specified and each model limited to two covariates; sparse predictors used Firth penalisation, and multiplicity was addressed by the Benjamini-Hochberg false discovery rate procedure. Results: Among 56 patients (73.2% male; 22 primary and 34 secondary prevention), appropriate intervention occurred in 24 (42.9%) over a median follow-up of 51.9 months. Atrial fibrillation (HR 2.75, 95% CI 1.22-6.23) and radiographic cardiomegaly (HR 3.99, 95% CI 1.28-12.40) showed the largest unadjusted associations, whereas individual congenital lesions and disease-severity descriptors did not. No predictor remained significant after Benjamini-Hochberg correction (q ≈ 0.10). The very high hazard ratios for sparsely represented thyroid disorders (hypothyroidism, n = 2; hyperthyroidism, n = 4) were unstable under penalised estimation and are reported as exploratory. A pre-specified two-variable model (atrial fibrillation plus radiographic cardiomegaly) showed modest discrimination (optimism-corrected Harrell's C-index 0.63). Conclusions: In a heterogeneous real-world ACHD cohort, a history of atrial fibrillation and radiographic cardiomegaly were associated with appropriate ICD intervention. Given the small number of events and the sparsity of several predictors, these findings should be regarded as hypothesis-generating and require validation in larger multicentre ACHD cohorts.