Abstract 14378: Implantable Cardioverter-Defibrillator Use in Nonischemic Cardiomyopathy - A Bayesian Analysis

Abstract

Background: Prophylactic implantable Cardioverter-Defibrillator (ICD) therapy has been a cornerstone of sudden cardiac death prevention in patients with left ventricular systolic failure. While the benefit of ICD in ischemic cardiomyopathy has been well established, its benefit in nonischemic cardiomyopathy (NICM) is debated. Methods: We conducted a post-hoc Bayesian analysis of the DANISH randomized clinical trial (N=1,116) to estimate the probability of clinical benefit of ICD compared with medical therapy (control) to reduce all-cause mortality among participants with NICM and who were not on cardiac resynchronization therapy (N=471). Bayesian modeling was used to ascertain the posterior probability of a reduction in all-cause mortality of ICD compared with medical therapy using 1) a noninformative prior (Beta[1,1] distribution), and 2) using an informative prior (derived from the random-effect meta-analysis comparing ICD versus control in the NICM cohorts of the SCD-HeFT, CAT, and DEFINITE trials, N=1,354). Results: Using a noninformative Beta[1,1] prior, Bayesian analysis of the DANISH trial indicates that there is a 74.2% probability of a reduction in all-cause mortality (ARR [absolute risk reduction]: -2.6% [-10.5 to 5.3%]; RRR [relative risk reduction]: 0.90 [0.67-1.22]) among NICM patients receiving an ICD compared with control. Using an informative prior, Bayesian analysis of the DANISH trial indicates a 99.7% probability of a reduction in all-cause mortality (ARR: -5.2% [-9.0 to -1.4%]; RRR: 0.80 [0.67-0.95]) and a 98.6% probability of an ARR<1% among NICM patients who received an ICD compared with control. Conclusion: Using a Bayesian approach, our post-hoc analysis of the DANISH trial provides probabilities of reducing the all-cause mortality in NICM patients using ICD compared with medical therapy.