Heart failure complicated by atrial fibrillation (AF) is associated with excessive mortality and morbidity. The aim of the study was to determine the role of amiodarone or implantable cardioverter/defibrillator (ICD) in patients with AF and heart failure.

Patients were determined to be in sinus rhythm (SR) or AF on the baseline electrocardiogram. Mortality, ICD discharge, or change in rhythm was assessed.

Of the 2521 patients at baseline, 2328 were in SR and 173 were in AF. Overall, after adjusting for differences in baseline variables, there was no difference in mortality between patients with SR and patients with AF (P = .98), nor within assigned groups: placebo (P) (P = .82), amiodarone (A) (P = .68), and ICD (P = .40). For patients with SR, ICD decreased mortality (P vs ICD, P = .004; A vs ICD, P = .004; P vs A, P = .75). For patients with AF, there were no differences in mortality among groups (P vs ICD, P = .99; A vs ICD, P = .88; P vs A, P = .88). Of patients with SR at baseline, 11% (264) developed AF by any electrocardiogram during follow-up (P 12%, A 8%, ICD 15%; A vs P, P = .019; A vs ICD, P = .001; P vs ICD, P = .044). Of patients with AF, 70% (121) developed SR during follow-up (P 66%, A 67%, ICD 75%, all P = not significant against each other). Any ICD shock was seen in 52% (34) of patients with AF vs 30% (222) of patients with SR (P = .001). Inappropriate shocks were seen in 37% (24) of patients with AF vs 14% (107) of patients with SR (P = .001). Appropriate shocks were more often seen in AF vs SR (P = .03).

After adjustments for baseline differences, patients with AF and patients with SR have similar overall mortality rates. Compared to P or A, ICD improves survival in patients with SR, but may not in patients with AF. Amiodarone is effective in reducing new AF, but not in converting AF to SR. Implantable cardioverter/defibrillator, inappropriate, and appropriate shocks were more often seen in AF than in SR.