Aims: Three randomized trials of implantable cardioverter defibrillator (ICD) therapy vs medical treatment for the prevention of death in survivors of ventricular fibrillation or sustained ventricular tachycardia have been reported with what might appear to be different results. The present analysis was performed to obtain the most precise estimate of the efficacy of the ICD, compared to amiodarone, for prolonging survival in patients with malignant ventricular arrhythmia.

Methods and results: Individual patient data from the Antiarrhythmics vs Implantable Defibrillator (AVID) study, the Cardiac Arrest Study Hamburg (CASH) and the Canadian Implantable Defibrillator Study (CIDS) were merged into a master database according to a pre-specified protocol. Proportional hazard modelling of individual patient data was used to estimate hazard ratios and to investigate subgroup interactions. Fixed effect meta-analysis techniques were also used to evaluate treatment effects and to assess heterogeneity across studies. The classic fixed effects meta-analysis showed that the estimates of ICD benefit from the three studies were consistent with each other (P heterogeneity=0.306). It also showed a significant reduction in death from any cause with the ICD; with a summary hazard ratio (ICD:amiodarone) of 0.72 (95% confidence interval 0.60, 0.87;P=0.0006). For the outcome of arrhythmic death, the hazard ratio was 0.50 (95% confidence interval 0.37, 0.67;P<0.0001). Survival was extended by a mean of 4.4 months by the ICD over a follow-up period of 6 years. Patients with left ventricular ejection fraction < or = 35% derived significantly more benefit from ICD therapy than those with better preserved left ventricular function. Patients treated before the availability of non-thoracotomy ICD implants derived significantly less benefit from ICD therapy than those treated in the non-thoracotomy era.

Conclusion: Results from the three trials of the ICD vs amiodarone are consistent with each other. There is a 28% reduction in the relative risk of death with the ICD that is due almost entirely to a 50% reduction in arrhythmic death.