Improved Survival With An Implanted Cardioverter Defibrillator


This report by Moss describes the results of a random- ized clinical trial testing the hypothesis that use of the implantable cardioverter defibrillator (ICD) would improve survival among a group of patients at high risk for sudden death. In order to be eligible for the Multicenter Automatic Defibrillator Implantation Trial (MADIT), all patients had to have a history of myocardial infarction more than three weeks before study entry, a left ventricular ejection fraction (LVEF) of 35% or less, no current indication for surgical or catheter revascularization, and asymptomatic, nonsustained ventricular tachycardia (VT). Patients who met these criteria were referred for a screening electrophysiologic study. Patients were eligible for randomization if sustained VT or ventricular fibrillation was induced with programmed ventricular stimulation both at baseline and after an intravenous procainamide infusion. Patients were then randomized to receive either a transthoracic (n = 45) or transvenous (n = 50) ICD or therapy without an ICD at discretion of the patient’s physician.

One hundred ninety-six patients were enrolled in the trial. The two groups had similar clinical findings prior to enrollment. The mean LVEFs for the ICD and non-ICD groups were 27% ± 7% and 25% ± 7%, respectively. Discretionary antiarrhythmic therapy in the non-ICD group was highly variable. Many patients initially received amiodarone, but only 45% were taking this drug at last contact. More patients in the ICD group received beta blockers or digitalis, but these drugs were probably prescribed in the ICD patients to slow rates in sinus rhythm or atrial arrhythmias rather than prevent ventricular arrhythmias. ACE inhibitor use was similar. Eleven patients in the non-ICD group crossed over to receive an ICD after initial randomization. Five patients randomized to receive an ICD never received a device because the patient either refused or had a high defibrillation threshold, and two implanted devices were later deactivated.

The study was terminated early when an excess mortality was noted in the non-ICD group. During an average follow-up of 27 months, there were 39 deaths in the non-ICD group vs. 15 in the ICD group (hazard ratio for the ICD group, 0.46; P = 0.009). The non-ICD group had 27 cardiac deaths (13 arrhythmic) compared to only 11 (3 arrhythmic) in the ICD group. A Cox regression analysis failed to identify any baseline or later treatment characteristics that influenced the hazard ratio. The authors conclude that implantation of an ICD in patients with prior myocardial infarction, spontaneous nonsustained ventricular tachycardia, and inducible sustained ventricular tachycardia improves survival. (Moss, AJ. N Engl J Med 1996;335:1933-1940.)


Sudden cardiac death remains a major clinical problem, claiming more than 350,000 victims annually in the United States. Despite advances in emergency medical care, rates of resuscitation and long-term survival after an episode of out-of-hospital cardiac arrest remain low, with only a small fraction of patients surviving without significant disability. Numerous studies on the use of antiarrhythmic drug therapy in high-risk groups have failed to demonstrate improved survival, and with some agents, harm has been observed. There is no question that an ICD can effectively terminate episodes of acute ventricular arrhythmias. However, the magnitude of the effects of the ICD on total mortality, even among cardiac arrest survivors, is still uncertain. Several clinical trials assessing ICD effects on long-term mortality are now underway. The MADIT study report is the first of these trials to report final results.

The survival difference between the two groups in this trial is impressive—a 54% reduction in total mortality was observed. This large difference led the study’s data safety monitoring trial to recommend stopping the trial before its planned time of completion. Thus, MADIT entered only a relatively small number of highly selected patients and many questions remain to be answered by future trials.

The control group in MADIT was treated at the local investigator’s discretion. Many, but not all, of these patients received amiodarone at some point, but fewer than half received amiodarone throughout the course of the study. Beta-blocker use was more frequent in the ICD group. Since both amiodarone and beta blockers have been shown to produce favorable effects, the survival difference between the drug and ICD groups may have been smaller if drug therapy had been more carefully controlled.

Patients in MADIT were required to undergo invasive electrophysiologic testing and had to manifest inducible VT that was not suppressed with procainamide. It is not known how many patients underwent screening in this fashion, and the patients with negative studies were not followed. Therefore, we do not know if induction of a drug-resistant sustained VT is a critical predictor of outcome for identifying patients likely to benefit from an ICD. The Multicenter Unsustained Tachycardia Trial is following patients without inducible VT and should provide this information.

The accelerated monitoring plan used by the trial enabled the investigators to detect the early benefit of ICD therapy. However, the magnitude of any long-term survival advantage might be lessened if many of these patients died from nonarrhythmic causes within the next year or two. More long-term data would have been helpful for any future cost/benefit calculations.

It should not be surprising that implantation of an ICD decreases sudden death. The MADIT investigators were able to define a group of very high-risk patients who were still functional in whom a marked short-term improvement in survival was clearly demonstrated. Improvements in ICD technology have made safe implantation possible in virtually any patient. ICD therapy, however, remains expensive, and more accurate estimates of long-term costs and benefits are needed before this strategy can be widely recommended.