Appropriate Utilization of ICDs

Abstract & Commentary

By John P. DiMarco, MD, PhD

Source: Packer, DL, Impact of implantable cardioverter-defibrillator, amiodarone, and placebo on the mode of death in stable patients with heart failure. Circulation. 2009; 120:2170-2176

In this paper, Packer et al report on the modes of death in patients in the Sudden Cardiac Death-Heart Failure Trial (SCD-HeFT). That trial compared the effects of single-chamber implantable cardioverter-defibrillators (ICDs), amiodarone, and placebo in patients with ischemic or nonischemic cardiomyopathy, a left ventricular ejection fraction < 35%, and NYHA Class II or III heart failure. Overall, SCD-HeFT demonstrated superiority of ICD therapy over both placebo and amiodarone in this population.

SCD-HeFT enrolled a total of 2,521 subjects who were randomized in equal proportions to a single-lead ICD programmed to a shock-only mode, amiodarone, or placebo. An independent events committee blinded to treatment reviewed available documentation after treatment-identifying data had been removed. Deaths during the study were classified as witnessed or unwitnessed, sudden or nonsudden, cardiac or noncardiac. Cardiac deaths were also evaluated to see if they were associated with ventricular tachyarrhythmias, bradyarrhythmias, worsened heart failure, or other cardiac causes. Death during sleep was considered to be due to a ventricular tachyarrhythmia if the event was unexpected and occurred in the absence of acceleration of heart failure symptoms. In the case of an event due to a bradyarrhythmia, ECG documentation at the onset of the event was required. Death occurring in a subject with progressively worsening heart failure over the preceding three to four months in whom long-term survival was not expected was classified as a heart-failure death, even if the final event was an arrhythmia.

During follow-up, there were 666 deaths. These included 182, 240, and 244 patients randomized to ICD, amiodarone, or placebo, respectively. Compared to placebo, ICD therapy resulted in a 23% reduction in all-cause mortality and a 24% reduction in cardiac mortality. There was no significant difference in total or cardiac mortality between the amiodarone and placebo groups. The mode of cardiac death differed between groups. In the ICD group, there were 122 cardiac deaths. Of these, 37 (30%) were due to tachyarrhythmia, one due to bradyarrhythmia (0.8%), 72 (59%) due to heart failure, and 12 (10%) due to other causes. In the amiodarone group, there were 162 cardiac deaths. Of these, 75 (46%) were due to tachyarrhythmia, five (3%) due to a bradyarrhythmia, 67 (41%) due to heart failure, and 15 (9%) due to other causes. There were 167 cardiac deaths in the placebo group. This included 95 (57%) due to tachyarrhythmia, three (2%) due to bradyarrhythmia, 66 (40%) due to heart failure, and three (2%) due to other causes. The adjusted hazard ratio for tachyarrhythmic death in subjects randomized to amiodarone compared to those randomized to placebo was not statistically significant (adjusted HR 0.84; p = 0.25), indicating that amiodarone did not decrease arrhythmia mortality in SCD-HeFT. By contrast, the adjusted hazard ratio for tachyarrhythmic death for ICD therapy vs. placebo was highly significant, with a hazard ratio of 0.40, p < 0.001. There were no statistically significant differences between the rates for heart failure death between the three groups. Noncardiac and unclassifiable deaths accounted for 23% of the deaths in the study. There was a slight increase in the number of unclassified deaths in the amiodarone and placebo groups (24 and 24) vs. the ICD therapy group (12), but the total of other than cardiac deaths was not statistically significant between the three groups.

The authors also examined the influence of baseline New York Heart Association (NYHA) heart failure class on mortality. In patients with NYHA class II heart failure, ICD therapy reduced cardiac mortality and sudden tachyarrhythmic deaths (adjusted HR 0.026) compared with placebo. However, there was no difference between treatment groups in heart-failure mortality among the class II patients. By contrast, ICD therapy had no effect on any mode of death in those with New York Heart Association class III heart failure. Amiodarone was associated with an increase in noncardiac mortality in the group with New York Heart Association class III heart failure (adjusted HR 1.68). The effects of ICD therapy were independent of the etiology for heart failure, with similar improvements in cardiac and sudden cardiac mortality seen in patients with both ischemic and nonischemic causes for heart failure.

The authors conclude that amiodarone decreased sudden death, which was presumed to be caused ventricular tachyarrhythmia in patients with New York Heart Association class II heart failure. Patients with class III heart failure have higher relative rates of nonsudden death and receive little benefit from ICD therapy.


This paper helps us further understand the potential and the limitations of ICD therapy. As has also been reported by the MADIT II investigators (see Goldenberg et al. J Am Coll Cardiol. 2008;51:288-296), there seems to be a "sweet spot" for ICD therapy. Among patients at low risk for sudden death, the costs and possible complications of ICD therapy outweigh the relatively small, potential, absolute benefits. Among sicker patients, there are more sudden deaths, but heart-failure deaths begin to predominate, and ICD therapy, although technically effective, offers little long-term benefit. The key is to identify patients in that "sweet spot." In SCD-HeFT, patients with Class II CHF symptoms benefitted from ICD therapy, but patient with Class III symptoms did not. In MADIT II, several risk factors were identified. Patients with either none or more than three risk factors failed to benefit. These observations are sobering in view of the fact that the national ICD registry shows that almost half of the ICDs implanted for primary sudden death today in the United States are given to patients classified as having Class III symptoms. Admittedly, some of these patients also receive resynchronization therapy devices, and that added feature should substantially improve the overall risk-to-benefit ratio in appropriate patients. Despite a wealth of trial-based evidence, we still have much to learn about how to best use ICD therapy.