Ethical issues with implantable defibrillators.

Implantable cardioverter defibrillators (ICDs) reduce the risk of sudden cardiac death in high risk populations.1 These devices were initially used only in patients who had survived an episode of life-threatening ventricular tachyarrhythmia,2−4 but recent clinical trials have yielded favorable results in populations of patients who have not experienced such arrhythmias but who are at high risk.5−7 In The Second Multicenter Automatic Defibrillator Implantation Trial (MADIT II), for example, myocardial infarction survivors with poor left ventricular function were randomized to ICD or no ICD therapy.7 Patients treated with an ICD had a 3-year mortality rate of approximately 20%, while those receiving no ICD had a mortality rate of about 30% in the same period. The difference in mortality rates was statistically significant. Based on this study, a number of professional societies (including North American Society of Pacing and Electrophysiology [NASPE]) have recently released clinical practice guidelines endorsing ICD implantation for the primary prevention of sudden cardiac death in survivors of myocardial infarction who have a left ventricular ejection fraction < 0.30.8 This raises troubling ethical issues which have not yet received due attention. Consider a hypothetical clinical trial in which survivors of myocardial infarction with low ejection fraction are randomized to receive drug X or placebo. After 3 years mortality is 30% in the placebo-treated group and 20% in the drug Xtreated group. The difference is statistically significant. Mortality rates in the two groups are linear functions of time. The results are shown in Figure 1. For each treatment the area under the curve represents patient years lost due to death. The area between the curves represents the number of life years gained by the drug X-treated patient group over the placebo-treated group. This amounts to 15 years per 100 patients. These 15 years could be distributed evenly to each patient receiving drug X, in which case the benefit per patient would be 0.15 years (or 55 days). Alternatively, the benefit might be unevenly distributed with some patients getting more than 55 days and