LAUSR

SEE PAGE 2606 W e commonly use absolute risk to determine candidacy for therapy in cardiovascular disease. For example, in patients with atrial fibrillation, annualized stroke estimates guide decision making for anticoagulation (1) and for patients with coronary heart disease, 10-year cardiovascular event risk serves as the basis for statin qualification (2). Inherent in this formulation of risk is the concept of absolute risk reduction and its inverse form, number needed to treat. The higher the absolute risk, the greater the absolute risk reduction and the lower the number needed to treat to benefit 1 patient. At least 1 arena where this concept has not proven consistently true is the prevention of sudden cardiac death with implantable cardioverter-defibrillators (ICDs) (3). Given that ICD therapy is designed to treat potentially fatal ventricular arrhythmias, it would seem sensible that the greatest mortality benefit of ICD therapy would be in those at highest absolute risk of arrhythmic death (4). However, a key missing consideration in this formulation of ICD benefit is the impact of the competing risk of nonsudden deaths, which mitigate the potential lifesaving benefits of ICD therapy. The implications of competing risk are not just theoretical. The majority of patients undergoing ICD implantation never use