Objective: Our purpose was to evaluate whether baseline characteristics predictive of implantable cardioverter defibrillator (ICD) efficacy in the Canadian Implantable Defibrillator Study (CIDS) are predictive in the Antiarrhythmics Versus Implantable Defibrillators (AVID) Trial.
Background: ICD therapy is superior to antiarrhythmic drug use in patients with life-threatening arrhythmias. However, identification of subgroups most likely to benefit from ICD therapy may be useful. Data from CIDS suggest that 3 characteristics (age > or =70 years, ejection fraction [EF] < or =0.35, and New York Heart Association class >II) can be combined to reliably categorize patients as likely (> or =2 characteristics) versus unlikely to benefit (<2 characteristics) from ICD therapy.
Methods: The utility of the CIDS categorization of ICD efficacy was assessed by Kaplan-Meier analysis and Cox hazards modeling. The accuracy of the CIDS score was formally tested by evaluating for interaction between categorization of benefit and treatment in a Cox model.
Results: ICD therapy was associated with a significantly lower risk of death in the 320 patients categorized as likely to benefit (relative risk [RR] 0.57, 95% confidence interval [CI] 0.37-0.88, P =.01) and a trend toward a lower risk of death in the 689 patients categorized as unlikely to benefit (RR 0.70, 95% CI 0.48-1.03, P =.07). Categorization of benefit was imperfect, as evidenced by a lack of statistical interaction (P =.5). Although 32 of the 42 deaths prevented by ICD therapy in AVID were in patients categorized as likely to benefit, all 42 of these patients had EF values < or =0.35. Neither advanced age nor poorer functional class predicted ICD efficacy in AVID.
Conclusion: Of the 3 characteristics identified to predict ICD efficacy in CIDS, only depressed EF predicted ICD efficacy in AVID. Thus physicians faced with limited resources might elect to consider ICD therapy over antiarrhythmic drug use in patients with severely depressed EF values.