Abstract 107: Predictors and Frequency of ICD Implantation for Primary Prevention in ICD Eligible Patients: Results from a Prospective Multicenter Registry
Background . The use of implantable cardiac defibrillators has been associated with improved survival in several well-defined patient (pt) subsets. Its utilization for primary prevention in eligible pts, however, is unclear. We sought to examine the frequency of ICD implantation (ICD-IMP) for primary prevention in a cohort prospectively enrolled in a prospective, multicenter registry of ICD candidates. Methods . We identified 961 pts enrolled in the AdreView Myocardial Imaging for Risk Evaluation in Heart Failure (ADMIRE-HF) study, a prospective, multicenter study evaluating the prognostic usefulness of 123I-mIBG scintigraphy in a heart failure population. Inclusion criteria limited patients to those meeting guideline criteria for ICD implantation; these criteria included left ventricular ejection fraction ≤35% and New York Heart Association functional class II-III. We excluded pts with an ICD at the time of enrollment, leaving a study cohort of 934 patients. Pts were followed up for 24 months after enrollment. Pts undergoing ICD-IMP after enrollment for secondary prevention were censored at the time of intervention. The association between ICD-IMP utilization and demographic, clinical, laboratory, and imaging data was examined using Cox proportional hazards analysis (CPH). Results . Of 934 pts, 196 (21%) were referred for ICD-IMP over a mean follow-up of 612±242 days. Implantations occurred 167±164 days after enrollment. Patients referred for ICD were younger (61±12 vs. 63±12), but did not differ with respect to proportion female (17% vs. 21%), African-American race (12% vs. 15%), diabetics (37% vs. 36%) (All p=NS). The frequency of ICD-IMP did not differ as a function of age, race, sex, LVEF, or imaging result (All p=NS). CPH revealed that a model including age, race, sex, diabetes, smoking, BMI, NYHA class, hypertension, heart failure etiology, and prior MI identified none of these as predictive of ICD-IMP. Conclusion: This analysis of prospective registry data reveals that in patients who are guideline-defined candidates for ICD-IMP, only about one in five receive an ICD over a two year follow-up interval. Multivariable modeling failed to identify any factor associated with ICD use.