Efficacy of ICD in ischemic cardiac disease determined by reconstructing patient-level data from Kaplan-Meier survival curves

Abstract

Background: Implantable cardioverter defibrillators (ICDs) are indicated for the prevention of sudden death in ischemic heart disease. Four randomized trials are the main source of efficacy data. Aims and Methods: Reconstruction of patient-level data from survival curves is a new technique (called the Shiny method or IPDfromKM method) for studying effectiveness. We used this method to analyze the four afore-mentioned trials based on the endpoint of all-cause mortality. After reconstruction by the IPDfromKM method, patient-level data were pooled according to three therapeutic options: ICDs, medical therapy, and no active treatment. Standard statistical comparisons were then made between these three groups of patients. Time-to-event data for reconstructed patients were evaluated using standard Kaplan-Meier analysis. The hazard ratio was estimated to compare and interpret these survival data. Results: A total of 4,621 patients were enrolled (follow-up, 60 months). Of these, 1,827 were treated with ICDs and 1,493 received medical therapy; the control group included 1,301 patients who received no active treatment. Our analysis, based on pooled Kaplan-Meier curves, showed that ICD was the most effective, followed by medical therapy and no active treatment. ICD was associated with significantly better survival compared with medical therapy (hazard ratio, 0.6523; 95% confidence interval [CI], 0.5580 to 0.7622; p<0.001) and no active treatment (hazard ratio, 0.6340; 95%CI, 0.5417 to 0.7424; p<0.001). In the subgroup of patients receiving an ICD, the heterogeneity between trials was negligible, whereas it was significant in the subgroups receiving medical therapy or no active treatment. Conclusions: Our results provide original evidence on long-term survival in ischemic heart disease. Methodologically, our study confirmed the advantages of the Shiny method.