An individual patient-based meta-analysis of the effects of dronedarone in patients with atrial fibrillation.

Abstract

Dronedarone is a non-iodinated benzofuran derivative with antiarrhythmic properties. In placebo-controlled atrial fibrillation (AF) trials, the drug was found to have divergent effects on endpoints such as cardiovascular death or hospitalization. The objective of this meta-analysis of all placebo-controlled studies was to provide insights on possible reasons for these divergent effects. Individual data on 9664 patients were used from all AF placebo-controlled studies. The primary outcome measure was cardiovascular death. Cardiovascular hospitalization and hospitalization for heart failure were secondary endpoints. Predefined procedures were used to reduce inter-study heterogeneity adjusting for important baseline variables using a Cox model. Despite adjustments, a significant inter-trial heterogeneity of the outcome of cardiovascular mortality persisted (P-value of 0.005 for the treatment effect × study interaction). Further analyses were conducted in subgroups based on baseline clinical criteria: digoxin co-prescription, advanced heart failure, coronary artery disease, or the presence of permanent AF. These analyses allowed the calculation of a global treatment effect in two important patient subgroups, those with permanent AF in whom there was harm with respect to cardiovascular mortality [hazard ratio (HR) = 2.32; 95% confidence interval (CI) 1.13-4.75] and hospitalization for heart failure (HR = 1.674; 95% CI 1.05-2.67); and those with non-permanent AF in whom there was benefit in terms of cardiovascular hospitalization [HR = 0.751 95% CI (0.68-0.83)]. This meta-analysis demonstrates significant heterogeneity of dronedarone treatment effects across the placebo-controlled randomized trials. The most important predictor of a harmful effect of dronedarone on cardiovascular death and heart failure hospitalization was the presence of permanent AF.