Seven RCTs (10,676 participants) were included. Study size ranged from 174 to 4,628 participants. Follow-up ranged from three to 26 months.
All studies were considered high quality. No details were reported..
There were more deaths in the dronedarone groups than in the control groups, but pooled data showed no statistically significant difference for all-cause mortality (Ι²=53%, seven trials) and cardiovascular mortality (Ι²=75%). Removal of one large study removed heterogeneity for both outcomes, and led to statistically significant worse effects with dronedarone. Removal of a further study on people with heart failure, led to no statistically significant effect on total mortality, but maintained the statistically significant worse effect on cardiovascular mortality (five trials). The effect of removing the study on people with permanent atrial fibrillation had similar results.
The effects on risk of heart failure exacerbations were not statistically significant (Ι²=52%, six trials) until the removal of the same large study that contributed to heterogeneity, resulting in dronedarone being associated with a statistically significant worsening in heart failure exacerbations.
In subgroup analysis dronedarone decreased rehospitalisation rates in people with paroxysmal or persistent atrial fibrillation (two trials). All other secondary outcomes (apart from acute coronary syndrome) showed trends towards worse outcomes with dronedarone, but none were statistically significant.
Sensitivity analyses showed that removing the study that used an active comparator did not change results.