The aim of the study was to assess and compare the safety of antiarrhythmic drugs (AADs) in an unselected real-world population of patients with atrial fibrillation (AF).

This is a study of all patients with diagnosed AF in the Swedish Patient register who filled a prescription for sotalol, amiodarone, dronedarone, flecainide or disopyramide during 2010 to 2015. The main end point consisted of arrhythmic death, successful resuscitation, new diagnosis of sustained ventricular tachycardia, ventricular fibrillation or implantation of ICD. All-cause mortality was a secondary end point. Minimum follow up was 1 year. Falsification end points were used to assess hidden confounding.

44,995 AF patients on AAD and 267,518 AF patients without AAD were studied during a total time at risk of over 1.1 million years. Compared to sotalol, the risk for the main end point was decreased with dronedarone (hazard ratio [HR] 0.58, 95% confidence interval [CI] 0.37–0.90), similar with flecainide (HR 0.95, 0.69–1.32) and disopyramide (HR 1.30, CI 0.83–2.05). All-cause mortality was lower with dronedarone (HR 0.44, CI 0.34–0.57) and flecainide (HR 0.55, CI 0.44–0.68) than with sotalol. Hidden confounding prevented reliable assessment of amiodarone.

Dronedarone was the only anti-arrhythmic drug with significantly lower risk for arrhythmic death, sustained ventricular arrhythmia or ICD implantation than sotalol among patients with atrial fibrillation. Both dronedarone and flecainide were associated with lower all-cause mortality than sotalol.