LAUSR

We read with great interest the reassessment of confidence of our network meta-analysis, which compared antiarrhythmic drugs for cardioversion of atrial fibrillation (AF), performed by deSouza et al. [1, 2]. We acknowledge that they undertook a laborious task in re-analyzing our published results based on the crucial choice of duration of AF before pharmacologic cardioversion. In their similar network meta-analyses published earlier, they only included studies with AF duration less than 48 h since a longer duration without prior anticoagulation is contrary to guidelines [3]. They included 21 studies and concluded that there is insufficient high quality evidence to determine which treatment is superior for pharmacologic cardioversion of recent-onset AF. They also analyzed 18 studies that examined pharmacologic cardioversion of recent-onset AF within 4 h, again without uncovering sufficient evidence to determine superiority of any treatment [4]. Regarding indications for pharmacologic cardioversion, current guidelines clearly state that immediate cardioversion is indicated in patients already on anticoagulation regardless of episode duration (Figure 16, ESC guidelines for the diagnosis and management of atrial fibrillation) [5]. This is precisely the population with a paroxysm of AF included in the 41 studies of our network meta-analysis, 9 of them extending the duration of AF to 72 h and 11 to a duration of 7 days; therefore, indirectness should not be an issue. This population represents the majority of patients with a paroxysm of AF presenting for cardioversion in the emergency department, since 89% of the total AF population in Western Europe is on anticoagulation based on the CHA2DS2 VASc score [6]. The transitivity assumption underlying indirect comparisons, and hence network meta-analysis, requires that the distribution of effect modifiers is similar for all sources of direct evidence. Transitivity cannot be empirically tested but should be conceptually considered. Considering that the AF duration is effect modifier whose distribution is in balance across the comparisons (patients already on anticoagulation with paroxysmal AF presenting for cardioversion in the emergency department), the assumption of transitivity is met, and it is valid to conduct a network meta-analysis. Crucially, our network meta-analysis identified with sufficient power and consistency the most effective antiarrhythmics for pharmacologic cardioversion over different time settings, with vernakalant and flecainide exhibiting a safer and more efficacious profile towards faster cardioversion [2]. Essentially these are the antiarrhythmics used for cardioversion in the emergency department in patients presenting with AF duration of either 48 h or 7 days, and deSouza et al. needlessly excluded 6 large RCTs examining Vernakalant and 4 additional ones assessing iv flecainide.