Complete Atrioventricular Block does Reduce Mortality in Patients with Atrial Fibrillation Treated with Cardiac Resynchronization Therapy: Reply

We appreciate the comments from Gasparini et al. in relation to the contradictory results of our study, as compared with other populations. We strongly believe that the difference in the analysed populations may explain this discrepancy. Specifically, our population includes patients with previous atrioventricular (AV) block and also those submitted to AV ablation due to poor control.1 A recent meta-analysis2 described a lower mortality in patients with AF treated with CRT if they had undergone ablation of the AV junction (AVJ). The conclusion is based on three observational studies;345 however, the non-randomized nature of the studies certainly weakens the conclusions. Mortality in our AF + AVJ block group is higher than in other series. The different patient profiles may explain this difference. Unlike previous studies, we included in the same group patients with AF and spontaneous complete AV block and patients who underwent AVJ ablation due to poor rate control. This reduces the percentage of patients with fast ventricular rates and tachycardiomyopathy in which the ‘ablate and pace’ strategy provides excellent results and prognosis just by controlling the heart rate. This fact may worsen the prognosis of this group of patients. Gasparini et al.3 6 described similar mortality between patients in sinus rhythm and patients in AF + AVJ ablation treated with CRT. However, AF has been recognized as an independent predictor of mortality in CRT patients.789 Again we believe that differences in patient characteristics may explain the discrepancy. One of the criticisms of the authors is the small number of patients of our series, considering that this weakens our conclusions. Nevertheless, based on previous data, our study has a statistical power of 89% to obtain a hazard ratio (HR) of 0.42 in a 0.050 level two-sided log-rank test for equality of survival curves, providing sufficient statistical power.2 Additional tests were made to analyse our Kaplan–Meier curves: the Breslow test that weights each time point to the number of patients at risk, over-representing early stages in comparison with later ones. This did not show significant differences (P = 0.234). Additionally, Renyi-type statistics provide a specific approach for scenarios where survival curves cross, testing for significance at the time point when differences are maximal. Again, the difference did not achieve statistical significance in non-weighted (P = 0.404) or Breslow-weighted (P = 0.26) Renyi-type statistics. Overall, new analyses show a robust lack of statistical significance at the aforementioned statistical power of 89%. A maximum percentage of ventricular pacing is necessary to obtain a response to CRT; however, based on our observations, we still believe AVJ ablation is probably not mandatory, unless there is an insufficient rate control.