We thank Emir Festic for his comment on our Article.1 Overall, 217 (16·4%) of 1325 patients used anti-arrhythmic drugs (chiefly amiodarone) at baseline in the SERVE-HF cohort (89 [13·5%] of 659 patients randomised to the control group and 128 [19·2%] of 666 patients randomised to the adaptive servoventilation [ASV] group [p<0·05]). Adjustment for this baseline covariate (as a potential confounding variable or as a marker for the perception of a higher propensity to arrhythmia) did not change the estimates of the increased all-cause mortality or cardiovascular mortality in the intention to treat analyses of these secondary endpoints, as reported in the primary manuscript and appendix figures 1A, 1B, and 1C.2 In the multistate analysis,1 we did not present analyses adjusted for antiarrhythmic medication. When this adjustment is made, the unadjusted hazard ratio (HR) of 2·59 (95% CI 1·54–4·37) for ASV versus control changes to 2·52 (1·49–4·26) for the endpoint of cardiovascular death without previous hospital admission for worsening heart failure or lifesaving intervention. Separating analyses of this endpoint with respect to anti-arrhythmic use at baseline reveals that the point estimate for the HR of ASV therapy versus control is higher for those who were taking an anti-arrhythmic (5·53, 95% CI 1·23–24·88), than for the majority who were not taking such a drug (2·19, 95% CI 1·24–3·86) but with overlapping 95% CIs and non-significant interaction (p=0·432). Thus, a confounding or interacting effect of anti-arrhythmic drugs at baseline (or an increased propensity for arrhythmia) is unlikely to be the explanation for the increased risk of cardiovascular death reported in SERVE-HF.
