We are grateful for the interest in our study1 and for the insightful comments by Lu et al2 and Wang et al3. Lu et al astutely observed that our study did not use any bleeding risk prediction tool such as the HAS-BLED score. The technical reason for this was the unavailability of certain variables in our database that are required to calculate HAS-BLED. The substantive reason was that stroke or bleeding risk prediction scores are derived in general populations and perform poorly when applied to individuals with kidney failure receiving dialysis.