In reply
In the opinion of Filion and colleagues, immortal time bias may have affected our 90-day risk for thrombosis and death hazard ratio (HR) estimates for warfarin therapy resumption following gastrointestinal (GI) tract bleeding. We carefully noted the date of warfarin therapy resumptions following index GI tract bleeding during data collection. As noted in our article, we had, in fact, performed the subanalyses suggested by Filion and colleagues.1 Because they did not materially alter our results, we chose not to include them in our original article. We now welcome the opportunity to report the results of our time-varying warfarin exposure and competing risk for death subanalyses.