To the Editor: We thank Baiges et al. 1 for their keen interest and observations regarding our study. 2 We completely agree that the definition of 'failure to control bleeding' is difficult to apply in individual clinical cases and in general its interpretation is dependent upon many factors including how sick the patients is, whether they have ongoing bleeding during an endoscopy session and the severity of bleeding. In our study, 2 we defined 'failure to control bleeding' strictly as failure to achieve haemostasis despite 2 endoscopies or need for adjuncts such as Sengstaken-Blakemore tube (SBT) or a stent within 5 days of the first bleed in combination with vasoactive drugs.Baiges et al. raise several important questions.