>0.7 and 26% (152/587) >1 at the time of presentation with their GI bleed. This is a very different patient cohort from that of Saffouri et al 2 As shock index is a simple calculation of systolic blood pressure/ heart rate, it is perhaps predictable that it will perform more poorly than scoring systems which include heart rate and systolic blood pressure as components (Glasgow-Blatchford and Rockall scores). AIMS65 includes systolic blood pressure alone. In the paper by Saffouri et al 45.5% of patients (1368/3012) had an endoscopically confirmed upper GI bleed. The remainder had a normal endoscopy (297), no endoscopy (937) or are unaccounted for in the paper (410). The NCEPOD report made 26 recommendations to improve the care of patients with GI bleeding. These included a recommendation to encourage the development of risk stratification tools relevant to all GI bleeding presentations and sites of bleeding as clinical presentations do not reliably predict the site or type of bleeding. Embedding any risk assessment tools in care pathways is essential. A scoring system applicable to all GI bleeds at presentation may be better adopted.