Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

BACKGROUND: Transfusion thresholds for acute upper gastrointestinal bleeding are controversial. So far, only three small, underpowered studies and one single-centre trial have been done. Findings from the single-centre trial showed reduced mortality with restrictive red blood cell (RBC) transfusion. We aimed to assess whether a multicentre, cluster randomised trial is a feasible method to substantiate or refute this finding. METHODS: In this pragmatic, open-label, cluster randomised feasibility trial, done in six university hospitals in the UK, we enrolled all patients aged 18 years or older with new presentations of acute upper gastrointestinal bleeding, irrespective of comorbidity, except for exsanguinating haemorrhage. We randomly assigned hospitals (1:1) with a computer-generated randomisation sequence (random permuted block size of 6, without stratification or matching) to either a restrictive (transfusion when haemoglobin concentration fell below 80 g/L) or liberal (transfusion when haemoglobin concentration fell below 100 g/L) RBC transfusion policy. Neither patients nor investigators were masked to treatment allocation. Feasibility outcomes were recruitment rate, protocol adherence, haemoglobin concentration, RBC exposure, selection bias, and information to guide design and economic evaluation of the phase 3 trial. Main exploratory clinical outcomes were further bleeding and mortality at day 28. We did analyses on all enrolled patients for whom an outcome was available. This trial is registered, ISRCTN85757829 and NCT02105532. FINDINGS: Between Sept 3, 2012, and March 1, 2013, we enrolled 936 patients across six hospitals (403 patients in three hospitals with a restrictive policy and 533 patients in three hospitals with a liberal policy). Recruitment rate was significantly higher for the liberal than for the restrictive policy (62% vs 55%; p=0·04). Despite some baseline imbalances, Rockall and Blatchford risk scores were identical between policies. Protocol adherence was 96% (SD 10) in the restrictive policy vs 83% (25) in the liberal policy (difference 14%; 95% CI 7-21; p=0·005). Mean last recorded haemoglobin concentration was 116 (SD 24) g/L for patients on the restrictive policy and 118 (20) g/L for those on the liberal policy (difference -2·0 [95% CI -12·0 to 7·0]; p=0·50). Fewer patients received RBCs on the restrictive policy than on the liberal policy (restrictive policy 133 [33%] vs liberal policy 247 [46%]; difference -12% [95% CI -35 to 11]; p=0·23), with fewer RBC units transfused (mean 1·2 [SD 2·1] vs 1·9 [2·8]; difference -0·7 [-1·6 to 0·3]; p=0·12), although these differences were not significant. We noted no significant difference in clinical outcomes. INTERPRETATION: A cluster randomised design led to rapid recruitment, high protocol adherence, separation in degree of anaemia between groups, and non-significant reduction in RBC transfusion in the restrictive policy. A large cluster randomised trial to assess the effectiveness of transfusion strategies for acute upper gastrointestinal bleeding is both feasible and essential before clinical practice guidelines change to recommend restrictive transfusion for all patients with acute upper gastrointestinal bleeding. FUNDING: NHS Blood and Transplant Research and Development.

Original publication




Journal article



Publication Date





137 - 144


Acute Disease, Adult, Aged, Aged, 80 and over, Erythrocyte Transfusion, Feasibility Studies, Female, Gastrointestinal Hemorrhage, Guideline Adherence, Hemoglobins, Humans, Male, Middle Aged, Patient Selection, Research Design, Selection Bias