Individual Patient Data Meta-Analysis of Red Blood Cell Transfusion Trials Comparing Liberal versus Restrictive Thresholds

Project Summary Accumulating evidence from individual clinical trials and conventional trial level meta-analysis suggests that restrictive transfusion threshold of 7 to 8 g/dL is as safe and effective as the 9 to10 g/dL threshold, based largely on an analysis of the primary outcome of mortality. However, it is much less clear whether the safety of

restrictive hemoglobin thresholds applies to all patient groups. In this revised application, we focus the aims on patients with underlying cardiovascular disease given recent results that suggest patients with acute myocardial infarction could be harmed by restrictive transfusion strategy and that it is especially important to

evaluate restrictive transfusion in patients with pre-existing cardiovascular disease. Individual clinical trials do not have adequate sample size and conventional trial level meta-analysis lack the specific detail to examine the effect of transfusion in specific subgroups. An individual patient data meta-analysis (IPDMA) will have the

power and detail to fully explore the effects of transfusion thresholds across clinically important subgroups. For this IPDMA, randomized clinical trials that assigned patients red blood cell transfusions based on transfusion threshold (sometimes also referred to as trigger) have been identified from systematic searches of

the literature. The search will be updated and supplemented by direct query of experts in the field prior finalizing the studies that are included. Data use agreements and letters of support document the commitments of investigators to provide individual patient data from 89% of the participants included in these

trials. The IPDMA will be combined into a single comprehensive analysis database. The aims of the analysis are to 1) Primary Aim: To estimate the treatment effect of liberal versus restrictive transfusion thresholds in patients with cardiovascular disease including those with myocardial infarction, pre-existing cardiovascular

disease and cardiac surgery. We will evaluate clinically important pre-specified risk factors including older age, sex, heart failure, type of MI, baseline hemoglobin concentration, and others on primary and secondary outcomes. 2) Secondary Aim: To estimate the treatment effect of liberal versus restrictive transfusion

thresholds in other clinically important pre-specified subgroups with severe chronic illnesses including cancers and chronic renal diseases, GI bleeding and by age and sex. 3)Exploratory: Use a personalized medicine approach to create models designed to identify risk factors and combinations of risk factors that modify the

effect of transfusion strategy on the primary and secondary outcomes. The results will advance our knowledge about the impact of transfusion threshold in cardiovascular and other clinically important patient groups, where there is concern about the safety of applying a universal `restrictive' hemoglobin threshold for transfusion. If liberal transfusion is shown to be superior to restrictive transfusion in

patient groups, guidelines will be updated and clinical practice will need to change.