Dr Annemarie Docherty, University of Edinburgh, Edinburgh, UK.
What is known?
Anaemia is prevalent in critically ill patients, and is associated with poor outcomes including acute myocardial infarction, heart failure, chronic kidney disease and risk of death. In critically ill patients, the standard method of reversing anaemia is with transfusion of red blood cells, with the aim of improving oxygen delivery to the tissues. However, blood transfusion is not without risks. These include immunosuppression, risk for infection, transfusion reactions and transfusion-related acute lung injury. There is conflicting evidence surrounding the association between red blood cell transfusion and mortality, with some studies suggesting a higher risk of death in transfused patients, and others finding a lower risk of death.
What did this paper set out to examine?
The authors have set out to examine whether there is an association between red blood cell transfusion and mortality in critically ill patients. The authors have performed a meta-analysis of all published retrospective and prospective observational studies comparing red blood cell transfused with non-transfused ICU patients, looking at all-cause in-hospital mortality, and risk factors of death in transfused patients.
What did they show?
The authors identified 18 observational studies which looked at mortality of transfused patients. Eight studies were prospective, and the other ten retrospective, six studies were very high overall quality, nine studies high overall quality and three studies median overall quality. The overall pooled risk ratio of in-hospital mortality of transfused patients compared to non-transfused patients was 1.431 (95%CI 1.105 to 1.854). However, in order to account for the impact of the observational design of the studies on the results, they performed several sensitivity analyses, including only studies that adjusted for confounders, only high quality studies, and only studies that included risk or hazard ratios. When only including studies that adjusted for confounding (of particular importance in observational studies), the RR was 1.211 (95%CI 0.795 to 1.505). The authors performed a subgroup analysis looking at different types of admission (sepsis and shock, surgical, trauma, and other). There was no association between RBC transfusion and mortality in each type of admission, however the pooled effect estimate suggested that type of admission was a significant predictor of in-hospital mortality. Other significant predictors were age of patient, and year of publication. Recent studies were more likely to report lower risk ratios, which the authors suggest means that blood transfusion may have got safer over time.
What are the implications for practice and for future work?
As a result of these observational limitations, although this systematic review suggests that RBC transfusion is not linked to in-hospital mortality, a randomised controlled trial designed and powered to answer this question would be required to determine causality. This review suggests that in the heterogenous ICU population, there is no association between RBC transfusion and in-hospital mortality after adjustment for confounders. Clinicians can perhaps be reassured that there does not appear to be an inherent risk with RBC transfusion, and that the decision to transfuse should be based on assessment of the patient’s physiological status and comorbidity.