Park and colleagues performed an observational study evaluating the impact of transfusion in patients admitted to the ICU with severe sepsis and septic shock. Like most observational studies, the unadjusted analysis found transfusion to be associated with increased mortality. However, when they performed a propensity matched analysis in 152 pairs of patients, transfusion was associated with lower mortality. This interesting finding is at odds with most observational studies in that it suggested transfusion was beneficial.

So, are these results valid? Should they influence care? The authors used propensity matched analysis which is the state of the art approach to analysis of observational data. However, in the propensity analysis, only 304 of the 1054 patients were included. Furthermore, the clinical issue is not whether transfusion should be given but under what circumstances and hemoglobin concentration should transfusion be administered. This analysis examines the effect of all transfusions rather than transfusion at specific hemoglobin levels or physiological triggers. Lastly, these results are inconsistent with the evidence from clinical trials. Thus, I would advise against clinicians using liberal transfusion in ICU patients with sepsis.

– Jeffrey L. Carson