Sheth and colleagues evaluate the impact of RBC transfusion on mortality in patients presenting to the emergency department with image-confirmed intracerebral hemorrhage. The design is a large single-centre retrospective observational study of 546 patients using prospectively collected data. Of the 546 patients, 144 were anemic at admission and a further 250 developed anemia over the 30-day observational period. One hundred (18%) of the 546 patients were transfused. Their primary analysis adjusting for important prognostic risk factors demonstrated an approximate 3-fold increase in improved 30-day survival for patients transfused at least one unit of RBCs compared to those that did not receive a transfusion. The study has many strengths including the use of consecutive patients to minimize selection bias and robust data such as CT-confirmed intracerebral hemorrhage to minimize both selection and information bias. However, a serious threat to the validity of the primary finding is that of time-dependent bias introduced when the design and analysis does not consider the fact that the longer the patient survives the greater the likelihood of receiving a transfusion. In light of close to 50% of patients developing anemia over their hospital stay, further analyses examining the time-dependent nature of hemoglobin values, transfusion status, and mortality are important. In addition, including non-anemic patients in the primary analysis is problematic as the relevant comparator would be anemic patients that did not receive an RBC transfusion. Nevertheless, the study underscores the need for randomized controlled trials assessing transfusion thresholds across important patient populations.

– Dean A. Fergusson