However, despite a highly sophisticated matching strategy that compares unique transfused with unique non-transfused patients, one cannot exclude that further undocumented clinical criteria of illness caused the responsible physicians to see an indication for transfusion. Particularly, this must be considered in view of the relatively low number of patients (mean 24) of the numerous hospitals participating (146).
Therefore, it is still open to discussion whether transfused patients were more ill and therefore received blood transfusions showing worse outcome despite transfusions, or whether transfusions really caused the worse outcome. As long as there is no randomization for transfusion (yes or no) after the clinicians decision to transfuse (yes or no), the question about the independent effect of transfusion cannot be reliably answered. The data regarding the outcomes are all the more so surprising as leukodepleted RBCs have been used most of the time by 46% of the ICUs and never in only 19%.
The Editorial of Hebert and Fergusson in the same journal is recommended for reading as it reflects a rather objective opinion on this study.
– V. Kretschmer