The author mentions that other databased studies, including ours, have examined outcomes and lowest hematocrit, found bad outcomes associated with low hematocrits but failed to examine the effect of transfusion in their analyses.
I would like to point out that this limitation was acknowledged in our report published in the British Journal of Anaesthesia (1998;1:38-45) and that, conversely, the same applies to the article of Engoren et al. who could not isolate the effect of anemia on long term outcome.
In summary, databased reviews generate interesting hypotheses but cannot replace prospective randomized studies such as those published recently in critical care patients (TRICC trial).
– Jean-Franois Hardy