Several studies in the past have looked into the topic of red blood cell (RBC) transfusions in the ICU and each one of them supports a rather restrictive approach in the ICU. Still though various guidelines around the world vary due to the lack of evidence (see below).
This october the New England Journal of Medicine (unnoticeably and slowly transforming into a critical care journal ;) published a large multi-centered, partially blinded trial that randomised septic patients in intensive care units to receive RBC at a threshold of 70g/L or 90g/L. The primary outcome was mortality after 90 days. A total of 998 patients finally underwent randomisation and as a result there was no significant difference in mortality after 90 days. Also there were no statistically significant differences in all secondary endpoints like use of life supporting measures, ischemic events, and severe adverse reactions.
This trial adds up to a list of studies showing that a liberal transfusion strategy is not beneficial for patients in critical care. This seems to be especially true for patients with sepsis. And not to forget: a considerable amount of packed RBC can be saved this way.
A higher transfusion threshold of 90g/L in patients with sepsis is non-superior to a lower threshold of 70g/L.
Get an insight into this topic yourself, here's the 'must read's about transfusions:
The TRICC trial
The CRIT study
Sherwood M et al. JAMA. 2014 Feb 26;Vol 311, No.8
The FOCUS trial
A short educational overview can be found here: http://lifeinthefastlane.com/education/ccc/blood-transfusion-in-icu/
Clinical Practice Guidelines from the AABB 2012: Ann Inten Med. 2012;157:49-58
Clinical Practice Guidelines 2009: Red blood cell transfusion in adult trauma and critical care, Crit Care Med 2009