Source: Hill, S et al. Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Data Syst Rev. 2010 Oct 6;10:CD002042.
Hajjar, LA et al. Transfusion requirements after cardiac surgery: the TRACS randomized control trial. JAMA. 2010 Oct 13;304(14):1610-1.
Gould S et al. Packed red blood cell transfusion in the intensive care unit: limitations and consequences. Am J Crit Care. 2007 May;16(3):206
Jacob, EK. Anderson, CF. Blood Transfusion Risks: By Mayo Clinic Staff
Dhingra, N. Making Safe Blood Available in Africa. World Health Organization June 27, 2006
Efficacy Endpoints: Probability of red cell transfusion
Harm Endpoints: Mortality, cardiac events (MI, arrhythmias, cardiac arrest, pulmonary edema and angina), length of hospital stay
Narrative: : In 2001 nearly 5 million patients in the United States received transfusion of approximately 14 million units of packed red blood cells (Gould et al 2007). In the last few decades the screening process in the United States has become quite rigorous, reducing the risks of infectious disease transmission (Jakob, Anderson 2009), however, risks in the form of adverse reactions to blood transfusion remain substantial. In addition, in developing countries unable to perform advanced screening some estimates have put the risk of contracting HIV from a blood transfusion between 5 and 10% (Dhingra 2006), which, if accurate, will dramatically affect regional risks associated with transfusion. One obvious strategy to reduce risks is to limit unnecessary transfusion.
The Cochrane Review identified 10 randomized control trials involving 1780 patients examining conservative vs. liberal transfusion thresholds. There were no statistically significant differences in mortality, adverse cardiac events, or length of hospital stay between the two groups. Two of the included studies did find an increased incidence of pulmonary edema among subjects randomized to liberal transfusion protocols.
It should be noted that the studies that the Cochrane Review included did not enroll patients with serious heart conditions, and the review authors made a special point that the safety of conservative transfusion thresholds for this subset of patients cannot be determined from these data. However, a recently published randomized trial of 502 patients undergoing cardiopulmonary bypass surgery (Hajjar et al 2010), which was not available during the last Cochrane update, showed no statistically significant difference in mortality or adverse events between patients randomized to liberal or conservative transfusion strategies.
Caveats: The majority of the evidence in the Cochrane Review was provided by one study (Herbert, PC et al. 1999) of 838 ICU patients that was stopped early due to lagging enrollments. Only two of the ten studies recruited more than 100 subjects. Thus their analysis was skewed heavily by the results of this single trial involving ICU patients, which compromises the generalizability of the data. A second issue is the heterogeneity of the transfusion protocols among the included trials. While the largest trial used a hemoglobin of 10.0 and 7.0 for the liberal and conservative transfusion thresholds respectively, other studies used varying hematocrit levels as the trigger, or instead followed a protocol of immediate post-operative transfusion vs. transfusing at a hemoglobin below 9.0. So while no benefit was associated with ‘liberal’ transfusion triggers, there was no consensus on what defined 'conservative'.
Despite these limitations, the enrollment of nearly 1800 subjects in these studies, the relatively high quality of the largest single study in the review, and the fairly consistent results (no benefit to liberal transfusion strategies) suggest that there is no reason to expose patients to the risks associated with liberal transfusion strategies.
Finally, as previously mentioned, there are potentially dramatic variations in the safety of blood transfusion based upon the setting and screening processes, suggesting that the reduction of transfusions may lead to more pronounced benefits (via reductions in harms and adverse events) in some settings.
Overall, the liberal use of blood transfusions showed little to no benefit in these studies but did cause harm in the form of pulmonary edema. It seems virtually certain that transmission of infectious diseases will also occur with increasing numbers of transfusions, though the frequency of these harms will depend heavily upon methods of screening and regional variations in risk. Because harms appear to be certain and because no benefit can be identified, despite the small numbers of subjects enrolled and other minor weaknesses in the data, we have decided that this data suggests a rank of 'Caution.'
Author: Dan Runde, MD and Jarone Lee, MD, MPH
Published/Updated: November 7, 2010
The title bar is color-coded with our overall recommendation.
If you have suggestions, requests, or questions about a particular NNT review, please send us a message and we’ll try to address it as soon as possible.