Red blood cell transfusions increase hospital-acquired infections (meta-analysis)

Red Cell Transfusions Increase Risk for Nosocomial Infection: Meta-Analysis

Transfusing blood to anemic patients has an almost irresistible intuitive and theoretical appeal both to physicians and the patients who get transfused. It's perhaps the archetypal example of the "find it - fix it" approach to doctoring: correct all laboratory abnormalities and ipso facto, the patient is healthy. For decades, consensus transfusion guidelines advised transfusing anemic patients to normal hemoglobin levels (especially after surgery), or at least to 10 g/dL.

After a century or so, the scientific method began to be applied to transfusion practices, and they haven't held up well at all. Most importantly, the TRICC trial showed that among critically ill patients, giving red blood cell transfusions only when hemoglobin fell below 7 g/dL likely saved lives and improved other outcomes, compared to transfusing to maintain hemoglobin of 10 g/dL. (I avoid the adjective "restrictive" because it wrongly implies we are withholding a helpful therapy.)

Preservatives, red cell deterioration during storage, and immunomodulating substances in donor blood are believed responsible for the harmful effects of blood transfusion. White blood cells are likely bad actors, but are removed from most blood transfused in the U.S., in a process called leukoreduction. Thanks to DNA testing, infectious agents have been almost entirely eliminated from the supply of donor blood.

Guidelines in the wake of TRICC advised blood transfusions only for hemoglobin below 7-8 g/dL for most hospitalized patients, including the critically ill and those post-surgery. But evidence suggests that physicians haven't given up their unhealthy love of red blood cell transfusion since TRICC's publication. Among university hospitals, surgeons varied from 0% to 91% in their rates of blood transfusion for similar surgeries on similar patients, with most of the variation determined by local hospital practices, in one study. Whatever harm physicians might cause by excessive blood transfusion is delayed and lacks clear cause-and-effect, reducing opportunities for feedback and improvement.

Multiple subsequent randomized trials have generally supported TRICC's finding that liberally transfusing red blood cells above hemoglobin of 7-8 g/dL does not improve outcomes, and might be harmful. Nosocomial infection risk appeared to be higher among liberally transfused patients in many of these trials.

For those who still need an extra nudge, a new meta-analysis in JAMA by Rohde et al further consolidates the evidence base against liberal red blood cell transfusion practices. Authors combined 18 randomized trials including 7,593 patients, all of which tested lower vs. higher transfusion thresholds (at varying set points) in the U.S. and Europe, with hospital-acquired infections as an outcome. The absolute risk for nosocomial infection was 12% among conservatively-transfused patients and 17% among liberally-transfused. In total, for every 38 patients treated under a liberal transfusion strategy, another hospital-acquired infection resulted (or was prevented by a more cautious strategy).

The findings were most consistent among orthopedic surgery patients. Interestingly, liberal transfusion did not appear to increase infection risk among critically ill patients, cardiac patients, or patients with gastrointestinal bleeding. Most of the trials' conservative or "restrictive" thresholds were higher than TRICC's hemoglobin of 7.0 g/dL -- most were 8 g/dL, and several 9 -10 g/dL. Among the 4 trials using a threshold of 7.0 g/dL, the number needed to transfuse liberally to cause an infection was only 20.

Leukocyte reduction did not mitigate the infection risk from a liberal transfusion strategy (so other microscopic intermediaries besides WBCs appear responsible for blood's harmful effects).

Since each unit of blood costs >$220 (retail price for a hospital buy it from the supplier) and the patient's insurance company is charged ~$340, with 14 million units of red blood cells transfused in 2011, even a small change in transfusion practice would result in billions of dollars of cost savings in addition to improved outcomes for patients.

This trial does not resolve the ongoing controversy over appropriate transfusion thresholds for patients with ongoing gastrointestinal bleeding or who have acute coronary syndrome. Severe or massive bleeding is obviously exempt as well.

Clinical Takeaway: Stephen Colbert famously said that reality has a well-known liberal bias, but when it comes to blood transfusion, conservative is the way to go for almost all patients. Guidelines strongly recommend limiting red blood cell transfusion in stable medical ward patients until hemoglobin falls <7 - 8 g/dL. The Choosing Wisely agendas for multiple specialties including hematology, hospital medicine and critical care concur. The American Society of Hematology advises using a Hb ≤ 7 g/dL for most nonbleeding medical patients, and ≤ 8 g/dL for those with acute coronary syndrome.

Jeffrey Rohde et al. Health Care–Associated Infection After Red Blood Cell Transfusion: A Systematic Review and Meta-analysis. JAMA. 2014;311(13):1317-1326. doi:10.1001/jama.2014.2726