A Randomized Clinical Trial of Red Blood Cell Transfusion Triggers in Cardiac Surgery

Class I evidence supporting a threshold for transfusion in the cardiac surgical setting is scarce. We randomly allocated patients to a transfusion hematocrit trigger of 24% versus 28% to compare morbidity, mortality, and resource use.

From March 2007 to August 2014, two centers randomly assigned 722 adults undergoing coronary artery bypass graft surgery or valve procedures to a 24% hematocrit trigger (n = 363, low group) or 28% trigger (n = 354, high group). One unit of red blood cells was transfused if the hematocrit fell below the designated threshold. The primary endpoint was a composite of postoperative morbidities and mortality. Treatment effect was primarily assessed using an average relative effect generalized estimating equation model.

At the second planned interim analysis, the a priori futility boundary was crossed, and the study was stopped. There was no detected treatment effect on the composite outcome (average relative effect odds ratio, low versus high, 0.86, 95% confidence interval: 0.29 to 2.54, p = 0.71). However, the low group received fewer red blood cell transfusions than the high group (54% versus 75%, p < 0.001), mostly administered in the operating room (low group, 112 [31%]; high group, 208 [59%]), followed by intensive care unit (low, 105 [31%]; high, 115 [34%]) and floor (low, 41 [12%]; high, 42 [13%]). The low group was exposed to lower hematocrits: median before transfusion, 22% (Q1 = 21%, Q3 = 23%) versus 24% (Q1 = 22%, Q3 = 25%).

Negative exposures differed between treatment groups, with lower hematocrit in the 24% trigger group and more red blood cells used in the 28% group, but adverse outcomes did not differ. Because red blood cell use was less with a 24% trigger without adverse effects, our randomized trial results support aggressive blood conservation efforts in cardiac surgery.