What Is the Optimal Transfusion Threshold After Cardiac Surgery?

Authors: Spertus J., N Engl J Med 2015 Mar 12; 372:1069

A randomized trial suggests no significant differences between liberal and restricted thresholds — but leaves room for debate.

The optimal hemoglobin threshold for transfusion after surgical procedures is a longstanding controversy. Observational studies of patients undergoing cardiac surgery have suggested worse outcomes with liberal transfusion thresholds (i.e., transfusion at higher hemoglobin levels), but randomized trials in patients with various indications have shown no meaningful differences. To test the hypothesis that a more-restrictive threshold would result in lower postoperative morbidity and cost, researchers randomized 2003 patients undergoing elective cardiac surgery whose postoperative hemoglobin level dropped below 9 g/dL to either a liberal transfusion hemoglobin threshold (9 g/dL) or a relatively restrictive threshold (7.5 g/dL). All patients received a transfusion of 1 unit of packed red blood cells if their hemoglobin dropped below the assigned threshold and an additional unit if the hemoglobin remained below that threshold.

The liberal-threshold group received a median of 2 units transfused (92.2% received at least one transfusion after randomization) compared with 1 unit in the restrictive threshold (53.4% received at least one transfusion). The primary composite outcome — serious infection or an ischemic event within 3 months after surgery — did not differ significantly between treatments (33.0% for liberal threshold, 35.1% for restrictive threshold), even in risk-based subgroups. Similarly, most secondary outcomes, including costs, showed no differences. The restrictive transfusion group had a significantly higher rate of death at 3 months (4.2% vs. 2.6%). Costs were similar.

Comment
The principal results of this trial suggest no differences in outcomes between the two transfusion strategies following cardiac surgery. The difference in the secondary mortality endpoint, and the fact that the study was designed to detect an absolute 6% difference in the primary endpoint (and some will consider the 2% difference they found to be important), are likely to generate debate and, perhaps, more study. Regardless of one’s interpretation, this trial should at least stimulate institutions to develop specific protocols for managing postoperative anemia. As an editorialist notes, the remarkable variability in U.S. transfusion rates suggests that this is an important target for quality improvement.