The paradox of on-bypass transfusion thresholds in blood conservation

While most reports on blood conservation define a specific transfusion trigger, few have primarily focused on the role of the predefined transfusion threshold in initiating blood utilization. This study was undertaken to test the hypothesis that rigid adherence to an arbitrarily defined protocol paradoxically increases homologous blood usage during isolated primary coronary artery bypass graft.

Prospectively, 100 consecutive patients were transfused on bypass solely for low venous oxygen saturation (SvO2), ie, <55%, without regard to hematocrit (Hct), postoperative for Hct <20, or if clinically warranted. During bypass the lowest Hct value was <25% in 72 patients, <22% in 52 patients, <20% in 39 patients, <18% in 23 patients, and <15% in 2 patients. These data, then, represent the percentage of patients who would have received blood on bypass had each respective level been used as a trigger, and hence the minimum number of patients who would have been transfused overall. In this study only 13 patients received 2.2+/-0.3 U of red blood cells; 4 on bypass, 5 in the intensive care unit within 24 hours, and 4 on postoperative days 2 or 3. Of the 87 patients not transfused, 15 arrived in the intensive care unit with Hct <25%, 4 with Hct <22%. By postoperative day 1, there were 7 patients with Hct <25% and only 1 <22%, confirming that many of these patients would have been unnecessarily transfused had we adhered to any of the noted on-bypass transfusion triggers. There were no deaths, no strokes, one Q wave myocardial infarction, and one sternal infection. Postoperative blood loss and discharge Hct were 741+/-131 mL and 29.3+/-0.5 versus 573+/-27 mL and 29.1+/-1.0 in transfused and nontransfused patients (P=.24 and P=.88, respectively).

These data suggest that avoiding use of a numerical on-bypass hematocrit trigger is safe and extremely effective in minimizing the use of homologous blood during isolated primary coronary artery bypass graft. Conversely, unless the chosen level is sufficiently low, ie, <15, setting arbitrary thresholds will paradoxically increase homologous blood utilization; data are mean+/-SEM.