Predicting the need for blood during cardiopulmonary bypass

Background: Haematocrit (Hct) values < 18%-20% during cardiopulmonary bypass (HctCPB) are potentially unsafe. Aims: 1. To predict when banked-blood should be pre-issued. 2. To evaluate the sparing-effect of banked-blood by autologous blood transfusions. Methods: An equation for prediction of HctCPB (Hctpred), based on weight and pre-operative haemoglobin concentration was used to forecast which patients would develop HctCPB = 20%. Perioperative blood and fluid administration were recorded in 80 patients requiring CPB. Blood and fluid administration strived for HctCPB = 18% on CPB and 33% in the ICU. Results: Hctpred bias and precision were 2.6% and 13.1%. A Hctpred cut-off value of 23% reliably forecast a HctCPB = 20% (15 patients with mean HctCPB 16.5%). Despite a 31% false positive rate (FPR), there is emphasis on safety associated with the 23% Hctpred cutoff-point. (100% negative predictive value; zero negative likelihood ratio). Applying the same predictive criterion to all blood transfusions performed in the OR, increased positive predictive values from 43% to 63% so that the FPR decreased to 24%. Autologous transfusion comprised 72% of transfused blood and was the only transfusion in 67% of patients. Banked-blood recipients weighed less and had lower pre-operative haemoglobin concentrations, Hctpred and HctCPB. They received larger transfusions of which autologous blood formed 46%. Conclusions: 1. It is possible to predict which patients will develop potentially low HctCPB. 2. Autologous transfusions result in considerable reduction of banked blood usage.