Methods	A randomised controlled trial was conducted to assess the effectiveness of two mechanical methods of blood conservation in reducing the need for allogeneic red blood cells or coagulation products during cardiac surgery. Patient allocations were generated from random number tables by an independent observer and concealed in sealed opaque envelopes
Participants	256 patients undergoing elective coronary artery bypass surgery were randomly allocated to one of three groups: Group 1 (Autotransfusion group): n=84; M//F=75//9; mean (sd) age = 63.8 (7.8) years Group 2 (Combined intervention group): n=84; M//F=74//10; mean (sd) age = 63. 1 (8.2) years Group 3 (Control group): n=84; M//F=74//10; mean (sd) age = 63.4 (9.1) years
Interventions	Group 1: during surgery blood from the operation site was collected in a storage system. At the termination of cardiopulmonary bypass, blood remaining in the bypass circuit was added to the storage system. This blood was then centrifuged using a cell salvage system (Dideco Compact cell saver) leaving a concentrated solution of red blood cells with a haematocrit of 50-60%. This autologous blood was then retransfused to the patient during the intra-operative period. Group 2: were treated with cell salvage and acute normovolaemic haemodilution (ANH). After induction of anaesthesia 10ml/kg of blood was removed from a central venous line while being replaced at the same time with an equivalent volume of modified gelatin (Gelofusine). The Harvest Blood Stream Recovery System, an autologous recovery system, was used to remove blood. The recovered blood was stored at room temperature. Group 3: were treated without the use of cell salvage or acute normovolaemic haemodilution (ANH).
Outcomes	Outcomes reported: number of patients transfused allogeneic blood, number of patients receiving any blood product, amount of allogeneic blood transfused, blood loss, reoperation for bleeding, hospital length of stay, infection, stroke, renal failure, myocardial infarction
Notes	Transfusion threshold: allogeneic red blood cells were transfused in all groups when the haemoglobin level fell below 9.0g/dL
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Patient allocations were generated from random number tables by an independent observer
Allocation concealment (selection bias)	High risk	Allocation concealment was inadequate.
Blinding (performance bias and detection bias) All outcomes	High risk