Methods	A randomised controlled trial was conducted to evaluate the safety and effectiveness of the collection and re-transfusion of post-operatively shed mediastinal blood in patients undergoing cardiac surgery. Method of randomisation and allocation concealment were not described
Participants	113 consecutive adult patients undergoing cardiac surgical procedures requiring cardiopulmonary bypass were randomised to one of two groups: Group 1 (Autotransfusion group): n=54; M//F=48//6; mean (range) age = 55.9 (24-72) years Group 2 (Control group): n=59; M//F=55//4; mean (range) age = 54.8 (38-73) years
Interventions	Group 1: Autotransfusion group had their shed mediastinal blood collected postoperatively by an autotransfusion system (Sorenson). Suction was applied (−20cmH2O), allowing shed blood to flow into the upper bag of the system and then through two 170 micron filters into a lower 800ml collection bag. The lower bag was then disconnected from the system and its contents infused, the collected blood being transfused through an in-line 40 micron filter. No blood was allowed to remain in the system longer than 4 hours. Shed blood that was not utilised during this time period was discarded. When notable bleeding ceased (4-8 hours) retransfusion was no longer employed. Group 2: Control group received usual care without the use of cell salvage. NB: Intra-operative and post-operative haemodilution was performed in all patients but not equally distributed between groups
Outcomes	Outcomes reported: amount of blood collected by the cell saver, amount of blood retransfused from the cell saver, amount of allogeneic blood transfused, adverse events, myocardial infarction, mortality, post-operative infections, renal function impariment, fluid replacement, blood loss
Notes	Transfusion threshold: intra-operative blood replacement was left to the discretion of the staff surgeon and anaesthesiologist. In patients who were unstable haemodynamically and in those patients whom complete revascularisation was not possible the haematocrit was raised to 30% or higher
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method used to generate allocation sequences was not described
Allocation concealment (selection bias)	Unclear risk	Method used to conceal treatment allocation was unclear.
Blinding (performance bias and detection bias) All outcomes	High risk