Maintaining adequate oxygen delivery during cardiopulmonary bypass (CPB)-assisted cardiac surgery is crucial, but hyperoxia has been suggested to cause organ injury. We compared the effects of restrictive vs liberal oxygenation during CPB and weaning from CPB on clinical outcomes in cardiac surgery.

We conducted a single-centre, patient- and assessor-blinded randomised trial on adults undergoing CPB-assisted coronary artery bypass grafting, aortic valve replacement, or both. Participants were randomly assigned (1:1) to restrictive (Fio₂=50%) or liberal (Fio₂=100%) oxygen therapy during and for the first hour after weaning from CPB. The primary composite outcome was the time to death, stroke, renal failure requiring dialysis, or new-onset or worsening heart failure during follow-up.

Among 1389 participants (mean age, 67 yr [range, 29–85 yr]; 17% female), randomisation to receive Fio₂ 50% resulted in median Pao₂ levels of 19–23 kPa during CPB, compared with >60 kPa in participants receiving Fio₂ 100%. During a median follow-up period of 5.9 yr (interquartile range, 2.5–8.3), 167/695 (24%) participants in the restrictive oxygenation group and 168/694 (24%) participants in the liberal oxygenation group met the primary endpoint (hazard ratio, 1.01 [95% confidence interval, 0.8–1.3]; P=0.92). There was no difference in adverse event rates between restrictive and liberal oxygen therapy.

Among patients undergoing elective or urgent CPB-assisted coronary artery bypass grafting, aortic valve replacement, or both, no significant differences were observed in mortality, dialysis-dependent renal failure, stroke, or new-onset or worsening heart failure between a restrictive oxygenation strategy (Fio₂ 50%) and a liberal oxygenation strategy (Fio₂ 100%) during CPB and the subsequent weaning period.