William Tarnow-Mordi, Ben Stenson, Adrienne Kirby, Edmund Juszczak, Mark Donoghoe, Sanjeev Deshpande, Colin Morley, Andrew King, Lex W Doyle, Brian W Fleck, Peter G Davis, Henry L Halliday, Wendy Hague, Pamela Cairns, Brian A Darlow, Alistair R Fielder, Val Gebski, Neil Marlow, Karen Simmer, Win Tin, Alpana Ghadge, Cathy Williams, Anthony Keech, Stephen P Wardle, Zsuzsoka Kecskes, Martin Kluckow, Glen Gole, Nicholas Evans, Girvan Malcolm, Melissa Luig, Ian Wright, Jacqueline Stack, Kenneth Tan, Margo Pritchard, Peter H Gray, Scott Morris, Bevan Headley, Peter Dargaville, R John Simes, Peter Brocklehurst

Background The safest ranges of oxygen saturation in preterm infants have been the subject of debate. Methods In two trials, conducted in Australia and the United Kingdom, infants born before 28 weeks' gestation were randomly assigned to either a lower (85 to 89%) or a higher (91 to 95%) oxygen-saturation range. During enrollment, the oximeters were revised to correct a calibration-algorithm artifact. The primary outcome was death or disability at a corrected gestational age of 2 years; this outcome was evaluated among infants whose oxygen saturation was measured with any study oximeter in the Australian trial and those whose oxygen saturation was measured with a revised oximeter in the U.K. trial. Results After 1135 infants in Australia and 973 infants in the United Kingdom had been enrolled in the trial, an interim analysis showed increased mortality at a corrected gestational age of 36 weeks, and enrollment was stopped. Death or disability in the Australian trial (with all oximeters included) occurred in 247 of 549 infants (45.0%) in the lower-target group versus 217 of 545 infants (39.8%) in the higher-target group (adjusted relative risk, 1.12; 95% confidence interval [CI], 0.98 to 1.27; P=0.10); death or disability in the U.K. trial (with only revised oximeters included) occurred in 185 of 366 infants (50.5%) in the lower-target group versus 164 of 357 infants (45.9%) in the higher-target group (adjusted relative risk, 1.10; 95% CI, 0.97 to 1.24; P=0.15). In post hoc combined, unadjusted analyses that included all oximeters, death or disability occurred in 492 of 1022 infants (48.1%) in the lower-target group versus 437 of 1013 infants (43.1%) in the higher-target group (relative risk, 1.11; 95% CI, 1.01 to 1.23; P=0.02), and death occurred in 222 of 1045 infants (21.2%) in the lower-target group versus 185 of 1045 infants (17.7%) in the higher-target group (relative risk, 1.20; 95% CI, 1.01 to 1.43; P=0.04). In the group in which revised oximeters were used, death or disability occurred in 287 of 580 infants (49.5%) in the lower-target group versus 248 of 563 infants (44.0%) in the higher-target group (relative risk, 1.12; 95% CI, 0.99 to 1.27; P=0.07), and death occurred in 144 of 587 infants (24.5%) versus 99 of 586 infants (16.9%) (relative risk, 1.45; 95% CI, 1.16 to 1.82; P=0.001). Conclusions Use of an oxygen-saturation target range of 85 to 89% versus 91 to 95% resulted in nonsignificantly higher rates of death or disability at 2 years in each trial but in significantly increased risks of this combined outcome and of death alone in post hoc combined analyses. (Funded by the Australian National Health and Medical Research Council and others; BOOST-II Current Controlled Trials number, ISRCTN00842661 , and Australian New Zealand Clinical Trials Registry number, ACTRN12605000055606 .).