Trade-Off between Lower or Higher Oxygen Saturations for Extremely Preterm Infants: The First Benefits of Oxygen Saturation Targeting (BOOST) II Trial Reports Its Primary Outcome

Abstract

See related article, p 30Administration of supplemental oxygen to very preterm infants is exacting because clinicians must strike a fine balance between the competing risks of insufficient oxygen delivery to the tissues and oxygen-induced damage to multiple organs including the eye, lung, and brain. The range of oxygen saturations that will best accomplish this goal is uncertain.1Tin W. Oxygen therapy: 50 years of uncertainty.Pediatrics. 2002; 110: 615-616Crossref PubMed Scopus (73) Google Scholar To address this knowledge gap, 5 randomized trials were initiated between 2005 and 2007 to compare oxygen saturation target ranges of 85%-89% vs 91%-95% by continuous pulse oximetry. These 2 narrow ranges were chosen because they were within the “pragmatically determined” and commonly accepted wider range of 85%-95%.2American Academy of Pediatrics and American College of Obstetricians and Gynecologists Guidelines for perinatal care, 6th ed. Clinical considerations in the use of oxygen. American Academy of Pediatrics, Elk Grove Village, IL2007: 259-262Google Scholar The different study teams discussed important aspects of the trial protocols such as target populations, study oximeters, saturation targets, sample sizes, and primary outcomes during several planning meetings and agreed to perform an individual patient data meta-analysis after the completion and publication of all 5 trials.3Askie L.M. Brocklehurst P. Darlow B.A. Finer N. Schmidt B. Tarnow-Mordi W. et al.NeOProM: Neonatal Oxygenation Prospective Meta-analysis Collaboration study protocol.BMC Pediatr. 2011; 11: 6PubMed Google Scholar However, each of the 5 trials was individually designed and managed.Until now, only 2 of the 5 trials—the Surfactant, Positive Pressure, and Pulse Oximetry Randomized Trial (SUPPORT) and the Canadian Oxygen Trial (COT)—have reported long-term outcomes of their study participants.4Vaucher Y.E. Peralta-Carcelen M. Finer N.N. Carlo W.A. Gantz M.G. Walsh M.C. et al.Neurodevelopmental outcomes in the early CPAP and pulse oximetry trial.N Engl J Med. 2012; 367: 2495-2504Crossref PubMed Scopus (139) Google Scholar, 5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar Neurodevelopmental disability was defined differently in these 2 trials; nevertheless, rates for the composite outcome of death or disability at 18 months did not differ between the 2 groups in either trial. The SUPPORT investigators reported increased mortality in infants assigned to the lower target range, and a greatly increased rate of severe retinopathy of prematurity (ROP) in infants assigned to the higher target range,6Finer N.N. Carlo W.A. Walsh M.C. Rich W. Gantz M.G. et al.SUPPORT Study Group of the Eunice Kennedy Shriver NICHD Neonatal Research NetworkTarget ranges of oxygen saturation in extremely preterm infants.N Engl J Med. 2010; 362: 1959-1969Crossref PubMed Scopus (724) Google Scholar whereas the COT investigators found similar rates for both of these outcomes in the 2 comparison groups.5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar These differences between the SUPPORT and COT results are perplexing because the distributions of true median oxygen saturations in the 2 treatment groups overlapped more in SUPPORT than in COT.5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar Consequently, we would expect to find smaller rather than greater differences between the 2 saturation target groups in SUPPORT than in COT.In this issue of The Journal, the first Benefits Of Oxygen Saturation Targeting (BOOST) II study group reports its primary outcome of death or major disability at age 2 years.7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar Consistent with SUPPORT and COT, Darlow et al7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar found no significant difference between the 2 groups for this composite outcome. As in COT, but in contrast to SUPPORT, the risks of mortality before the scheduled follow-up and of severe ROP were very similar in infants assigned to lower compared with higher oxygen saturation target ranges. SUPPORT and BOOST II New Zealand (NZ) are the only 2 of the 5 trials that did not replace the calibration software in their study oximeters.8Johnston E.D. Boyle B. Juszczak E. King A. Brocklehurst P. Stenson B.J. Oxygen targeting in preterm infants using the Masimo SET Radical pulse oximeter.Arch Dis Child Fetal Neonatal Ed. 2011; 96: F429-F433Crossref PubMed Scopus (65) Google Scholar Of note, mortality at follow-up in both study groups combined was similar in BOOST II NZ and in COT, 52 of 340 (15.3%) and 185 of 1162 (15.9%), respectively, and much lower than the combined mortality rate at discharge of 21.8% in BOOST II United Kingdom (UK).5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar, 7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar, 9Stenson B.J. Tarnow-Mordi W.O. Darlow B.A. et al.BOOST II United Kingdom Collaborative GroupBOOST II Australia Collaborative GroupBOOST II New Zealand Collaborative GroupOxygen saturation and outcomes in preterm infants.N Engl J Med. 2013; 368: 2094-2104Crossref PubMed Scopus (351) Google Scholar The relatively small sample size of 340 infants in BOOST II NZ is a limitation that has been acknowledged by the authors.7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar However, BOOST II NZ is the only member of the BOOST II family of trials that was not stopped early before enrolling its target sample size,9Stenson B.J. Tarnow-Mordi W.O. Darlow B.A. et al.BOOST II United Kingdom Collaborative GroupBOOST II Australia Collaborative GroupBOOST II New Zealand Collaborative GroupOxygen saturation and outcomes in preterm infants.N Engl J Med. 2013; 368: 2094-2104Crossref PubMed Scopus (351) Google Scholar, 10Stenson B. Brocklehurst P. Tarnow-Mordi W. U.K. BOOST II trialAustralian BOOST II trialNew Zealand BOOST II trialIncreased 36-week survival with high oxygen saturation target in extremely preterm infants.N Engl J Med. 2011; 364: 1680-1682Crossref PubMed Scopus (159) Google Scholar and this is a major strength. Although Darlow et al observed neither benefit nor harm in either arm of their own trial, they note that “a lower saturation target in the first few weeks of life may reduce the risk of ROP, but if it also increases mortality, this would not be acceptable.”7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google ScholarThe BOOST II NZ report follows on the heels of the first published meta-analysis of the 5 oxygen saturation targeting trials. Saugstad and Aune pooled data for the secondary neonatal outcome data from the 3 BOOST II trials with primary and secondary outcome data at or before a corrected age of 18 months from SUPPORT and COT.11Saugstad O.D. Aune D. Optimal oxygenation of extremely low birth weight infants: a meta-analysis and systematic review of the oxygen saturation target studies.Neonatology. 2014; 105: 55-63Crossref PubMed Scopus (201) Google Scholar There was no significant effect of saturation targeting on mortality when all 1281 SUPPORT participants with known vital status at 18 months were combined with the 1259 BOOST II and 549 COT participants who were studied with oximeters containing the original calibration software (relative risk 1.04; 95% CI, 0.88-1.22). In contrast, in the subgroup of 1182 BOOST II and 543 COT participants who were studied with the revised calibration software, the relative risk of mortality was 1.41 (95% CI, 1.14-1.74) in favor of the higher target range. In addition, for all study participants, the relative risk for necrotizing enterocolitis was significantly increased in infants assigned to lower as compared with higher saturation target ranges whereas the relative risk for severe ROP was significantly reduced. The authors concluded that “functional SpO2 should be targeted at 90%-95% in infants with gestational age <28 weeks until 36 weeks' postmenstrual age.”11Saugstad O.D. Aune D. Optimal oxygenation of extremely low birth weight infants: a meta-analysis and systematic review of the oxygen saturation target studies.Neonatology. 2014; 105: 55-63Crossref PubMed Scopus (201) Google ScholarBefore clinicians accept this interpretation of the trade-off between benefits and risks of lower vs higher target saturations, we need to explore more carefully the potential threats to the validity of the various mortality estimates reported to date and ask: How strong is the evidence that the revised oximeter calibration algorithm improved saturation targeting in BOOST II UK, Australia, and COT?; How appropriate was the interim subgroup analysis by calibration algorithm that led to the early stopping of enrollment in BOOST II for increased survival with the higher oxygen saturation target range?; and What are the implications of the early stopping of BOOST II UK and Australia on any pooled estimates of mortality in saturation targeting trials?Although the BOOST II investigators reported that the revised calibration software was associated with improved oxygen saturation targeting, there was little evidence in COT and only modest evidence in BOOST II to support this claim.5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar, 9Stenson B.J. Tarnow-Mordi W.O. Darlow B.A. et al.BOOST II United Kingdom Collaborative GroupBOOST II Australia Collaborative GroupBOOST II New Zealand Collaborative GroupOxygen saturation and outcomes in preterm infants.N Engl J Med. 2013; 368: 2094-2104Crossref PubMed Scopus (351) Google Scholar, 10Stenson B. Brocklehurst P. Tarnow-Mordi W. U.K. BOOST II trialAustralian BOOST II trialNew Zealand BOOST II trialIncreased 36-week survival with high oxygen saturation target in extremely preterm infants.N Engl J Med. 2011; 364: 1680-1682Crossref PubMed Scopus (159) Google Scholar Moreover, some study centers in the UK started recruitment later and never used the original calibration software. However, the comparison of distributions of median saturations by calibration algorithm was not restricted to those centers that enrolled infants both before and after the revision of the calibration algorithm.9Stenson B.J. Tarnow-Mordi W.O. Darlow B.A. et al.BOOST II United Kingdom Collaborative GroupBOOST II Australia Collaborative GroupBOOST II New Zealand Collaborative GroupOxygen saturation and outcomes in preterm infants.N Engl J Med. 2013; 368: 2094-2104Crossref PubMed Scopus (351) Google Scholar The belief of the BOOST II UK investigators in the importance of the calibration software version led them to request that the Data Safety Monitoring Boards conduct a pooled interim subgroup analysis by original vs revised calibration algorithm. Late in 2010, the UK investigators considered an extension of recruitment beyond the original target sample size to increase the number of babies who would be studied with the revised algorithm, and one possible consequence of examining the interim data would have been a judgment that trial recruitment would need to be extended. Thus, concerns about safety after the publication of the neonatal outcomes in SUPPORT were not the only reason for this investigator-initiated and unscheduled interim subgroup analysis.The interim BOOST II subgroup analysis showed a puzzling reversal of the direction of the observed treatment effect on mortality at a postmenstrual age of 36 weeks in favor of the higher saturation target range following installation of the revised calibration software.10Stenson B. Brocklehurst P. Tarnow-Mordi W. U.K. BOOST II trialAustralian BOOST II trialNew Zealand BOOST II trialIncreased 36-week survival with high oxygen saturation target in extremely preterm infants.N Engl J Med. 2011; 364: 1680-1682Crossref PubMed Scopus (159) Google Scholar The observed absolute mortality risk difference favoring the higher saturation range after the software change in BOOST II UK and Australia was 8.5%.10Stenson B. Brocklehurst P. Tarnow-Mordi W. U.K. BOOST II trialAustralian BOOST II trialNew Zealand BOOST II trialIncreased 36-week survival with high oxygen saturation target in extremely preterm infants.N Engl J Med. 2011; 364: 1680-1682Crossref PubMed Scopus (159) Google Scholar This is as large an effect on survival as we can expect from antenatal corticosteroids in threatened preterm birth or from animal-derived surfactant therapy in infants with respiratory distress syndrome.12Roberts D. Dalziel S. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth.Cochrane Database Syst Rev. 2006; 3: CD004454Crossref PubMed Scopus (67) Google Scholar, 13Seger N. Soll R. Animal derived surfactant extract for treatment of respiratory distress syndrome.Cochrane Database Syst Rev. 2009; 2: CD007836PubMed Google ScholarTrials that are stopped early for benefit overestimate the treatment effect.14Bassler D. Briel M. Montori V.M. Lane M. Glasziou P. Zhou Q. et al.Stopping randomized trials early for benefit and estimation of treatment effects: systematic review and meta-regression analysis.JAMA. 2010; 303: 1180-1187Crossref PubMed Scopus (439) Google Scholar However, contingent on a number of conditions including the presence of appropriate stopping rules, the early stopping of clinical trials is not a major source of bias in systematic reviews and meta-analyses.15Bassler D. Montori V.M. Briel M. Glasziou P. Walter S.D. Ramsay T. et al.Reflections on meta-analyses involving trials stopped early for benefit: is there a problem and if so, what is it?.Stat Methods Med Res. 2013; 22: 159-168Crossref PubMed Scopus (35) Google Scholar, 16Schou I.M. Marschner I.C. Meta-analysis of clinical trials with early stopping: an investigation of potential bias.Stat Med. 2013; 32: 4859-4874Crossref PubMed Scopus (28) Google Scholar Consequently, the debate about the impact of the truncated BOOST II UK and Australia trials on all pooled analyses of mortality in trials of oxygen saturation targeting will largely focus on the question whether an ad-hoc and investigator-initiated interim subgroup analysis can be considered “appropriate.”Pending the resolution of this debate, what should be current evidence-based guidelines for oxygen therapy in extremely preterm infants? We strongly recommend the prescription of oximeter alarm settings instead of target ranges, and the enforcement of adherence with those alarm settings through frequent audits and feedbacks. Oximeter alarm settings were not consistently mandated or monitored in the 5 trials. The COT protocol defined and prescribed alarms and provided monthly feedback to the clinical centers on their ability to maintain the displayed saturations within the study alarm limits during oxygen therapy,5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar whereas in BOOST II NZ alarm settings were recommended but not mandated.7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar A low alarm setting was not standardized at all in the BOOST II UK protocol.17Brocklehurst P. Boost-II UK: Benefits of oxygen saturation targeting. Protocol. Which oxygen saturation level should we use for very premature infants? A randomised controlled trial. Version 3. University of Oxford, 2011 Available at: https://wwwnpeuoxacuk/downloads/files/boost/BOOSTII-Protocol-Version-3-Nov-2011pdf. Accessed December 29, 2012.Google ScholarThe trade-off between the potential benefits and risks of lower vs higher saturations may not be the same in each nursery. For example, hospitals with low rates of mortality and necrotizing enterocolitis but high rates of severe ROP may choose lower alarms between 85% and 88%, and upper alarms between 93% and 94% while in supplemental oxygen. Conversely, hospitals with high rates of mortality and necrotizing enterocolitis but low rates of severe ROP may be compelled to set their lower alarms at 89% or 90% and their upper alarms at 95%. As we try to strike the right balance for our patients between oxygen deprivation and oxygen toxicity, we must remember that severe ROP remains an adverse outcome of neonatal intensive care with poor prognosis for child development.18Schmidt B. Davis P.G. Asztalos E.V. Solimano A. Roberts R.S. Association between severe retinopathy of prematurity and non-visual disabilities at age 5 years: results from the Caffeine for Apnea of Prematurity Trial.JAMA. 2014; 311: 523-525Crossref PubMed Scopus (48) Google Scholar See related article, p 30Administration of supplemental oxygen to very preterm infants is exacting because clinicians must strike a fine balance between the competing risks of insufficient oxygen delivery to the tissues and oxygen-induced damage to multiple organs including the eye, lung, and brain. The range of oxygen saturations that will best accomplish this goal is uncertain.1Tin W. Oxygen therapy: 50 years of uncertainty.Pediatrics. 2002; 110: 615-616Crossref PubMed Scopus (73) Google Scholar To address this knowledge gap, 5 randomized trials were initiated between 2005 and 2007 to compare oxygen saturation target ranges of 85%-89% vs 91%-95% by continuous pulse oximetry. These 2 narrow ranges were chosen because they were within the “pragmatically determined” and commonly accepted wider range of 85%-95%.2American Academy of Pediatrics and American College of Obstetricians and Gynecologists Guidelines for perinatal care, 6th ed. Clinical considerations in the use of oxygen. American Academy of Pediatrics, Elk Grove Village, IL2007: 259-262Google Scholar The different study teams discussed important aspects of the trial protocols such as target populations, study oximeters, saturation targets, sample sizes, and primary outcomes during several planning meetings and agreed to perform an individual patient data meta-analysis after the completion and publication of all 5 trials.3Askie L.M. Brocklehurst P. Darlow B.A. Finer N. Schmidt B. Tarnow-Mordi W. et al.NeOProM: Neonatal Oxygenation Prospective Meta-analysis Collaboration study protocol.BMC Pediatr. 2011; 11: 6PubMed Google Scholar However, each of the 5 trials was individually designed and managed. See related article, p 30 See related article, p 30 Until now, only 2 of the 5 trials—the Surfactant, Positive Pressure, and Pulse Oximetry Randomized Trial (SUPPORT) and the Canadian Oxygen Trial (COT)—have reported long-term outcomes of their study participants.4Vaucher Y.E. Peralta-Carcelen M. Finer N.N. Carlo W.A. Gantz M.G. Walsh M.C. et al.Neurodevelopmental outcomes in the early CPAP and pulse oximetry trial.N Engl J Med. 2012; 367: 2495-2504Crossref PubMed Scopus (139) Google Scholar, 5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar Neurodevelopmental disability was defined differently in these 2 trials; nevertheless, rates for the composite outcome of death or disability at 18 months did not differ between the 2 groups in either trial. The SUPPORT investigators reported increased mortality in infants assigned to the lower target range, and a greatly increased rate of severe retinopathy of prematurity (ROP) in infants assigned to the higher target range,6Finer N.N. Carlo W.A. Walsh M.C. Rich W. Gantz M.G. et al.SUPPORT Study Group of the Eunice Kennedy Shriver NICHD Neonatal Research NetworkTarget ranges of oxygen saturation in extremely preterm infants.N Engl J Med. 2010; 362: 1959-1969Crossref PubMed Scopus (724) Google Scholar whereas the COT investigators found similar rates for both of these outcomes in the 2 comparison groups.5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar These differences between the SUPPORT and COT results are perplexing because the distributions of true median oxygen saturations in the 2 treatment groups overlapped more in SUPPORT than in COT.5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar Consequently, we would expect to find smaller rather than greater differences between the 2 saturation target groups in SUPPORT than in COT. In this issue of The Journal, the first Benefits Of Oxygen Saturation Targeting (BOOST) II study group reports its primary outcome of death or major disability at age 2 years.7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar Consistent with SUPPORT and COT, Darlow et al7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar found no significant difference between the 2 groups for this composite outcome. As in COT, but in contrast to SUPPORT, the risks of mortality before the scheduled follow-up and of severe ROP were very similar in infants assigned to lower compared with higher oxygen saturation target ranges. SUPPORT and BOOST II New Zealand (NZ) are the only 2 of the 5 trials that did not replace the calibration software in their study oximeters.8Johnston E.D. Boyle B. Juszczak E. King A. Brocklehurst P. Stenson B.J. Oxygen targeting in preterm infants using the Masimo SET Radical pulse oximeter.Arch Dis Child Fetal Neonatal Ed. 2011; 96: F429-F433Crossref PubMed Scopus (65) Google Scholar Of note, mortality at follow-up in both study groups combined was similar in BOOST II NZ and in COT, 52 of 340 (15.3%) and 185 of 1162 (15.9%), respectively, and much lower than the combined mortality rate at discharge of 21.8% in BOOST II United Kingdom (UK).5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar, 7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar, 9Stenson B.J. Tarnow-Mordi W.O. Darlow B.A. et al.BOOST II United Kingdom Collaborative GroupBOOST II Australia Collaborative GroupBOOST II New Zealand Collaborative GroupOxygen saturation and outcomes in preterm infants.N Engl J Med. 2013; 368: 2094-2104Crossref PubMed Scopus (351) Google Scholar The relatively small sample size of 340 infants in BOOST II NZ is a limitation that has been acknowledged by the authors.7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar However, BOOST II NZ is the only member of the BOOST II family of trials that was not stopped early before enrolling its target sample size,9Stenson B.J. Tarnow-Mordi W.O. Darlow B.A. et al.BOOST II United Kingdom Collaborative GroupBOOST II Australia Collaborative GroupBOOST II New Zealand Collaborative GroupOxygen saturation and outcomes in preterm infants.N Engl J Med. 2013; 368: 2094-2104Crossref PubMed Scopus (351) Google Scholar, 10Stenson B. Brocklehurst P. Tarnow-Mordi W. U.K. BOOST II trialAustralian BOOST II trialNew Zealand BOOST II trialIncreased 36-week survival with high oxygen saturation target in extremely preterm infants.N Engl J Med. 2011; 364: 1680-1682Crossref PubMed Scopus (159) Google Scholar and this is a major strength. Although Darlow et al observed neither benefit nor harm in either arm of their own trial, they note that “a lower saturation target in the first few weeks of life may reduce the risk of ROP, but if it also increases mortality, this would not be acceptable.”7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar The BOOST II NZ report follows on the heels of the first published meta-analysis of the 5 oxygen saturation targeting trials. Saugstad and Aune pooled data for the secondary neonatal outcome data from the 3 BOOST II trials with primary and secondary outcome data at or before a corrected age of 18 months from SUPPORT and COT.11Saugstad O.D. Aune D. Optimal oxygenation of extremely low birth weight infants: a meta-analysis and systematic review of the oxygen saturation target studies.Neonatology. 2014; 105: 55-63Crossref PubMed Scopus (201) Google Scholar There was no significant effect of saturation targeting on mortality when all 1281 SUPPORT participants with known vital status at 18 months were combined with the 1259 BOOST II and 549 COT participants who were studied with oximeters containing the original calibration software (relative risk 1.04; 95% CI, 0.88-1.22). In contrast, in the subgroup of 1182 BOOST II and 543 COT participants who were studied with the revised calibration software, the relative risk of mortality was 1.41 (95% CI, 1.14-1.74) in favor of the higher target range. In addition, for all study participants, the relative risk for necrotizing enterocolitis was significantly increased in infants assigned to lower as compared with higher saturation target ranges whereas the relative risk for severe ROP was significantly reduced. The authors concluded that “functional SpO2 should be targeted at 90%-95% in infants with gestational age <28 weeks until 36 weeks' postmenstrual age.”11Saugstad O.D. Aune D. Optimal oxygenation of extremely low birth weight infants: a meta-analysis and systematic review of the oxygen saturation target studies.Neonatology. 2014; 105: 55-63Crossref PubMed Scopus (201) Google Scholar Before clinicians accept this interpretation of the trade-off between benefits and risks of lower vs higher target saturations, we need to explore more carefully the potential threats to the validity of the various mortality estimates reported to date and ask: How strong is the evidence that the revised oximeter calibration algorithm improved saturation targeting in BOOST II UK, Australia, and COT?; How appropriate was the interim subgroup analysis by calibration algorithm that led to the early stopping of enrollment in BOOST II for increased survival with the higher oxygen saturation target range?; and What are the implications of the early stopping of BOOST II UK and Australia on any pooled estimates of mortality in saturation targeting trials? Although the BOOST II investigators reported that the revised calibration software was associated with improved oxygen saturation targeting, there was little evidence in COT and only modest evidence in BOOST II to support this claim.5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar, 9Stenson B.J. Tarnow-Mordi W.O. Darlow B.A. et al.BOOST II United Kingdom Collaborative GroupBOOST II Australia Collaborative GroupBOOST II New Zealand Collaborative GroupOxygen saturation and outcomes in preterm infants.N Engl J Med. 2013; 368: 2094-2104Crossref PubMed Scopus (351) Google Scholar, 10Stenson B. Brocklehurst P. Tarnow-Mordi W. U.K. BOOST II trialAustralian BOOST II trialNew Zealand BOOST II trialIncreased 36-week survival with high oxygen saturation target in extremely preterm infants.N Engl J Med. 2011; 364: 1680-1682Crossref PubMed Scopus (159) Google Scholar Moreover, some study centers in the UK started recruitment later and never used the original calibration software. However, the comparison of distributions of median saturations by calibration algorithm was not restricted to those centers that enrolled infants both before and after the revision of the calibration algorithm.9Stenson B.J. Tarnow-Mordi W.O. Darlow B.A. et al.BOOST II United Kingdom Collaborative GroupBOOST II Australia Collaborative GroupBOOST II New Zealand Collaborative GroupOxygen saturation and outcomes in preterm infants.N Engl J Med. 2013; 368: 2094-2104Crossref PubMed Scopus (351) Google Scholar The belief of the BOOST II UK investigators in the importance of the calibration software version led them to request that the Data Safety Monitoring Boards conduct a pooled interim subgroup analysis by original vs revised calibration algorithm. Late in 2010, the UK investigators considered an extension of recruitment beyond the original target sample size to increase the number of babies who would be studied with the revised algorithm, and one possible consequence of examining the interim data would have been a judgment that trial recruitment would need to be extended. Thus, concerns about safety after the publication of the neonatal outcomes in SUPPORT were not the only reason for this investigator-initiated and unscheduled interim subgroup analysis. The interim BOOST II subgroup analysis showed a puzzling reversal of the direction of the observed treatment effect on mortality at a postmenstrual age of 36 weeks in favor of the higher saturation target range following installation of the revised calibration software.10Stenson B. Brocklehurst P. Tarnow-Mordi W. U.K. BOOST II trialAustralian BOOST II trialNew Zealand BOOST II trialIncreased 36-week survival with high oxygen saturation target in extremely preterm infants.N Engl J Med. 2011; 364: 1680-1682Crossref PubMed Scopus (159) Google Scholar The observed absolute mortality risk difference favoring the higher saturation range after the software change in BOOST II UK and Australia was 8.5%.10Stenson B. Brocklehurst P. Tarnow-Mordi W. U.K. BOOST II trialAustralian BOOST II trialNew Zealand BOOST II trialIncreased 36-week survival with high oxygen saturation target in extremely preterm infants.N Engl J Med. 2011; 364: 1680-1682Crossref PubMed Scopus (159) Google Scholar This is as large an effect on survival as we can expect from antenatal corticosteroids in threatened preterm birth or from animal-derived surfactant therapy in infants with respiratory distress syndrome.12Roberts D. Dalziel S. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth.Cochrane Database Syst Rev. 2006; 3: CD004454Crossref PubMed Scopus (67) Google Scholar, 13Seger N. Soll R. Animal derived surfactant extract for treatment of respiratory distress syndrome.Cochrane Database Syst Rev. 2009; 2: CD007836PubMed Google Scholar Trials that are stopped early for benefit overestimate the treatment effect.14Bassler D. Briel M. Montori V.M. Lane M. Glasziou P. Zhou Q. et al.Stopping randomized trials early for benefit and estimation of treatment effects: systematic review and meta-regression analysis.JAMA. 2010; 303: 1180-1187Crossref PubMed Scopus (439) Google Scholar However, contingent on a number of conditions including the presence of appropriate stopping rules, the early stopping of clinical trials is not a major source of bias in systematic reviews and meta-analyses.15Bassler D. Montori V.M. Briel M. Glasziou P. Walter S.D. Ramsay T. et al.Reflections on meta-analyses involving trials stopped early for benefit: is there a problem and if so, what is it?.Stat Methods Med Res. 2013; 22: 159-168Crossref PubMed Scopus (35) Google Scholar, 16Schou I.M. Marschner I.C. Meta-analysis of clinical trials with early stopping: an investigation of potential bias.Stat Med. 2013; 32: 4859-4874Crossref PubMed Scopus (28) Google Scholar Consequently, the debate about the impact of the truncated BOOST II UK and Australia trials on all pooled analyses of mortality in trials of oxygen saturation targeting will largely focus on the question whether an ad-hoc and investigator-initiated interim subgroup analysis can be considered “appropriate.” Pending the resolution of this debate, what should be current evidence-based guidelines for oxygen therapy in extremely preterm infants? We strongly recommend the prescription of oximeter alarm settings instead of target ranges, and the enforcement of adherence with those alarm settings through frequent audits and feedbacks. Oximeter alarm settings were not consistently mandated or monitored in the 5 trials. The COT protocol defined and prescribed alarms and provided monthly feedback to the clinical centers on their ability to maintain the displayed saturations within the study alarm limits during oxygen therapy,5Schmidt B. Whyte R.K. Asztalos E.V. Moddemann D. Poets C. Rabi Y. et al.Effects of targeting higher vs lower arterial oxygen saturations on death or disability in extremely preterm infants: a randomized clinical trial.JAMA. 2013; 309: 2111-2120Crossref PubMed Scopus (293) Google Scholar whereas in BOOST II NZ alarm settings were recommended but not mandated.7Darlow B.A. Marschner S.L. Donoghoe M. Battin M.R. Broadbent R.S. Elder M.J. et al.Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes.J Pediatr. 2014; 165: 30-35Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar A low alarm setting was not standardized at all in the BOOST II UK protocol.17Brocklehurst P. Boost-II UK: Benefits of oxygen saturation targeting. Protocol. Which oxygen saturation level should we use for very premature infants? A randomised controlled trial. Version 3. University of Oxford, 2011 Available at: https://wwwnpeuoxacuk/downloads/files/boost/BOOSTII-Protocol-Version-3-Nov-2011pdf. Accessed December 29, 2012.Google Scholar The trade-off between the potential benefits and risks of lower vs higher saturations may not be the same in each nursery. For example, hospitals with low rates of mortality and necrotizing enterocolitis but high rates of severe ROP may choose lower alarms between 85% and 88%, and upper alarms between 93% and 94% while in supplemental oxygen. Conversely, hospitals with high rates of mortality and necrotizing enterocolitis but low rates of severe ROP may be compelled to set their lower alarms at 89% or 90% and their upper alarms at 95%. As we try to strike the right balance for our patients between oxygen deprivation and oxygen toxicity, we must remember that severe ROP remains an adverse outcome of neonatal intensive care with poor prognosis for child development.18Schmidt B. Davis P.G. Asztalos E.V. Solimano A. Roberts R.S. Association between severe retinopathy of prematurity and non-visual disabilities at age 5 years: results from the Caffeine for Apnea of Prematurity Trial.JAMA. 2014; 311: 523-525Crossref PubMed Scopus (48) Google Scholar Randomized Controlled Trial of Oxygen Saturation Targets in Very Preterm Infants: Two Year OutcomesThe Journal of PediatricsVol. 165Issue 1PreviewTo assess whether an oxygen saturation (Spo2) target of 85%-89% compared with 91%-95% reduced the incidence of the composite outcome of death or major disability at 2 years of age in infants born at <28 weeks' gestation. Full-Text PDF