Tarnow-mordi W, Stenson B, Kirby A, et al. Outcomes of Two Trials of Oxygen-Saturation Targets in Preterm Infants. N Engl J Med. 2016;374(8):749-60. PMID: 26863265
Emöke Deschmann M.D. M.M.Sc.
Karolinska University Hospital, Division of Neonatology
Mikael Norman M.D. PhD
Professor of Pediatrics and Neonatology
Karolinska University Hospital, Division of Neonatology
TYPE OF INVESTIGATION
P: in infants born <28 weeks’ gestation
I: is lower (85 to 89%) oxygen saturation target range superior
C: compared with higher (91 to 95%) saturation target range
O: with primary outcome of death or disability
T: at corrected age of 2 years
- Design: Multicenter randomized controlled trials – 2 trials: Australian (15 centers) and U.K. trial (34 centers)
- Allocation: Randomization was performed centrally by computer and was performed separately for each trial (Australian and U.K. trial). Infants from multiple births underwent randomization individually. Minimization procedures were used to balance group assignments according to sex, gestational age, and center, additionally in the Australian trial, according to whether the infant was a singleton or part of a multiple birth and whether the infant was enrolled in the hospital of birth.
- Blinding: All clinicians, outcome assessors and analysts were unaware of the group assignments. The study oximeters were modified so that the observers were unaware of the true reading. The clinical staff targeted displayed readings of 88 to 92% to achieve the intended target ranges. Outcomes were evaluated without knowledge of treatment-group assignments.
- Follow-up period: Up to a corrected age of 2 years.
- Setting: Two trials: (1) the Australian trial involved 15 centers, was started on March 25, 2006; (2) the U.K. trial involved 34 centers, and was started on September 29, 2007. Both trials closed recruitment on December 24, 2010.
***After the trials were initiated, the U.K. trial investigators found that the pulse oximeters had an artifact in their calibration algorithm: approximately a third fewer oxygen-saturation values between 87 and 90% were displayed than expected, and values above 87% were shifted up by 1 to 2 % points. After that, revised oximeters were used.
- Patients: Infants were eligible if they were born <28 weeks’ gestation, and were <24 hours’ age. Infants who had major congenital abnormalities or were unlikely to survive or to be followed up were excluded.
- Intervention: The study oximeters were modified so that the observers were unaware of the true reading: readings in the range of 85 to 95% displayed an oxygen saturation that was up to 3% points higher than the actual oxygen saturation in the lower-target group and lower than actual oxygen saturation in the higher-target group. The clinical staff targeted displayed readings of 88 to 92% to achieve the intended target ranges. At readings above 95% and below 85%, the oximeters reverted to the true reading. Only study oximeters were used until 36 weeks’ postmenstrual age. If infants were in stable condition while breathing in room air before then, oximetry was discontinued. If oximetry was resumed before 36 weeks, a study oximeter was used.
- Primary outcome: death or disability (serious neurosensory disability in the U.K. trial or major disability in the Australian trial) at corrected age of 2 years. Disability was defined as a score of <85 on the Bayley-III cognitive or language assessment, as being legally blind or partially sighted in the U.K. trial or legally blind with less than 6/60 vision in the better eye in the Australian trial, or as having a severe cerebral palsy (GMCSF ≥2 or not walking without aid at 2 years)
- Secondary outcomes: death, a Bayley-III cognitive or language score of <85, blindness, cerebral palsy, deafness, developmental delay of 12 months or more on a pediatric assessment, and in the Australian trial, late-onset infection, respiratory illness, death attributed to pulmonary causes, days of endotracheal intubation, days on CPAP, days of supplemental oxygen in the hospital and at home, and the number of hospital readmissions at corrected 2 years of age. Treatment for severe ROP, use of oxygen at 36 weeks’ postmenstrual age, PDA, NEC resulting in surgery or death, IVH gr III or IV, and brain injury – which were previously reported.
- Analysis and Sample Size: Infants who were randomly assigned to the higher oxygen-saturation target range were expected to have a 30 to 40% incidence of the primary outcome. Assuming an incidence of 35%, it was calculated that each trial would need to enroll 1200 infants for the study to have 80% power to detect an 8% absolute difference between the groups, at a two-sided 5% level of significance. Data were analyzed according to “intention-to-treat”, i.e., assignment. In the Australian trial, the primary analysis population comprised all enrolled infants, however, in the U.K. trial, the primary analysis population comprised infants whose oxygen-saturation levels were evaluated with the revised oximeters.
- Patient follow-up: 1,135 infants were enrolled in the Australian trial, and 973 in the U.K. trial. The primary outcome was determined for 1,094 of 1,135 infants (96.4%) in the primary analysis population of the Australian trial, and 723 of 745 infants (97%) in the primary analysis of the U.K. trial (revised oximeter patient group). The primary outcome in each trial was death or disability at a corrected gestational age of 2 years.
Table 1. Baseline characteristics of the primary analysis population:
|Australian trial (all oximeters)||U.K. trial (revised oximeters)|
|Male sex %||51.6||52.2||52.5||53.5|
|Birth weight g||817±177||833±190||821±182||818±189|
|Gestational age wk||26±1.2||26±1.2||26±1.3||26±1.3|
|Multiple birth %||24.3||23.8||28.4||29.4|
|Born outside of study center %||7.7||7.4||12.6||11.2|
|Antenatal steroid %|
|Birth by cesarian section %||51.9||54.4||42.6||40.3|
|Temp. at admission °C||36.0±1.1||36.1±0.9||36.6±0.9||36.7±0.9|
Data is presented as means±SD, * p<0.05
The baseline characteristics were similar in the two treatment groups in each trial, with the exception of a lack of antenatal steroids, which was more common among infants in the lower-target group than in the higher-target group in the Australian trial (P = 0.03). The relative risks of the primary outcome in the lower-target group versus the higher-target group were unchanged after adjustment for use of antenatal glucocorticoids.
Relative Risk (95%CI)
|Death or disability
|492/1022 (48.1)||437/1013 (43.1)||1.11 (1.01–1.23)||0.02|
|144/587 (24.5)||99/586 (16.9)||1.45 (1.16–1.82)||0.001|
In combined unadjusted analyses of pooled data (both trials), death or disability occurred more frequently in the lower-target group than in the higher-target group among all infants, without evidence of heterogeneity according to oximeter calibration algorithm. Mortality before 2 years was higher in the lower-target group than in the higher-target group among all infants (with both oximeters), however, there was evidence of heterogeneity between the original and the revised oximeter group. There were no significant differences between the groups in either trial in the rates of disability. There were no significant differences between the groups in the secondary outcomes (either trial): Bayley-III cognitive or language score <85, blindness, cerebral palsy, deafness, developmental delay of 12 months or more, late/onset infection, respiratory illness, days of endotracheal intubation, days on CPAP, days of treatment with supplemental oxygen, and number of hospital readmissions.
In the primary and individual analyses of the Australian and the U.K. trials, there were no significant differences between the study groups in the rates of death or disability. In contrast, the post hoc combined, unadjusted analyses of the two trials demonstrated that the mortality or disability rate was significantly higher in the lower-target group (Table 2). Looking at data from revised oximeters only, the relative risk of death was found to be 45% higher in the lower oxygen-saturation target group (corresponding to an absolute difference of 76 infants per 1,000 liveborns) than in the higher-target group. The higher mortality among infants with the lower oxygen-saturation target was also observed when data from all oximeters were used. There was no difference between the groups in the rates of disability at 2 years, neither in the individual trials or in the combined analyses. In the combined analysis, more infants in the higher-target group were treated for ROP, without any between-group difference in the rate of blindness.
Oxygen saturation target range in extremely preterm infants is an important question of the daily practice in the NICUs, since important complications such as ROP and BPD can be associated with oxygen exposure outside physiological boundaries. Observational data1 suggested that saturation below 90% was associated with lower rate of ROP without increased rate of CP or death. These data led to change in practice targeting lower oxygen saturation in the NICUs without strong evidence of its safety and efficacy from large RCTs. To answer the question if a lower saturation target range (85-89%) is superior to a higher saturation target range (91-95%) on an outcome of death or major disability at 18 to 24 months corrected age, five comparative effectiveness trials of the targeting of oxygen saturation in infants <28 weeks’ gestation (the Neonatal Oxygen Prospective Meta-analysis (NeOProM) Collaboration) were designed. To facilitate patient data meta-analysis, the trials were designed in a fairly similar fashion and in total, almost 5,000 preterm infants less than 28 weeks of gestation were enrolled. Two of these trials are known as BOOST-II Australian and BOOST-II U.K. trial. This report presented the outcomes of the Australian and U.K. BOOST-II trials in children up to a corrected age of 2 years.
As previously reported,2 the NeOProM trials encountered several problems: 1. An oxygen saturation calibration problem was found in the pulse oximeters used in two of the NeOProM trials: COT and the U.K. BOOST-II: approximately a third fewer oxygen-saturation values between 87 and 90% were displayed than expected, and values above 87% were shifted up by 1-2% points; 2. Additionally, when the SUPPORT trial found that targeting the lower oxygen saturation range resulted in significantly lower rate of ROP, but higher mortality at discharge compared with the higher oxygen saturation range,3 the BOOST-II trial was stopped early, because the safety monitoring committees found that the pooled safety analysis of the two trials showed similar results to SUPPORT at 36 weeks’ postmenstrual age; 3. None of the NeOProM trials manage to keep the infants’ oxygen saturation in the target range: all median oxygen saturations were higher (>89%) than the predetermined in the lower saturation target group and there was also significant overlap in oxygen saturations achieved.
This meta-analysis of the two BOOST-II trials4 confirms the findings of the SUPPORT, that targeting lower oxygen saturation does not increase major disability, however increases the rate of death. The authors conclude that at present, the most rigorously evaluated evidence for policy is that targeting an oxygen saturation of 91 to 95% is safer than targeting an oxygen saturation of 85 to 89%. However, controversy still remains around dynamic saturation target ranges and alarm settings2 over the neonatal period. Tissue oxygenation, which depends on multiple factors, including oxygen saturation, is affected by fetal hemoglobin concentration and blood hemoglobin concentration, along with several other factors.
Although the quality of the existing evidence has been graded as moderate to low,5 we think that as for now and until more sophisticated methods and individual patient data are available, targeting oxygen saturation between 91% and 95% is a reasonable initial approach for extremely preterm infants. Still, in units with low rates of mortality and but high rates of severe ROP, lower saturation targets for supplemental oxygen treatment has been suggested.3
- Tin W, Milligan DW, Pennefather P, Hey E. Pulse oximetry, severe retinopathy, and outcome at one year in babies of less than 28 weeks gestation. Archives of disease in childhood Fetal and neonatal edition 2001;84:F106-10.
- Schmidt B, Whyte RK, Roberts RS. 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. The Journal of pediatrics 2014;165:6-8.
- Carlo WA, Finer NN, Walsh MC, et al. Target ranges of oxygen saturation in extremely preterm infants. The New England journal of medicine 2010;362:1959-69.
- Tarnow-Mordi W, Stenson B, Kirby A, et al. Outcomes of Two Trials of Oxygen-Saturation Targets in Preterm Infants. The New England journal of medicine 2016;374:749-60.
- Manja V, Lakshminrusimha S, Cook DJ. Oxygen saturation target range for extremely preterm infants: a systematic review and meta-analysis. JAMA pediatrics 2015;169:332-40.