C u r ren t R e c o m m e n d a t i o n s
a n d P r a c t i c e o f Ox y g e n
T h e r a p y in Pret e r m I n f a n t s
William Tarnow-Mordi,
Adrienne Kirby, MSc
BA, MB BChir, MRCP (UK), DCH*,
KEYWORDS
Oxygen Oxygenation Oximetry Infant mortality Necrotizing enterocolitis
Retinopathy of prematurity Extremely premature infant
KEY POINTS
Because low-oxygen saturation targets increased mortality in extremely preterm infants,
all guidelines since 2011, except one, have recommended targeting oxygen saturation
ranges of 91% to 95% or 90% to 95%, or avoiding 85% to 89%.
The quality of the evidence that the low target increased mortality has been rated as
“high,” “moderate,” or “low” despite the low heterogeneity between randomized trials
of oxygen saturation targeting, suggesting key differences in interpreting the GRADE
guidelines.
International surveys in 2015 and 2016 showed that oxygen saturation target ranges had
increased to levels that are more consistent with the evidence.
Randomized controlled trials on oxygen saturation targeting have excluded moderately
and late preterm infants. This is a major evidence gap.
Systematic reviews, guidelines, and consensus statements without biostatisticians or epidemiologists as coauthors should be considered potentially problematic.
INTRODUCTION
Authors of systematic reviews, editorial commentaries, opinions, consensus statements, and guidelines render a valuable service in summarizing the best available evidence for clinical practice. Similarly, those who undertake periodic questionnaire
surveys reveal whether this appraisal of the evidence subsequently impacts clinical
practice. After 2 systematic reviews, which were published in 2017 and 2018,1,2 had
Disclosure Statement: W. Tarnow-Mordi and A. Kirby were Chief Investigators of the BOOST II
Study, funded by the Australian National Health and Medical Research Council and coinvestigators of the NeOPrOM Collaboration.
NHMRC Clinical Trials Centre, University of Sydney, Sydney, New South Wales, Australia
* Corresponding author.
E-mail addresses: williamtm@med.usyd.edu.au; wotarnowmordi@gmail.com
Clin Perinatol 46 (2019) 621–636
https://doi.org/10.1016/j.clp.2019.05.015
0095-5108/19/ª 2019 Elsevier Inc. All rights reserved.
perinatology.theclinics.com
622
Tarnow-Mordi & Kirby
assessed the 5 Neonatal Oxygen Prospective Meta-analysis (NeOProM) trials of low(85%–89%) versus high- (91%–95%) oxygen saturation target ranges in infants of less
than 28 weeks’ gestation,3–7 the authors evaluated published (i) recommendations for
oxygen-saturation target ranges in these infants during oxygen therapy after admission to the neonatal unit and (ii) surveys of practice.
The GRADE Working Group Guidelines
Several systematic reviews, commentaries, and consensus statements have cited the
GRADE guidelines8–10 in assessing (a) the quality of evidence and (b) the strength of
recommendations made. The 4 grades of evidence in the GRADE guidelines are
high quality (indicating that further research is very unlikely to change the confidence
in the estimate of effect), moderate quality (further research is likely to have an important impact on the confidence in the estimate of effect and may change the estimate),
low quality (further research is very likely to have an important impact on the confidence in the estimate of effect and is likely to change the estimate), and very low quality (very uncertain about the estimate). Evidence from observational studies is initially
rated as of low quality, but ratings can be modified upward, while evidence from
randomized studies is initially rated as of high quality, but ratings can be modified
downward. After assessing the quality of evidence, GRADE gives criteria to rate recommendations for practice as “strong” or “weak.”11
The GRADE process begins with a Health Care Question (Patient Intervention,
Comparator, Outcome), progresses through systematic reviews of all evidence
addressing that question, rates the quality of evidence of each outcome across
studies and overall, and then rates the strength of the resulting recommendations,
as summarized in Fig. 1.
OXYGEN SATURATION TARGETS IN INFANTS LESS THAN 28 WEEKS’ GESTATION
Methods
Search strategy
The authors searched PubMed for systematic reviews, editorial commentaries, opinions, consensus statements, and guidelines on oxygen therapy in infants less than
28 weeks’ gestation, excluding (i) reports of individual randomized controlled trials
(RCTs) of low- versus high-oxygen saturation targets, (ii) guidelines focusing on a single measure of outcome, for example, retinopathy of prematurity (ROP), bronchopulmonary dysplasia (BPD), and (iii) systematic reviews or commentaries on oxygen
targeting during resuscitation or delivery room care. Further relevant articles were obtained from the reference lists of publications identified as above. The search was
limited to the period since the publication in 2010 of SUPPORT,3 the first of the NeOProM trials.12 The search terms were “((clinical practice guidelines) AND preterm AND
oxygen) NOT resuscitation,” which yielded 18 hits and “systematic review AND preterm AND oxygen NOT resuscitation,” which yielded 90 hits. Guidelines, systematic
reviews, narrative reviews, commentaries, opinions, and consensus statements
were selected from among these 108 hits. This search strategy yielded 23 publications
(Table 1).1,2,13–33 The authors also searched CINAHL using the terms
(“Infant, Premature”) AND (“Practice Guidelines”), which yielded 28 hits, none of which
were relevant.
These strategies also yielded 2 surveys of clinical practice, one conducted between
November 2015 and February 2016 in 193 European neonatal intensive care units
(NICUs),34 and another among representatives of 329/390 NICUs of the International
Network for Evaluating Outcomes in Neonates (iNeo), conducted in 2015.35
Practice of Oxygen Therapy in Preterm Infants
Fig. 1. Schematic view of GRADE’s process for developing recommendations. a Also labeled
“conditional” or “discretionary”. PICO, patient, intervention, comparator, outcome. (From
Guyatt G, Oxman AD, Akl EA, et al. GRADE guidelines: 1. Introduction-GRADE evidence profiles and summary of findings tables. J Clin Epidemiol 2011;64:383–94; with permission.)
Results
Recommendations for practice
The authors found 2 consensus guidelines,19,29 a clinical report from the Committee
on Fetus and Newborn of the American Academy of Pediatrics (AAP),26 5 systematic
reviews,1,2,20,22,28 and 13 commentaries or opinions (see Table 1).
European Consensus guidelines and American Academy of Pediatrics Clinical Report
In 2016, the European Consensus Guidelines on Management of Respiratory Distress
Syndrome,29 in an update of its earlier consensus statement,19 suggested an oxygen
saturation target of 90% to 94% for all preterm infants less than 28 weeks’ gestation.
Using the GRADE guidelines,8 the European Consensus group assessed the overall
quality of evidence for this recommendation overall, rather than for specific outcomes,
as “moderate” and the strength of this recommendation as “weak.”
In 2016, the AAP Committee on Fetus and Newborn26 concluded that an oxygen
saturation target range greater than 89% may be safer, at least for some infants.
623
624
Tarnow-Mordi & Kirby
Table 1
Recommendations for oxygen saturation target ranges and alarm limits
First Author, Year
Type of Publication
Recommended Target Range
Quality of Evidence (Strength
of Recommendation)
Saugstad et al,13 2011
Commentary
>89%
-
Bashambu et al,14 2012
Narrative review
>89%
-
Askie,15 2013
Opinion
>90%
-
Polin & Bateman16 2013
Commentary
90%–95%
-
Bancalari & Claure,17 2013
Commentary
90%–95%
-
Stenson,18 2013
Opinion
90–95
-
Sweet et al,19 2013
Consensus statement
90%–95%
Moderate
Saugstad & Aune,20 2014
Systematic review
90%–95%
-
Schmidt et al,21 2014
Commentary
85%–88% to 93%–94%a 89%–90% to 95%b
-
Systematic review
-
Lowc
Synnes & Miller,23 2015
Commentary
85%–95%
-
Isaacs24 2016
Commentary
91%–95%
-
Stenson25 2016
Commentary
90%–95%
-
Cummings et al,26 2016
Committee clinical report
90% to 95%d,e
-
Manja et al,
22
2015
Deschmann & Norman,27 2017
Commentary
91%–95%
Manja,28 2017
Systematic review
91%–95%e
Sweet,
29
2017
Askie,1 2017
Consensus statement
90%–94%
Systematic review
-
e
Moderatec
Moderate (weak)
Highc
Kayton,32 2018
Commentary
91%–95%
-
Askie et al,2 2018
Systematic review
-
-
Bizzarro,30 2018
Commentary
91%–95%
-
Einhorn,33 2018
Commentary
91%–95%
-
Stenson & Saugstad,31 2018
Commentary
91–95e
-
Practice of Oxygen Therapy in Preterm Infants
Bold indicates systematic review; italics indicates an outlier commentary which recommends 85-95% in contrast to other commentaries which recommend ranges
above 89%.
a
e.g. alarm limits in hospitals with low rates of mortality and necrotizing enterocolitis but high rates of severe ROP.
b
e.g. alarm limits in hospitals with low rates of severe ROP, but high rates of mortality and necrotizing enterocolitis.
c
For effect on mortality.
d
90 to 95% may be safer than 85% to 89% in some infants.
e
Lower alarm limit will generally need to extend somewhat below the lower target.
625
626
Tarnow-Mordi & Kirby
Although the disproportionately high rate of mortality in small-for-gestational-age infants in the low-target group was discussed, the committee did not define those infants in whom it considered a higher saturation target safer. The AAP Committee on
Fetus and Newborn did not apply the GRADE guidelines.
Systematic reviews
In 2014, Saugstad and Aune20 published a systematic review and meta-analysis on infants less than 28 weeks’ gestation reported in the 5 NeOProM trials. They reported no
difference in death or disability up to 24 months corrected gestational age in 2463 infants
in SUPPORT3 and COT,5 the 2 trials that had reported this outcome. However, all 5 trials
reported an increased relative risk (RR) for mortality at discharge or follow-up (4884 infants; RR 1.41; 95% confidence interval [CI] 1.14–1.74) and for necrotizing enterocolitis
(NEC) (4929 infants; RR 1.25, 95% CI 1.05–1.49). Results were inconclusive for ROP,
BPD, brain injury, and patent ductus arteriosus. They recommended adoption of an oxygen saturation target range of 90% to 95% until 36 weeks’ gestation but did not use the
GRADE guidelines to assess the quality of evidence or strength of this recommendation.
In 2015, Manja and colleagues22 published a systematic review in 4929 infants reported in the 5 NeoProM trials, which was the first to use the GRADE guidelines8 to
assess the quality of evidence for each outcome. It reported (i) no effect for death
or disability up to 24 months corrected gestational age in 3 trials (2716 infants; RR
1.02, 95% CI 0.94–1.14: quality of evidence downgraded to moderate8); (ii) increased
hospital mortality in the low-target group across 4 trials (3757 infants; RR 1.18, 95% CI
1.03–1.36: quality of evidence downgraded to low8 because, among other reasons,
the separation in oxygen saturation achieved between low- and high-target groups
was less than planned); (iii) that the low target increased NEC in 5 trials (4929 infants;
RR 1.24%, 95% 1.05–1.47; quality of evidence downgraded to moderate); (iv) no effect for severe ROP in 5 trials (4066 infants; RR 0.72, 95% CI 0.5–1.04; quality of evidence downgraded to low8 because of unexplained heterogeneity); and (v) no effect
for BPD, neurodevelopmental outcomes at 18 to 24 months, and hearing loss (quality
of evidence downgraded to moderate8).
In 2017, Manja and colleagues28 published a systematic review of RCTs evaluating
the effect of lower (85%–89%) versus higher (91%–95%) pulse oxygen saturation
(SpO2) target on mortality and neurodevelopmental impairment at 18 to 24 months.28
They concluded that the risks associated with restricting the upper SpO2 target limit to
89% outweighed the benefits. The quality of evidence was moderate. They speculated
that a wider target range (lower alarm limit, 89%, and upper alarm limit, 96%) might
increase time spent within the 91% to 95% range.
In 2017, Askie and colleagues1 published a Cochrane Review of the effects of targeting lower versus higher arterial oxygen saturations on death or disability in preterm
infants, based on the 5 NeOProM trials, in a total of 4965 infants less than 28 weeks’
gestation. They applied the GRADE guidelines to assess the quality of evidence for estimates of the effect of the low-target range on individual outcomes. Their assessments of the quality of evidence for these outcomes have been summarized
alongside the assessments of Manja and colleagues22 in Table 2.
In 2018, Askie and colleagues2 published the results of the previously planned12
prospective individual patient data meta-analysis of the effects of targeting 85% to
89% versus 91% to 95% oxygen saturation in all 4965 infants less than 28 weeks’
gestation in the 5 NeOProM Collaboration trials. This publication2 (which did not apply
the GRADE guidelines12) and the Cochrane Review by Askie and colleagues1 suggest
that for every 1000 infants, targeting low- versus high-oxygen saturation made no difference in death or major disability up to 18 to 24 months, nor in major disability,
Practice of Oxygen Therapy in Preterm Infants
Table 2
Observed effects of the low-target range 85% to 89% on individual outcomes in 2 systematic
reviews of 5 trials, with quality of evidence assessed by GRADE
Observed Effect
of 85%–89% Target Quality of Evidence
on This Outcome in as Assessed in
Manja et al,22 2015 Manja et al,22 2015
Observed Effect of
85%–89% Target on
This Outcome in Askie
et al,1 2017
Quality of
Evidence as
Assessed in
Askie et al,1
2017
Death or
disability
at 24 mo
No effect
Moderate
No effect
High
Hospital
death
Increased
Low
Increased
High
Moderate
No effect
High
Outcome
Disability at No effect
18–24 mo
NEC
Moderate
Increased
High
Treated ROP No effect
Increased
Low
No effect
Moderate
Hearing loss No effect
Moderate
No effect
Not assessed
Blindness
Moderate
No effect
Not assessed
No effect
including blindness, but led to 28 more deaths, 22 more cases of NEC, and 42 fewer
infants being treated for ROP.1,2,36
Table 3 presents the authors’ reasons for uprating the Quality of Evidence that the
low target increases mortality in the NeOProM trials in 2 systematic reviews using
GRADE from “low”22 to “high.”1
SURVEYS OF CLINICAL PRACTICE IN OXYGEN TARGETING AMONG INFANTS LESS
THAN 28 WEEKS’ GESTATION
In a Web-based survey among representatives of 390 NICUs of the International
Network for Evaluating Outcomes in Neonates (iNeo) conducted in 2015, responses
were received from 329 (84%).35 Of these, 60% had recently made changes to their
upper and lower SpO2 target limits, with the median value of the target range set higher
than previously by 2% to 3% in 8 of 10 networks.
In a similar survey of 193 European NICUs conducted in 2015 and 2016,34 there was
considerable variation in practice. The most frequently targeted oxygen-saturation
ranges were 90% to 95% (28%), 88% to 95% (12%), 90% to 94% (5%), and 91%
to 95% (5%), reflecting the most commonly recommended target ranges shown in
Table 1. A total of 156 NICUs (81%) had recently changed their oxygen saturation
target limits. The median values for the oxygen-saturation ranges in clinical practice
had increased by between 3% and 5% within the last 10 years.
Neither survey determined current practice in setting alarm limits nor in oxygen
saturation targeting for preterm infants of 30 to 36 weeks’ gestation.
Discussion
All 5 systematic reviews of trials of oxygen saturation targeting for infants of less than
28 weeks’ gestation that were identified since 2010 concluded that the low target 85%
to 89% increased mortality.1,2,20,22,28 Table 1 shows notable uniformity in recommendations for practice. Of 16 guidelines, commentaries, or opinions published since
SUPPORT in 2010 (see Table 1), 15 recommended targeting a range greater than
627
628
Tarnow-Mordi & Kirby
Table 3
Contrasting interpretations of the quality of evidence that the low target increased mortality
in randomized controlled trials of oxygen saturation targets in infants less than 28 wk using
GRADE8
Initial Reason for Downgrading
Quality of Evidence22
Our Interpretation
of the Effect on
Quality of Evidence
Reinterpretation Using the
GRADE Guidelines
1. Separation of achieved SpO2
values was less than planned
Despite less than expected
separation, mortality effects
were still observed. This
confounding is likely to have
reduced the effect (see Fig. 18)
2. The oximeter algorithm was
revised during enrollment to 3
trials5,7
The revised algorithm improved Rating should be
separation in oxygen
modified upward
saturation36 and thus study
power. The original algorithm
reduced separation (see Fig. 18),
thus underestimating the
average estimate of effect36
Rating should be
modified upward
3. Mortality in infants on the revised The BOOST II protocols specified a Rating should be
algorithm was not a
difference in a major endpoint
modified upward
prespecified outcome; yet 2
of >3 standard errors or >3
BOOST II trials7,37 were stopped
standard deviations (equivalent
to P<.0027) to justify early
early because of an increase in
stopping. After SUPPORT
this outcome
showed excess mortality,3 the
BOOST II trials7,37 were stopped
early because of a highly
significant increase in pooled
mortality in infants on revised
oximeters (RR 1.65; P 5 .0003)38
4. Only 43,6,37 of 5 trials reported
hospital death
The estimates of effect on
hospital death in these 4 trials
demonstrate low to moderate
heterogeneity39 (c2 4.76,
P 5 .19; I2 5 37%)
No effect on rating
Other issues
5. Effects on mortality using revised The mortality risk for low- vs high- Rating should be
vs original oximeters; among
target infants was greater in
modified upward
infants on revised oximeters,
those using revised vs original
low-target infants had
oximeters.8 This, and the
consistently higher mortality
finding that targeting was
than high-target infants, with
more accurate in infants using
(a) statistically significant
revised oximeters,36 is likely to
interactions in 3 meta-analyses
have caused the average effect
of mortality in low- vs highof the low target on mortality
target infants stratified by use
in infants on original oximeters
of original vs revised oximeters
to have been underestimated,
(P 5 .00638; P 5 .031,20;
as in 1 above (see Fig. 18)
P 5 .0422) (but interactions
were unreported in 2 metaanalyses20,22); (b) Infants using
revised oximeters spent more
time in their assigned target
ranges than those on original
oximeters36
(continued on next page)
Practice of Oxygen Therapy in Preterm Infants
Table 3
(continued )
Initial Reason for Downgrading
Quality of Evidence22
Reinterpretation Using the
GRADE Guidelines
Our Interpretation
of the Effect on
Quality of Evidence
Shows high quality
6. What was the potential for biased None. Death was assessed
assessment of mortality?
without interobserver variation
or bias
7. Was there loss to follow-up in
ascertainment of mortality?
No. 100% ascertainment of death Shows high quality
at 36 wk and before hospital
discharge
8. Was the magnitude of the effect Yes. The low target shows a 38% Rating should be
modified upward
on mortality large?
increase in RR of death in
infants on revised oximeters (RR
1.38, 95% CI 1.13–1.68;
P 5 .014)1 (see Fig. 18)
90% or avoiding the low-target range of 85% to 89%. The exception was an editorial
commentary,23 which suggested that targeting an oxygen-saturation range between
85% and 95% remained acceptable. Two international surveys34,35 in approximately
500 neonatal units reported considerable variation in clinical practice, but documented recent increases of 2% to 5% in the median values for target oxygensaturation ranges.
CONTRASTING INTERPRETATIONS OF CURRENT RANDOMIZED EVIDENCE
The recommendation23 that an oxygen-saturation target range of 85% to 89%
remained acceptable23 reflects a judgment in the systematic review of Manja and colleagues22 that the GRADE quality of evidence that the low target increased mortality
was low. This conclusion merits reconsideration. Reviewers may legitimately disagree
when interpreting quality of evidence using GRADE, which provides a transparent
framework guiding the assessment process (see Fig. 1).8,9 Table 2 shows judgments
from 2 systematic reviews1,22 about the quality of evidence for the effect of the low
target on various outcomes. Table 3 outlines the present authors’ reasons for reinterpreting the quality of evidence for the effect of the low target on mortality as “high”
rather than “low,” using GRADE. For example, a treatment effect for mortality was
observed despite the confounding effect of achieving less separation between study
arms in actual oxygen saturation than planned. This result supports an upward modification of the rating for quality (see Fig. 1).
Neither the European Consensus Guidelines nor the AAP Guidance has been
updated since these most recent data were published.1,2 Future recommendations
may benefit from inviting coauthorship by colleagues with advanced biostatistic and
epidemiologic expertise. Indeed, attempting to synthesise and interpret complex trial
data without such expertise might be compared with attempting brain surgery without
a neurosurgeon.
Future international clinical surveys32–35 and other evidence36,40 are warranted, to
assess the impact on practice of the final results of the NeOPrOM trials.1,2 Future practice may also be influenced by increasing attention to 5 questions:
i. Did the pulse oximeters which were used in the NeOProM trials estimate hypoxemia with progressively wider limits of accuracy as true oxygen-saturation
values (SaO2) decreased from 93% to 80%, as reported in 2012 by Rosychuk
629
630
Tarnow-Mordi & Kirby
et al (Fig. 2)?41 Did the pulse oximeters expose substantially more infants in the
lower-target group to values of oxygen tension less than 50 mm Hg (6.7 kPa).42
The study authors41 wrote, “The large sample size of each of these (NeOProM) trials
and the planned meta-analysis should address the question of whether infants assigned to the lower range are at higher risk of low PaO2-induced pulmonary vasoconstriction, patent ductus arteriosus, abnormal neuro-developmental outcome,
and potentially death. Our data may provide partial explanations if differences in
short- or long-term outcomes are observed.”41
ii. Could targeting an untested intermediate oxygen-saturation range, such as 87 to
93%, increase mortality versus a higher range because current oximeters inadvertently permit increasingly disproportionate exposure to hypoxemia as true oxygen saturation decreases below 93%? (see Fig. 2).40,41 At present, the most
rigorously evaluated evidence for policy is that targeting oxygen saturations of
91 to 95% is safer overall than targeting oxygen saturations of 85% to
89%.1,2,40 The most reliable evidence for an intermediate range will come from
an adequately powered RCT of two pre-specified target ranges, defined by
pre-specified alarm settings.
iii. Will future practice reflect recent analyses showing that infants with revised oximeters in the Australian and UK Benefits of Oxygen Saturation Targeting-II
(BOOST II) trials spent longer in their planned pulse oximeter saturation target
ranges than infants with the original oximeters (P<0.001)?36 There was no difference in separation of median oxygen saturation between infants with original vs
revised oximeters,43 but the increased targeting accuracy in the low-target range
Fig. 2. Proportions of SpO2 values that are 3% of the corresponding SaO2 value are
plotted. The numbers above the x-axis on the graph denote the number of measurements
at each value of SaO2, i.e. a total of 1,620 measurements. There is a highly significant
difference in the proportions of measurements in which SpO2 is within 3% of the
corresponding SaO2 value for values of SaO2 of 92% versus values of SaO2 >92%
(Chi squared >63, p<0.0001). Further analyses adjusted for clustering by infant will be
valuable. (From Rosychuk RJ, Hudson-Mason A, Eklund D, Lacaze-Masmonteil T. Discrepancies between arterial oxygen saturation and functional oxygen saturation measured
with pulse oximetry in very preterm infants. Neonatology 2012;101:14–9; with permission.)
Table 4
Sample sizes to show 10% or 20% reductions in relative risk of mortality or disability
Total Sample to Show Effect in a 2-Arm Comparison with 90%
Power, 2P 5 .05 and Nonadherence to Protocol Due to:
Relative Risk
Reduction
(1 L RR)
Number
Needed to
Benefit or
Harm (100/D)
0% Crossover in
Each Group (Perfect
Adherence to Protocol)
5% Crossover in
Each Group (Total
10% Crossover)
10% Crossover
in Each Group
(Total 20%
Crossover)
16
4
0.8
0.2
25
3868
4776
6044
18
2
0.9
0.1
50
16,166
19,960
25,260
10
8
2
0.8
0.2
50
8598
10,616
13,436
10
9
1
0.9
0.1
100
36,136
44,164
56,464
Event Rate
in Treatment
Group (T), %
20
20
Data from Sealed envelope. Trial sample size calculator. Available at: https://www.sealedenvelope.com/power/. Accessed 24 Jan 2018.
Practice of Oxygen Therapy in Preterm Infants
Risk Difference
(C L T 5 D), %
Relative Risk
or Risk Ratio
(RR 5 T/C)
Event Rate
in Control
Group (C), %
631
632
Tarnow-Mordi & Kirby
with revised oximeters36 may explain the larger mortality difference between lowand high-target infants with revised oximeters. It also suggests that average treatment effects in the BOOST II trials are underestimates, as they include data from
infants on the original oximeters.36
iv. Should future clinical policy be based on observational studies? No. David
Sackett, an early advocate of Evidence-Based Medicine, stated “If you’re scanning an article about therapy and it is not a randomized trial, why on earth are
you wasting your time?”44 Two major threats to the validity of therapeutic comparisons are random error and systematic bias. Randomization is the most reliable way to protect therapeutic comparisons from systematic bias, because it
tends to balance all confounding variables, including those which are currently
unknown, evenly between intervention and control groups.45 Large observational analyses can minimize random error, but cannot overcome errors arising
from systematic bias. Hence the most reliable evidence for future policy will
come from large randomized trials, preferably those which incorporate automated targeting of oxygen saturation with predefined settings for alarm limits
around each target.
v. What do parents and former patients think of the trade-offs between lower mortality and NEC vs lower survival and less ROP, or future trade-offs? Increasingly, using digital technology to seek input from parents and patients will allow their voices
to be heard in discussions about trade-offs between competing outcomes in oxygen trials. Such methods could allow parents, former patients, professionals, policymakers and other stakeholders to help prioritize core questions for large
international trials. Such trials could evaluate the effects of different oxygen targeting policies on survival as primary outcome, and on disability in survivors as a key
secondary outcome.46–48
LACK OF EVIDENCE FROM TRIALS IN MODERATELY PRETERM OR LATE PRETERM
INFANTS
No trials or systematic reviews have evaluated the effects of different oxygen saturation targets or saturation alarm limits on short- or long-term outcomes in preterm infants of 28 to 36 weeks’ gestation, who may outnumber extremely preterm infants
of less than 28 weeks’ gestation by 10-fold or 20-fold.49 This represents a major evidence gap.
OBTAINING EVIDENCE FOR FUTURE RECOMMENDATIONS ON OXYGEN TARGETS IN
PRETERM INFANTS
In 1998, Peto and Baigent wrote,50 “.medical research needs to find practicable ways
of greatly increasing the size of randomized studies; otherwise moderate but worthwhile benefits will continue to be missed .” As event rates fall, if mortality and
disability are to improve further, large, well-designed perinatal trials with increasingly
large sample sizes will be needed.51 Table 4 illustrates the sample sizes needed to
detect moderate, worthwhile reductions of 10% and 20% in key outcomes, such as
mortality or disability, with adequate power and realistic rates of nonadherence to protocol.52 These sample sizes might mean about 1 additional survivor without major
disability for every 50 to 100 patients treated, which many would consider worthwhile
for a widely available, affordable treatment like oxygen. All this underlines the need for
increasing international collaboration, a major goal of the newly conceived ALPHA
Collaboration for Advancing Large, collectively Prioritized perinatal trials of Health outcomes Assessment.46–48
Practice of Oxygen Therapy in Preterm Infants
Best Practices
What is the current practice for oxygen saturation targeting in infants of less than 28 weeks’
gestation?
Surveys conducted in 2015 and 2016 in more than 500 NICUs showed wide variation in practice,
with between 60% and 81% of NICUs having increased their median oxygen saturation target
ranges by 2% to 5%.34,35 The most up-to-date guidance for practice, published since the
NeOProM Collaboration results in 2017 and 2018,1,2 recommends a target range of 91 to
95.30,31
What changes in current practice are likely to improve outcomes?
Targeting an oxygen saturation of 91% to 95% will reduce mortality and NEC and increase
retinopathy without increasing blindness or disability, compared with targeting an oxygen
saturation of 85% to 89%.1,2,7 Targeting an intermediate oxygen-saturation range, such as
87% to 93%, is an untested practice that may increase mortality compared with targeting 91%
to 95%, because current oximeters permit increasingly disproportionate exposure to
hypoxemia as oxygen saturation decreases to less than 93%.7,41
Is there a Clinical Algorithm? No.
New trials are needed in infants of 28 weeks and 0 days to 35 weeks and 6 days gestation to
determine which targets minimize mortality and NEC and ROP. However, the incidence of ROP
is likely to be proportionately much lower than the incidence of the first 2 outcomes in these
infants.
Major Recommendations:
Adopt a target range of 91% to 95% for oxygen saturation in infants less than 28 weeks’
gestation.
Rating for the Strength of the Evidence: High.1
Bibliographic Source(s): Refs.1,2,36,41
ACKNOWLEDGMENT
We acknowledge valuable comments by Dr Balpreet Singh, IWK Hospital, Halifax.
REFERENCES
1. Askie LM, Darlow BA, Davis PG, et al. Effects of targeting lower versus higher
arterial oxygen saturations on death or disability in preterm infants. Cochrane
Database Syst Rev 2017;(4):CD011190.
2. Askie LM, Darlow BA, Finer N, et al. Association between oxygen saturation targeting and death or disability in extremely preterm infants in the neonatal oxygenation prospective meta-analysis collaboration. JAMA 2018;319:2190–201.
3. Carlo WA, Finer NN, Walsh MC, et al. Target ranges of oxygen saturation in
extremely preterm infants. N Engl J Med 2010;362:1959–69.
4. Vaucher YE, Peralta-Carcelen M, Finer NN, et al. Neurodevelopmental outcomes in the early CPAP and pulse oximetry trial. N Engl J Med 2012;367:
2495–504.
5. Schmidt B, Whyte RK, Asztalos EV, 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;309:2111–20.
6. Darlow BA, Marschner SL, Donoghoe M, et al. Randomized controlled trial of oxygen saturation targets in very preterm infants: two year outcomes. J Pediatr
2014;165:30–5.e2.
633
634
Tarnow-Mordi & Kirby
7. Tarnow-Mordi W, Stenson B, Kirby A, et al, BOOST II Australia and BOOST II UK
Collaborative Study Groups. Outcomes of two trials of oxygen-saturation targets
in preterm infants. N Engl J Med 2016;374:749–60.
8. Guyatt G, Oxman AD, Akl EA, et al. GRADE guidelines: 1. Introduction-GRADE
evidence profiles and summary of findings tables. J Clin Epidemiol 2011;64:
383–94.
9. Balshem H, Helfand M, Schunemann HJ, et al. GRADE guidelines: 3. Rating the
quality of evidence. J Clin Epidemiol 2011;64:401–6.
10. Guyatt GH, Oxman AD, Kunz R, et al. What is “quality of evidence” and why is it
important to clinicians? BMJ 2008;336:995–8.
11. Guyatt GH, Oxman AD, Kunz R, et al. Going from evidence to recommendations.
BMJ 2008;336:1049–51.
12. Askie LM, Brocklehurst P, Darlow BA, et al. NeOProM: neonatal oxygenation prospective meta-analysis collaboration study protocol. BMC Pediatr 2011;11:6.
13. Saugstad OD, Speer CP, Halliday HL. Oxygen saturation in immature babies: revisited with updated recommendations. Neonatology 2011;100:217–8.
14. Bashambu MT, Bhola M, Walsh M. Evidence for oxygen use in preterm infants.
Acta Paediatr 2012;101:29–33.
15. Askie LM. Optimal oxygen saturations in preterm infants: a moving target. Curr
Opin Pediatr 2013;25:188–92.
16. Polin RA, Bateman D. Oxygen-saturation targets in preterm infants. N Engl J Med
2013;368:2141–2.
17. Bancalari E, Claure N. Oxygenation targets and outcomes in premature infants.
JAMA 2013;309:2161–2.
18. Stenson BJ. Oxygen targets for preterm infants. Neonatology 2013;103:341–5.
19. Sweet DG, Carnielli V, Greisen G, et al. European consensus guidelines on the
management of neonatal respiratory distress syndrome in preterm infants–2013
update. Neonatology 2013;103:353–68.
20. Saugstad OD, Aune D. Optimal oxygenation of extremely low birth weight infants:
a meta-analysis and systematic review of the oxygen saturation target studies.
Neonatology 2014;105:55–63.
21. 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. J Pediatr 2014;165:6–8.
22. Manja V, Lakshminrusimha S, Cook DJ. Oxygen saturation target range for
extremely preterm infants: a systematic review and meta-analysis. JAMA Pediatr
2015;169:332–40.
23. Synnes A, Miller SP. Oxygen therapy for preterm neonates: the elusive optimal
target. JAMA Pediatr 2015;169:311–3.
24. Isaacs D. Optimal oxygen saturation target for preterm infants. J Paediatr Child
Health 2016;52:783.
25. Stenson BJ. Oxygen saturation targets for extremely preterm infants after the
NeOProM trials. Neonatology 2016;109:352–8.
26. Cummings JJ, Polin RA, Committee On F, et al. Oxygen targeting in extremely low
birth weight infants. Pediatrics 2016;138 [pii:e20161576].
27. Deschmann E, Norman M. Oxygen-saturation targets in extremely preterm infants. Acta Paediatr 2017;106:1014.
28. Manja V, Saugstad OD, Lakshminrusimha S. Oxygen saturation targets in preterm
infants and outcomes at 18-24 months: a systematic review. Pediatrics 2017;139
[pii:e20161609].
Practice of Oxygen Therapy in Preterm Infants
29. Sweet DG, Carnielli V, Greisen G, et al. European consensus guidelines on the
management of respiratory distress syndrome—2016 update. Neonatology
2017;111:107–25.
30. Bizzarro MJ. Optimizing oxygen saturation targets in extremely preterm infants.
JAMA 2018;319:2173–4.
31. Stenson B, Saugstad OD. Oxygen treatment for immature infants beyond the delivery room: lessons from randomized studies. J Pediatr 2018;200:12–8.
32. Kayton A, Timoney P, Vargo L, et al. A review of oxygen Physiology and Appropriate management of oxygen levels in premature neonates. Adv Neonatal
Care 2018;18:98–104.
33. Einhorn, J. Oxygen Saturation Limits for Premature Babies: The Final Word for
Now. NEJM Journal Watch. 25 June 2018. Available at: https://www.jwatch.org/
na46890/2018/06/25/oxygen-saturation-limits-premature-babies-final-word-now.
Accessed May 27, 2019.
34. Huizing MJ, Villamor-Martinez E, Vento M, et al. Pulse oximeter saturation target
limits for preterm infants: a survey among European neonatal intensive care units.
Eur J Pediatr 2017;176:51–6.
35. Darlow BA, Vento M, Beltempo M, et al. Variations in oxygen saturation targeting,
and retinopathy of prematurity screening and treatment criteria in neonatal intensive care units: an international survey. Neonatology 2018;114:323–31.
36. Stenson BJ, Donoghoe M, Brocklehurst P, et al. Pulse oximeter saturation targeting and oximeter changes in the benefits of oxygen saturation targeting
(BOOST)-II Australia and BOOST-II UK oxygen trials. J Pediatr 2019;204:
301–4.e2.
37. Stenson BJ, Tarnow-Mordi WO, Darlow BA, et al, for the Boost II United Kingdom,
Australian and New Zealand Collaborative Groups. Oxygen saturation and outcomes in preterm infants. N Engl J Med 2013;368:2094–104.
38. Stenson B, Brocklehurst P, Tarnow-Mordi W. Increased 36-week survival with high
oxygen saturation target in extremely preterm infants. N Engl J Med 2011;364:
1680–2.
39. Ryan R, Hill S. How to GRADE the quality of the evidence. Cochrane Consumers
and Communication Group; 2016. Available at: http://cccrg.cochrane.org/authorresources. Accessed June 14, 2019.
40. Tarnow Mordi W, Stenson B, for BOOST II Australia and BOOST II UK Collaborative Groups. Outcomes of two trials of oxygen saturation targets in preterm infants. N Engl J Med 2016;374:749–60.
41. Rosychuk RJ, Hudson-Mason A, Eklund D, et al. Discrepancies between arterial
oxygen saturation and functional oxygen saturation measured with pulse oximetry
in very preterm infants. Neonatology 2012;101:14–9.
42. Quine D, Stenson BJ. Arterial oxygen tension (PaO2) values in infants <29 weeks
of gestation at currently targeted saturations. Arch Dis Child Fetal Neonatal Ed
2009;94:F51–3.
43. Whyte RK, Nelson H, Roberts RS, et al. Benefits of oxygen saturation targeting
trials: oximeter calibration software revision and infant saturations. J Pediatr
2017;182:382–4.
44. Vines G. Quotation attributed to David Sackett. In: Is there a database in the
house? New Scientist 1995. Available at: https://www.newscientist.com/article/
mg14519612-300-is-there-a-database-in-the-house/. Accessed May 27, 2019.
45. Byar DP, Simon RM, Friedewald WT, et al. Randomized clinical trials. Perspectives on some recent ideas. N Engl J Med 1976;295:74–80.
635
636
Tarnow-Mordi & Kirby
46. Tarnow-Mordi W, Cruz M, Morris JM, et al. RCT evidence should drive clinical
practice: a day without randomisation is a day without progress. BJOG 2017;
124:613.
47. Fogarty M, Osborn DA, Askie L, et al. Delayed vs early umbilical cord clamping
for preterm infants: a systematic review and meta-analysis. Am J Obstet Gynecol
2018;218:1–18.
48. NHMRC Clinical Trials Centre, University of Sydney. Saving millions more lives
through megatrials; Introducing the ALPHA Collaboration. Available at: https://
www.youtube.com/watch?v5oy3J6iZzOfE. Accessed July 5, 2018.
49. Blencowe H, Cousens S, Oestergaard MZ, et al. National, regional, and worldwide estimates of preterm birth rates in the year 2010 with time trends since
1990 for selected countries: a systematic analysis and implications. Lancet
2012;379:2162–72.
50. Peto R, Baigent C. Trials: the next 50 years. Large scale randomised evidence of
moderate benefits. BMJ 1998;317:1170–1.
51. Tarnow-Mordi W, Cruz M, Morris J. Design and conduct of a large obstetric or
neonatal randomized controlled trial. Semin Fetal Neonatal Med 2015;20:
389–402.
52. Sealed envelope. Trial sample size calculator. Available at: https://www.
sealedenvelope.com/power/. Accessed January 24, 2018.