Beruflich Dokumente
Kultur Dokumente
Agradecimentos:
A Deus.
A todos que foram para o Simposio e contribuiram com material para confeccao deste compilado,
em especial a Dra. Suerda, Dra. Andrea Sueli e a Dra. Ana Caroline.
Saturacao-Alvo de Oxigenio para Prematuros Extremos apos
o Estudo NeOProM
Review
Key Words below 90% and indicate the importance of more trials to see
Extremely preterm infants Neonatal Oxygen Prospective if a further survival advantage can be identified. The safety
Meta-Analysis Pulse oximeter saturation of targets above 95% has not been evaluated. The five trials
were designed to be similar to facilitate an individual patient
data meta-analysis, and this Neonatal Oxygen Prospective
Abstract Meta-Analysis (NeOProM) may provide further insights.
Five randomized controlled trials comparing lower (8589%) 2016 S. Karger AG, Basel
versus higher (9195%) pulse oximeter saturation (SpO2) tar-
gets for extremely preterm infants have now been reported
from the United States of America, Canada, the United King- Historical Evidence
dom, Australia and New Zealand. These trials included more
than 4,800 infants, and they provide robust evidence to per- Prior to the recent randomized controlled trials of low-
mit comparison of these target ranges and consider the next er versus higher oxygen saturation (SpO2) target ranges,
steps for clinicians and researchers. The lower SpO2 range the evidence base to guide oxygen therapy in extremely
was associated with a significant increase in the risk of death. preterm infants was very weak. When oxygen therapy was
There was no significant difference between the two target first introduced into neonatal care, there was no way to
ranges in the rate of disability at 1824 months, including measure arterial oxygen levels. Oxygen was administered
blindness. A significant difference between groups in the to preterm infants at high inspired percentages for long
risk of the composite primary outcome of death or disability periods according to clinical judgements and it was in this
in favour of the higher SpO2 range was mainly attributable context that retinopathy of prematurity (ROP) was linked
to the difference between groups in the risk of death. The with prolonged exposure to high inspired oxygen (FiO2).
lower target range did not reduce bronchopulmonary dys- Trials conducted without measurements of oxygenation
plasia or severe visual impairment, but it did increase the risk demonstrated that a restrictive approach to oxygen sup-
of necrotizing enterocolitis requiring surgery or causing plementation could reduce the number of infants with
death. The trials provide no reason to prefer SpO2 targets retinopathy [13], but the oxygen restriction that fol-
128.111.121.42 - 6/9/2016 6:22:01 AM
Univ. of California Santa Barbara
E-Mail karger@karger.com
Edinburgh EH16 4SA (UK)
www.karger.com/neo
E-Mail ben.stenson@luht.scot.nhs.uk
lowed came with a cost of increased mortality and cere- values at any individual SpO2 is wide. The 95% confidence
bral palsy [48]. intervals for PO2 in oxygen-dependent preterm infants for
When it became possible to measure oxygen tension, the SpO2 range 8595% in the first week after birth include
first by intermittent arterial puncture and then by mea- a PO2 range of 2967 mm Hg (3.88.9 kPa) [13]. In other
surement of transcutaneous oxygen tension (PO2), con- words, the shift to SpO2 monitoring with a range of 85
trolled oxygen therapy began. Oxygen was titrated to 95% from transcutaneous PO2 monitoring was associated
achieve various PO2 ranges, but there were no trials to with an unevaluated and substantial shift downwards in
determine the upper and lower PO2 limits associated with the permitted range of PO2 values that preterm infants
the best outcomes. were being targeted to from previous limits of 5080 mm
A trial published in 1987 aimed to show whether using Hg (6.710.7 kPa) to a range of around 2967 mm Hg
continuous transcutaneous PO2 monitoring, in addition (3.88.9 kPa). Encouragingly, observational studies sug-
to intermittent arterial sampling, would improve out- gested the possibility that targeting lower SpO2 values
comes in comparison with intermittent arterial sampling might allow a reduction in ROP and bronchopulmonary
alone [9]. PO2 targets of 5070 mm Hg (6.79.3 kPa) were dysplasia (BPD), without an increase in the risk of mortal-
used in both groups. There were 296 infants with birth ity or disability [1417], giving rise to wide interest in the
weight <1,300 g randomized. The addition of transcuta- evaluation of different SpO2 targets.
neous monitoring was associated with an 8% decrease in
all stages of ROP (51 vs. 59%). However, there was no dif-
ference in more concerning severe retinopathy. There Recent Trial Evidence
were 8% more deaths in the transcutaneous monitoring
group (32 vs. 24%). These differences were not statisti- Five randomized controlled trials have now been un-
cally significant and may be chance findings, but in retro- dertaken since 2005 to compare the outcomes of infants
spect they may represent further early evidence that tight- targeted to lower (8589%) versus higher (9195%) SpO2
ening up the control of oxygen administration reduced targets [1821]. These trials were conducted as separately
retinopathy at a cost of increased mortality. In a later funded and managed components of a prospective indi-
analysis of the same study, ROP was associated with PO2 vidual patient data meta-analysis known as the Neonatal
levels greater than 80 mm Hg (10.7 kPa) [10]. The Amer- Oxygen Prospective Meta-Analysis (NeOProM) Collabo-
ican Academy of Pediatrics recommended a PO2 target ration [22]. The trials used Masimo Radical oximeters
range of 5080 mm Hg (6.710.7 kPa) [11]. (Masimo Corporation, Irvine, Calif., USA) that were in
Pulse oximetry gradually became the most widely used widespread use clinically at the time. In order to mask the
monitoring method, probably because of its ease of ap- group allocation from the caregivers, the trial oximeters
plication and not because of its demonstrated superiority were offset to read either 3 percentage points higher or
in optimizing clinical outcomes. There were no trials to lower than the underlying true SpO2 reading in the range
determine that using SpO2 monitoring improved out- 8595%, so that in the lower SpO2 target group a SpO2
comes when compared with transcutaneous PO2 moni- reading of 90% would be displayed at an underlying SpO2
toring. It remains an open question whether oxygen ten- of 87% and in the higher target group a SpO2 reading of
sion or oxygen saturation represents the safer measure for 90% would be displayed at an underlying SpO2 of 93%. If
guiding oxygen treatment in preterm infants. A small nursing staff targeted the displayed SpO2 range 8892%,
crossover study suggested that transcutaneous monitor- this was expected to achieve the intended ranges of 85
ing may facilitate avoidance of high and low oxygen ten- 89% and 9195% in the two randomization groups, re-
sion more effectively than saturation monitoring [12]. spectively, and in the middle of the target ranges there was
In the absence of trials to determine the optimal satura- expected to be a 6% difference between groups in mean
tion target range for preterm infants there was widespread SpO2. In order that clinical staff could be confident that
variation in practice. A survey of practice in UK neonatal they were not allowing what they might consider clini-
units interested in participating in the BOOST-II UK trial cally significant hypoxia or hyperoxia at SpO2 above 95 or
suggested that a SpO2 range of 8595% was broadly rep- below 85%, the oximeter transitioned back to displaying
resentative of practice before the recent trials; however, the underlying true reading. An oximeter offset of plus or
some units were using higher or lower limits [unpubl. minus 2% was effective in achieving separation in SpO2
data]. The relationship between PO2 and saturation varies distributions between randomization groups in the first
within and between infants, and the range of possible PO2 BOOST trial in 2003 [23].
128.111.121.42 - 6/9/2016 6:22:01 AM
Univ. of California Santa Barbara
Fig. 1. Meta-analysis of mortality at 1824 months for all infants reported, using both the original and revised
oximeters. Analysed using RevMan version 5.3, Cochrane Collaboration.
Summary Outcomes (43.7%) infants in the low SpO2 target group versus
967/2,372 (40.8%) infants in the high target group (RR
A meta-analysis of some of the outcomes from the 5 1.07, 95% CI 1.001.14; p = 0.04). When this analysis is
NeOProM trials was published in 2014 [27]. At this time restricted to infants treated with the revised oximeters,
the UK, Australian and New Zealand BOOST-II trials had death or disability occurred in 430/852 (50.5%) low target
reported their outcomes to hospital discharge, but their group infants versus 372/829 (44.9%) high target infants
2-year outcome data were not available. The 2014 meta- (RR 1.13, 95% CI 1.021.24; p = 0.02). A test for interac-
analysis reported data for 4,911 infants. The analyses tion was not statistically significant for the 2 different ox-
showed that there was a lower risk of severe ROP in infants imeter types for this composite outcome (p = 0.24).
targeted to lower SpO2 (10.7%) compared to the higher Considering the components of the primary outcome
SpO2 (14.5%), with a relative risk (RR) of 0.74 and 95% separately, death before 1824 months occurred in
confidence interval (CI) of 0.590.92. Infants randomized 484/2,433 (19.9%) low target infants and 418/2,440
to the lower SpO2 range had a higher risk of necrotizing (17.1%) of high target infants (RR 1.16, 95% CI 1.031.31;
enterocolitis (11.2 vs. 9.0%; RR 1.25, 95% CI 1.051.49). p = 0.01; fig.1). When this analysis was restricted to the
There was no significant difference between groups in in- results obtained using the revised oximeters, death before
cidence of BPD when this was defined physiologically or 1824 months occurred in 190/860 (22.1%) low target in-
as severe disease (37.6 vs. 39.7%; RR 0.95, 95% CI 0.86 fants versus 137/856 (16.0%) high target infants (RR 1.38,
1.04). There was no significant difference between groups 95% CI 1.131.68; p = 0.001). A test for interaction showed
in brain injury (14.2 vs. 14.1%; RR 1.02, 95% CI 0.881.19) that the results obtained using the revised oximeters were
or in treatment for patent ductus arteriosus (49.2 vs. 49%; significantly different from those obtained with the orig-
RR 1.10, 95% CI 0.951.08). There was no heterogeneity inal oximeters (p = 0.009). The effect of the change in ox-
between trials for any of these outcomes. imeters on mortality was similar in all 3 trials which
Data on outcomes to 1824 months have now been changed to the revised oximeters during enrollment
reported for all 5 trials [19, 20, 25, 28]. Data are available (fig.2).
for the composite outcome of death or disability in 4,751 There was no significant difference between groups in
infants, 2,379 who were treated with the lower SpO2 tar- the composite measure of disability. One or more of the
get and 2,372 who were treated with the higher SpO2 tar- measures of disability was identified in 566/1,895 (29.9%)
get. A fixed effects meta-analysis of the 5 trials combining infants randomized to the low SpO2 target and 539/1,954
the results of infants who were treated with the original (28.1%) infants randomized to the high SpO2 target (RR
and the revised oximeters shows that the composite out- 1.04, 95% CI 0.941.14). The results were similar when
come of death or disability occurred in 1,040/2,379 the analysis was restricted to the infants treated with re-
128.111.121.42 - 6/9/2016 6:22:01 AM
Univ. of California Santa Barbara
Fig. 2. Meta-analysis of mortality at 1824 months in the BOOST-II UK and Australian trials and COT, sub-
grouped according to whether infants were treated with the original or the revised oximeters. Analysed using
RevMan version 5.3, Cochrane Collaboration.
vised oximeters (RR 1.06, 95% CI 0.921.23). Disability that context, targeting SpO2 below 90% significantly in-
outcomes for COT are not in the trial outcomes paper and creased mortality and necrotizing enterocolitis. There was
were estimated by subtracting the number of infants who no reduction in BPD with lower SpO2 targets. Although
died from the published results for death or disability. there were fewer infants treated for ROP with lower SpO2
Severe visual impairment was present in 25/1,910 in- targets, ROP treatment is usually successful and there was
fants targeted to lower SpO2 and 24/1,965 infants targeted no long-term difference in severe visual impairment or any
to higher SpO2 (RR 1.13, 95% CI 0.651.97). Most of the other disability between groups. The hope that lower SpO2
trials did not report blindness due to retinal damage sep- targets might reduce ROP and BPD without any conse-
arately to that resulting from cortical visual impairment. quence was not fulfilled. The fact that higher SpO2 targets
In the BOOST-II UK trial [25] there were 4 infants (2 in increased survival without increasing morbidity should be
each group) whose severe visual impairment was due to considered quite exciting. At least in a developed world set-
retinal damage from a total of 18 cases (12 cortical, 4 ret- ting, there is no longer (as there was in the past) a trade-off
inal, 2 unknown). between mortality and blindness from ROP only between
mortality and a need for ROP treatment. This should not
be a difficult choice given the success of ROP treatment.
Discussion The advantage of higher SpO2 targets is clear. The message
is similar when the data from all oximeters combined are
The recent oxygen trials took place in a developed world considered and when the analysis is restricted to the re-
setting where SpO2 monitoring is universally available and vised oximeters; however, the difference in mortality ob-
ROP screening and treatment is carefully organized. In served after the oximeters were revised was greater.
128.111.121.42 - 6/9/2016 6:22:01 AM
Univ. of California Santa Barbara