ARTICLE IN PRESS

THE JOURNAL OF PEDIATRICS • www.jpeds.com ORIGINAL

ARTICLES
Optimal Target Range of Closed-Loop Inspired Oxygen Support in
Preterm Infants: A Randomized Cross-Over Study
Maria Elisabeth Nicoletta van den Heuvel, MD1, Henriette A. van Zanten, PhD2, Tom E. Bachman, MSc3,
Arjan B. te Pas, MD, PhD2, Anton H. van Kaam, MD, PhD1, and Wes Onland, MD, PhD1

Objective To investigate the effect of different pulse oximetry (SpO2) target range settings during automated frac-
tion of inspired oxygen control (A-FiO2) on time spent within a clinically set SpO2 alarm range in oxygen-
dependent infants on noninvasive respiratory support.
Study design Forty-one preterm infants (gestational age [median] 26 weeks, age [median] 21 days) on FiO2 >0.21
receiving noninvasive respiratory support were subjected to A-FiO2 using 3 SpO2 target ranges (86%-94%, 88%-
92%, or 89%-91%) in random order for 24 hours each. Before switching to the next target range, SpO2 was manu-
ally controlled for 24 hours (washout period). The primary outcome was the time spent within the clinically set alarm
limits of 86%-94%.
Results The percent time within the 86%-94% SpO2 alarm range was similar for all 3 A-FiO2 target ranges (74%).
Time spent in hyperoxemia was not significantly different between target ranges. However, the time spent in severe
hypoxemia (SpO2 <80%) was significantly reduced during the narrowed target ranges of A-FiO2 (88%-92%; 1.9%,
89%-91%; 1.7%) compared with the wide target range (86%-94%; 3.4%, P < .001). There were no differences between
the 88%-92% and 89-91% target range.
Conclusions Narrowing the target range of A-FiO2 to the desired median ±2% is effective in reducing the time
spent in hypoxemia, without increasing the risk of hyperoxemia. (J Pediatr 2018;■■:■■-■■).
Trial registration www.trialregister.nl: NTR4368.

P
reterm infants often require supplemental oxygen to prevent hypoxemia, a condition that has been associated with organ
damage and an increased mortality.1-3 However, too much oxygen supplementation resulting in hyperoxemia may lead
to systemic oxidative stress and long-term complications such as retinopathy of prematurity.4,5 For these reasons, oxygen
saturation is continuously measured in preterm infants with pulse oximetry (SpO2) aiming to keep it within a safe range by
manually titrating the fraction of inspired oxygen (FiO2). However, studies in preterm infants have shown that SpO2 targeting
is a clinical challenge with patients spending only 50% of the time within their intended SpO2 range because of clinical insta-
bility and the limited time nurses have to adjust the amount of oxygen.6,7 To improve SpO2 targeting, systems for automated
FiO2 (A-FiO2) control have been developed.
These closed-loop A-FiO2 systems are increasingly incorporated into standard neonatal ventilators and consist of an oxy-
genation monitoring device (pulse oximeter), gas (air/oxygen mixing) delivery device, and an algorithm that determines the
timing and magnitude of the FiO2 adjustments. Previous studies have shown that these algorithms improve maintenance of
oxygenation within an SpO2 target range compared with routine or dedicated manual FiO2 (M-FiO2) control.8-12 However, in
these studies, the SpO2 target set in the A-FiO2 arm was relatively wide and the SpO2 alarm setting was identical to the setting
used during normal clinical care. This approach may not use the full potential of the closed-loop A-FiO2 systems. The aggres-
siveness of adjustments is proportional to the deviation of the measured SpO2 from the set SpO2 limits. Therefore, narrowing
the SpO2 target range during A-FiO2 might increase the time spent within the SpO2 alarm limits even further and make the
algorithm more effective.
This study aimed to investigate the effect of different SpO2 target range set-
tings during A-FiO2 on time spent within a (wider) clinically desired SpO2 range
in oxygen-dependent infants on noninvasive respiratory support. We hypoth-
esized that narrowing the upper and lower limits of the A-FiO2 target range would From the 1Department of Neonatology, Emma Children’s
Hospital, Academic Medical Center Amsterdam;
result in an increased proportion of time spent within the (wider) clinically set 2
Department of Pediatrics, Division of Neonatology,
Leiden University Medical Center, Leiden, The
SpO2 alarm limits and reduced episodes of both hypoxemia and hyperoxemia. Netherlands; and 3Faculty of Biomedical Engineering,
Czech Technical University in Prague, Kladno, Czech
Republic
T.B. is an independent clinical research consultant and
provides services to the manufacturer of an automated
FiO2 system (CareFusion). A.v.K. received travel support
A-FiO2 Automated FiO2 and a speaker fee from CareFusion. The other authors
declare no conflicts of interest.
FiO2 Fraction of inspired oxygen
M-FiO2 Manual FiO2 0022-3476/$ - see front matter. © 2018 Elsevier Inc. All rights
SpO2 Pulse oximetry reserved.
https://doi.org10.1016/j.jpeds.2018.01.077

FLA 5.5.0 DTD ■ YMPD9783_proof ■ March 21, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com
Methods
This study was designed as a multicenter, randomized, cross-
over clinical trial (www.trialregister.nl: NTR4368) in 2 level III
neonatal intensive care units in The Netherlands (Emma Chil-
dren’s Hospital Academic Medical Center, Amsterdam; Leiden
University Medical Center, Leiden). The study was approved
by the institutional review boards, and written parental in-
formed consent was obtained before enrollment.
Patients
Preterm infants, born with a gestational age between 24 and
32 weeks, and a weight at study entry between 0.5 and 4 kg,
receiving nasal continuous positive airway pressure or nasal
intermittent positive pressure ventilation with a FiO2 >0.21 for
more than 18 hours per day and expected to require this non-
invasive support for the 5-day study period were considered
eligible for the study.
Infants with major congenital anomalies, hemodynamic in-
stability requiring vasopressor treatment within 48 hours prior
to enrollment, or culture proven sepsis within 72 hours prior
to enrollment were excluded.
Study Protocol
Respiratory support was provided by means of Avea ventila-
tor (CareFusion, Yorba Linda, California) with a built-in A-FiO2
adjustment function. A neonatal pulse oximeter (Radical,
Masimo Corporation, Irvine, California) integrated in the ven-
tilator was used to measure SpO2 (normal sensitivity, averag-
ing time 8 seconds). The algorithm determines the timing and
size of the FiO2 adjustments, if needed, every second. FiO2 is
reduced when SpO2 exceeds the target range or increased when
SpO2 falls below it. The magnitude of these adjustments is pro-
portional to the deviation of the measured SpO2 from the set
high or low SpO2 limit as well as the time the SpO2 has been
out of the target range. The algorithm calculates the SpO2 re-
sponse to hypoxemia and hyperoxemia in relation to the lower
and upper limits of the target range, respectively. It includes
adjustments in FiO2 based on upward or downward trends in
SpO2 that are tied to the midpoint of the target range. In ad-
dition, when in the target range, but above the midpoint, the
FiO2 is weaned slowly. Therefore, a narrow target range leads
to an amplification of the FiO2 response, for a given SpO2 value
in the hypoxemic or hyperoxemic zone. When the oximeter
identifies a condition that renders the SpO2 measurement
invalid or unreliable, in addition to creating a high priority
alarm, the FiO2 is set to a fallback level. Additional details of
the A-FiO2 function can be found elsewhere.13 This system is
approved for clinical use in The Netherlands. At the time of
the study, both sites aimed to keep the SpO2 between 86% and
94% during normal clinical practice and this SpO2 range was
also adopted for outcome assessment in this study and is here
from referred to as the SpO2 alarm range. This SpO2 alarm range
was similar for all 3 target ranges.
Enrolled infants underwent three 24-hour study periods of
A-FiO2 control using 3 different SpO2 target ranges with a fixed
midpoint of 90%: (1) 86%-94%; (2) 88%-92%; and (3) 89%-
2 van den Heuvel et al
FLA 5.5.0 DTD ■ YMPD9783_proof ■ March 21, 2018
Volume ■■ • ■■ 2018
91%. The order of SpO2 targeting was randomly assigned, using
sequentially numbered opaque sealed envelopes. To mini-
mize a possible effect of the previously assigned target range,
infants received 24 hours of standard M-FiO2 care after the first
and second target range study period, resulting in a total study
time of 120 hours. The nursing staff was unaware that data
were recorded during both M-FiO2 periods.
Demographic data including gestational age, birth weight,
sex, postnatal age, and weight were collected at time of en-
rollment. Ventilator settings and monitored measures includ-
ing SpO2, FiO2, pulse rate, and alarms were recorded every 5
seconds during the study period by an electronic data logger.
Off-line computerized analysis was used to evaluate the re-
corded data for each infant for all five 24-hour periods (3 A-FiO2
and 2 M-FiO2). Intention-to-treat analysis was applied to all
recorded data during A-FiO2 (ie, periods when the auto-
mated function was paused or the target range intermit-
tently set incorrectly were not excluded).
The primary outcome was defined as the proportion of time
spent within the (clinically) intended SpO2 alarm range of 86%-
94%, plus time with SpO2 above 94% and FiO2 was set at 0.21.
Secondary outcomes included the proportion of time below
or above the 86%-94% SpO2 alarm range, the proportion of
time below SpO2 <80% and above SpO2 >98%, and the fre-
quency of prolonged episodes (>1 minute) with SpO2 <80%
and >98% (excluding time when the FiO2 was 0.21). Data col-
lection continued during the 2 days of M-FiO2 to evaluate
M-FiO2 during usual care as a secondary outcome. Histo-
grams and median SpO2 and FiO2 levels in each of the 24-
hour periods were calculated for each infant and averaged for
all infants.
Statistical Analyses
Our hypothesis was that narrowing the target range of the
A-FiO2 adjustment to 88%-92%, and 89%-91% would in-
crease the proportion of time within the SpO2 alarm range of
86%-94% by 7%, and 10%, respectively.13 A sample size of 41
subjects achieves 80% power to detect this increase with a sig-
nificance level of 0.05. The power calculation was performed
using Nquery (Statistical Solutions Ltd, Cork, Ireland).
The primary analysis of the entire study population used a
linear mixed model with fixed effect for sequence and with the
interventions (3 A-FiO2 and 2 M-FiO2) as within subject re-
peated measures. Descriptive and secondary analyses in-
cluded within-subject comparisons with paired t tests for
normally distributed data or nonparametric Wilcoxon signed-
rank tests. Results are reported as mean and SD or median and
IQR. These analyses were conducted with XLSTAT v 6.01
(XLSTAT, Paris, France) software. P values of <.05 were con-
sidered statistically significant.
Results
Forty-three infants requiring supplemental oxygen by nonin-
vasive respiratory support were enrolled between January 2014
and October 2015. Six infants exhibited clinical deteriora-
tion shortly after the start of the study needing intubation and
■■ 2018 ORIGINAL ARTICLES

Table I. Patient characteristics at time of study Table III. Time and episodes of normoxemia, hypox-
Variables N = 41 emia, and hyperoxemia in SpO2 target subgroups*
Gestational age — wk 26 (25-27)* A-FiO2 M-FiO2
Birth weight — g 835 (692-978)* 86%-94% 86%-94%
Male sex — no. (%) 20 (49) % time in SpO2 range 74 (67-81)† 57 (50-67)
Amsterdam — no. (%) 31 (76) % time >SpO2 range 12 (5-23)† 30 (23-40)
Age at data collection — d 21 (12-29)* % time SpO2 >98% 0.4 (0.1-2)† 3.7 (2-9)
nCPAP/NIPPV — no. (%) 34 (83)/7 (17) % time <SpO2 range 14 (9-18) 14 (10-16)
FiO2 0.28 (0.25-0.31)* % time SpO2 <80% 3.4 (1.6-4.9)‡ 3.9 (2.5-5.9)
Episodes 1 min SpO2 >98%, (n/24 h) 1 (0-11)† 44 (29-86)
nCPAP, nasal continuous positive airway pressure; NIPPV, nasal intermittent positive pressure
ventilation.
Episodes 1 min SpO2 <80%, (n/24 h) 7 (2-12)† 18 (11-25)
*Median and IQR. Median FiO2 (%) 25 (22-29)‡ 27 (24-31)
Median SpO2 (%) 91 (90-92)‡ 93 (91-93)
Manual FiO2 adjustments/24h 1 (0-3)† 81 (51-115)

*Time in range includes time when SpO2 >94 and FiO2 = 0.21. Median and IQR.
invasive mechanical ventilation and, therefore, (temporarily) †P < .001, ‡P < .05 automated vs manual.
exited the study. The reasons for clinical deterioration were nec-
rotizing enterocolitis (n = 2) and sepsis (n = 4). Four of these hyperoxemia by narrowing the target range during A-FiO2
infants restarted the study once they recovered and were control to the desired median ±1%.
transitioned back to noninvasive support. Including those 4 Hypoxemia, assessed by the proportion of time spent below
infants, the 120-hour study period was completed in 41 infants the intended SpO2 alarm range (<86%), was significantly
without any adverse events. Of these infants, 7 received nasal reduced by narrowing the target range during A-FiO2 control
intermittent positive pressure ventilation and 34 nasal con- to either 88%-92% or 89%-91% (both 10%) compared with
tinuous positive airway pressure at the start of the study. The the widest 86%-94% target range (14%). Consistent with this
patient characteristics are shown in Table I. finding, the proportion of time and the number of pro-
Protocol deviations occurred in 3 infants and consisted in
longed episodes of severe hypoxemia were significantly reduced
2 patients of a short episode of incorrect SpO2 target range
by narrowing the SpO2 target range during A-FiO2 control.
setting during A-FiO2 control. In 1 patient, the SpO2 target range There were no differences between the 88%-92% and 89%-
was set at 86%-94% twice, resulting in missing data on the 89%- 91% target ranges. The increased hypoxemia during the 86%-
91% target range. In 4 cases, technical failure of the data logger 94% SpO2 target range is also illustrated in the distribution
occurred and this was corrected by adding extra time to data of SpO2 values in the Figure, showing a clear left shift com-
collection, assuring that data were available for a 24-hour mea- pared with the narrower SpO2 target ranges. The median SpO2,
surement period. FiO2, and number of manual FiO2 adjustments did not differ
Table II summarizes the primary and secondary out- between the 3 SpO2 target range subgroups.
comes in the 3e A-FiO2 target range subgroups. The propor-
Comparison of the two 24-hour periods with M-FiO2 control
tion of time spent within the 86%-94% SpO2 alarm range was
showed no significant difference in outcomes of oxygen-
not significantly different between the 3 A-FiO2 target range
ation, and for this reason all M-FiO2 periods were averaged
subgroups. This was also true for the outcomes assessing time for further analyses. Table III shows the oxygenation out-
and episodes above the SpO2 range (hyperoxemia), although comes for the subgroups of infants when assigned to the widest
there was a trend to more episodes of prolonged severe target range during A-FiO2 control (86%-94%) and M-FiO2
control using the same SpO2 alarm range. The time spent within
the intended SpO2 range significantly increased during A-FiO2
control (74%) compared with M-FiO2 control (57%, P < .001).
Table II. Time and episodes of normoxemia, hypox-
In addition, the proportion of time spent in severe hyperoxemia
emia, and hyperoxemia in SpO2 target subgroups*
was significantly reduced during A-FiO2 control compared with
A-FiO2 A-FiO2 A-FiO2 M-FiO2 control. This was reflective of a marked decrease in
86%-94% 88%-92% 89%-91%
the number of prolonged episodes spent in severe hyperoxemia
% time in SpO2 range 74 (67-81) 72 (64-83) 73 (58-85) during A-FiO2 control (1 [0-11]) compared with M-FiO2
% time >SpO2 range 12 (5-23) 19 (9-35) 17 (6-31)
% time SpO2 >98% 0.4 (0.1-1.7) 0.5 (0.2-3.0) 0.8 (0.2-3.5) control (44 [29-86], P < .001). The effect on hypoxemia was
% time <SpO2 range 14 (9-18)† 10 (6-13) 10 (6-13) less pronounced, showing only a significant reduction in the
% time SpO2 <80% 3.4 (1.6-4.9)† 1.9 (0.9-3.7) 1.7 (0.8-4.3) number of episodes with prolonged severe hypoxemia during
Episodes 1 min SpO2 >98%, 1 (0-11) 4 (0-13) 12 (0-36)
(n/24 h) A-FiO2. Consistent with these findings, the distribution of SpO2
Episodes 1 min SpO2 <80%, 7 (2-12)‡ 3 (1-6) 4 (1-9) values during manual control showed a right-sided shift com-
(n/24 h) pared with A-FiO2 control (Figure). The median FiO2 and SpO2
Median FiO2 (%) 25 (22-29) 26 (23-30) 27 (24-31)
Median SpO2 (%) 91 (90-92) 92 (91-93) 92 (91-93) over the 24-hour period was significant lower during A-FiO2
Manual FiO2 adjustments/24h 1 (0-3) 1 (0-3) 0 (0-3) compared with M-FiO 2 control. Manual adjustments of
*Time in range includes time when SpO2 > 94 and FiO2 = 0.21. Median and IQR.
FiO2 were infrequent during A-FiO2 control and significantly
†P < .001, ‡P < .01 compared with other target ranges. lower compared with M-FiO2 control.
Optimal Target Range of Closed-Loop Inspired Oxygen Support in Preterm Infants: A Randomized Cross-Over Study 3

FLA 5.5.0 DTD ■ YMPD9783_proof ■ March 21, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com
Figure. SpO2 histograms. Histogram of SpO2 exposure for the target ranges of A-FiO2 control and M-FiO2 control.
Discussion
This study shows that narrowing the target range setting during
A-FiO2 control on the Avea ventilator in preterm infants re-
ceiving noninvasive respiratory support does not result in an
increased time spent within the clinically set (wider) SpO2 alarm
range. However, narrowing of the target range of A-FiO2 control
to the desired median ±2% does reduce the time spent in hy-
poxemia without increasing the risk of hyperoxemia.
Up to now, most studies on A-FiO2 used a SpO2 target range
setting during A-FiO2 control identical to the clinically desired
SpO2 range. For instance, if the SpO2 range was 87%-93% or
90%-95% in daily clinical practice, this was also adopted as
the target range set during A-FiO2. We hypothesized that nar-
rowing this target range setting could potentially improve the
time spent within the clinically targeted (wider) SpO2 range.
There are several reasons why this hypothesis proved to be false.
First, it is possible that a narrower target range during A-FiO2
will mainly impact small “physiological” fluctuations in SpO2
that would have remained within the clinically set SpO2 alarm
range without any FiO2 adjustment. This is reflected in the small
difference in median and IQR FiO2 used between the 3 target
ranges. Therefore, automated FiO2 adjustments in response to
these fluctuations will not improve the time spent within this
SpO2 alarm range. Second, this effect may have been ampli-
fied by the fact that we included relatively stable infants in this
study. Differences between the 3 A-FiO2 target range settings
might be present in less stable subjects. Finally, the random-
ization of the different target ranges set during A-FiO2 across
a relatively long period of 5 days might have resulted in av-
eraging out small differences between them.
4 van den Heuvel et al
FLA 5.5.0 DTD ■ YMPD9783_proof ■ March 21, 2018
Volume ■■
Our study shows that narrowing the target range during
A-FiO2 control reduces the time in moderate (SpO2 < 86%) and
severe hypoxemia (SpO2 < 80%), and the number of pro-
longed episodes of severe hypoxemia. The risk of hyperoxemia,
however, was not significantly reduced by narrowing the target
range during A-FiO2; instead, we observed a trend to more epi-
sodes of prolonged hyperoxemia.
Again, these findings may be explained by the algorithm of
the A-FiO2 system. When using a narrower target range during
automated control, the FiO2 increase starts earlier and the mag-
nitude of the FiO2 increase will be greater as the SpO2 starts
to drop. If SpO2 drops below 86%, the speed of response is then
identical, only the magnitude of response differs. This may
reduce both the magnitude and the duration of the SpO2 de-
crease more effectively compared with a wider SpO2 target
range, thereby explaining the beneficial effect of a narrower
SpO2 target on hypoxemia. It is important to acknowledge that
the algorithm responds slower (15-90 seconds) to hyperoxemia
than to hypoxemia (within 10 seconds). This slower re-
sponse to a SpO2 above the target range may explain why a
narrower SpO2 target range did not reduce hyperoxemia. In
fact, narrowing the SpO2 target range during automated control
may result in more (inappropriate) increases in FiO2 in re-
sponse to small (physiological) decreases in SpO2. Such in-
creases in FiO2 may result in an overshoot of the SpO2 above
the upper alarm limit. This may explain the trend to modest
hyperoxemia in the narrowest target range. The reduction in
hypoxemia was similar in both narrowed SpO2 target ranges,
suggesting that there is no extra benefit of pursuing a very small
SpO2 target range (±1%) during A-FiO2. In fact, such extreme
narrowing might increase the risk of prolonged episodes of
hyperoxemia.
■■ 2018
One study compared a wider (87%-93%) with a narrower
(90%-93%) SpO2 target range using the same A-FiO2 system.14
Narrowing the target range resulted in significantly less time
in hypoxemia and more time in moderate hyperoxemia. Al-
though the finding of less hypoxemia is consistent with our
study, the increased risk of hyperoxemia is not. This might be
explained by the fact that our study used the same median SpO2
across all target ranges. In the study by Wilinska et al, the
median SpO2 of the narrower range (90%-93%) was higher
than the wider range (87%-93%) and this may have caused
more time in hyperoxemia.14
Despite differences in patient characteristics, mode of re-
spiratory support, and SpO2 targets, most of the findings in
this study are in agreement with other studies comparing
automated with manual control and using the same
algorithm.8-10,13-15 However, the difference in hyperoxemia
between automated and manual control, especially the number
of prolonged episodes of hyperoxemia, was more profound
in the present study compared with previous studies.13,15 This
was mainly due to significantly more hyperoxemic episodes
during manual control. This difference may have been caused
by the fact that, in contrast to other studies, the nurses were
unaware that SpO2 recording continued during the 2 manu-
ally controlled washout days. Previous studies have shown
that the nursing staff has a tendency to tolerate hyperoxemia
to avoid spells of hypoxemia.16 This finding also suggests
that studying A-FiO2 under nonblinded conditions may un-
derestimate the true treatment effects of A-FiO2 control on
SpO2 targeting.
Our study has some limitations. First, preterm infants require
supplemental oxygen for many weeks. Data collection was
evaluated during a relatively short period that may not be
representative for this prolonged period of support. Second,
data collection (5-second data logging) with a SpO2 value
averaged over 8 seconds might understate the rate of epi-
sodes of hyperoxemia and hypoxemia longer than 1 minute,
but should not affect the differences comparing the different
target ranges. Third, the infants were all supported by a non-
invasive modality and had to be clinically stable during the
entire 5 days of study. The results may differ in less stable
infants and/or on invasive respiratory support. Finally, this
study was done with 1 algorithm incorporated in 1 neonatal
ventilator. This might limit extrapolation of our results to
other ventilators. Our results would only be generalizable to
other automated oxygen control devices having a similar
algorithm to that used in the Avea ventilator, with SpO2
error calculated in relation to the upper and lower limit of
the set target range, rather than the midpoint. Future studies
need to confirm this.
Reducing time spent in hypoxemia by narrowing the SpO2
target range during automated control may have important
clinical implications. Hypoxemia is associated with an in-
creased mortality in preterm infants 2 and with adverse
neurodevelopmental outcome at 18 months.17 Our results also
suggest that using a too narrow SpO2 target range during A-FiO2
control does not improve efficacy of automated oxygen control
and might even have adverse effects. How these results trans-
Optimal Target Range of Closed-Loop Inspired Oxygen Support in Preterm Infants: A Randomized Cross-Over Study
FLA 5.5.0 DTD ■ YMPD9783_proof ■ March 21, 2018
ORIGINAL ARTICLES
late to clinical outcome will have to be addressed in future ran-
domized clinical trials.
In conclusion, this study shows that narrowing the target
range of A-FiO2 to the desired median ±2% is effective in re-
ducing the time spent in hypoxemia, without increasing the
risk of hyperoxemia. Future studies need to evaluate the impact
on clinical relevant outcomes. ■
Submitted for publication Aug 9, 2017; last revision received Dec 31, 2017;
accepted Jan 27, 2018
Reprint requests: Maria Elisabeth Nicoletta van den Heuvel MD, Department
of Neonatology (H3-113), Emma Children’s Hospital/Academic Medical
Center, PO Box 22660, 1100 DD, Amsterdam, The Netherlands. E-mail:
m.e.vandenheuvel@amc.uva.nl
References
1. Carlo WA, Finer NN, Walsh MC, Rich W, Gantz MG, Laptook AR, et al.
Target ranges of oxygen saturation in extremely preterm infants. N Engl
J Med 2010;362:1959-69.
2. Schmidt B, Whyte RK, Asztalos EV, 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;309:2111-20.
3. Stenson B, Brocklehurst P, Tarnow-Mordi W. Increased 36-week sur-
vival with high oxygen saturation target in extremely preterm infants. N
Engl J Med 2011;364:1680-2.
4. Askie LM, Darlow BA, Davis PG, Finer N, Stenson B, Vento M, 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.
5. Askie LM, Henderson-Smart DJ, Irwig L, Simpson JM. Oxygen-saturation
targets and outcomes in extremely preterm infants. N Engl J Med
2003;349:959-67.
6. Hagadorn JI, Furey AM, Nghiem TH, Schmid CH, Phelps DL, Pillers DA,
et al. Achieved versus intended pulse oximeter saturation in infants born
less than 28 weeks’ gestation: the AVIOx study. Pediatrics 2006;118:1574-
82.
7. Hagadorn JI, Sink DW, Buus-Frank ME, Edwards EM, Morrow KA, Horbar
JD, et al. Alarm safety and oxygen saturation targets in the Vermont
Oxford Network iNICQ 2015 collaborative. J Perinatol 2017;37:270-
6.
8. Claure N, Bancalari E, D’Ugard C, Nelin L, Stein M, Ramanathan R, et al.
Multicenter crossover study of automated control of inspired oxygen in
ventilated preterm infants. Pediatrics 2011;127:e76-83.
9. Claure N, D’Ugard C, Bancalari E. Automated adjustment of inspired
oxygen in preterm infants with frequent fluctuations in oxygenation: a
pilot clinical trial. J Pediatr 2009;155:640-5, e1-2.
10. Claure N, Gerhardt T, Everett R, Musante G, Herrera C, Bancalari E. Closed-
loop controlled inspired oxygen concentration for mechanically venti-
lated very low birth weight infants with frequent episodes of hypoxemia.
Pediatrics 2001;107:1120-4.
11. Hallenberger A, Poets CF, Horn W, Seyfang A, Urschitz MS. Closed-
loop automatic oxygen control (CLAC) in preterm infants: a random-
ized controlled trial. Pediatrics 2014;133:e379-85.
12. Plottier GK, Wheeler KI, Ali SK, Fathabadi OS, Jayakar R, Gale TJ, et al.
Clinical evaluation of a novel adaptive algorithm for automated
control of oxygen therapy in preterm infants on non- invasive
respiratory support. Arch Dis Child Fetal Neonatal Ed 2017;102:F37-
43.
13. van Kaam AH, Hummler HD, Wilinska M, Swietlinski J, Lal MK, te Pas
AB, et al. Automated versus manual oxygen control with different satu-
ration targets and modes of respiratory support in preterm infants. J Pediatr
2015;167:545-50, e1-2.
14. Wilinska M, Bachman T, Swietlinski J, Kostro M, Twardoch-Drozd M.
Automated FiO2-SpO2 control system in neonates requiring
5
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

respiratory support: a comparison of a standard to a narrow SpO2 control
range. BMC Pediatr 2014;14:130.
15. Lal M, Tin W, Sinha S. Automated control of inspired oxygen in venti-
lated preterm infants: crossover physiological study. Acta Paediatr
2015;104:1084-9.
16. Claure N, Bancalari E. Closed-loop control of inspired oxygen in pre-
mature infants. Semin Fetal Neonatal Med 2015;20:198-204.
17. Poets CF, Roberts RS, Schmidt B, Whyte RK, Asztalos EV, Bader D, et al.
Association between intermittent hypoxemia or bradycardia and late death
or disability in extremely preterm infants. JAMA 2015;314:595-603.

6 van den Heuvel et al

FLA 5.5.0 DTD ■ YMPD9783_proof ■ March 21, 2018