ARTICLE IN PRESS

THE JOURNAL OF PEDIATRICS • www.jpeds.com ORIGINAL

ARTICLES
Preterm Infant Outcomes after Randomization to Initial Resuscitation
with FiO2 0.21 or 1.0
Valerie Thamrin1, Ola D. Saugstad, MD, PhD2, William Tarnow-Mordi, MBBS, MRCP3,4, Yueping Alex Wang, BMed, MPH, PhD5,
Kei Lui, MBBS, FRACP, MD1,6, Ian M. Wright, MBBS, MRCP (Paeds), FRACP7,8, Koert De Waal, MBBS, FRACP, PhD8,9,
Javeed Travadi, MBBS, FRACP8,9, John P. Smyth, MBBS, FRACP1,2, Paul Craven, MBBS, FRACP8,9,
Rowena McMullan, MBBS, FRACP10, Elisabeth Coates, BSc3, Meredith Ward, MBBS, FRACP1,6, Parag Mishra, MBBS, DM1,6,
Kwee Ching See, MBBS, MRCP11, Irene G. S. Cheah, MBBS, MRCP12, Chin Theam Lim, MBBS, MRCP13,
Yao Mun Choo, MBBS, MRCP13, Azanna Ahmad Kamar, MBBS, MRCP13, Fook Choe Cheah, MD, FRACP, PhD14,
Ahmed Masoud, MD15, and Ju Lee Oei, MBBS, FRACP, MD1,3,6

Objective To determine rates of death or neurodevelopmental impairment (NDI) at 2 years corrected age (primary
outcome) in children <32 weeks’ gestation randomized to initial resuscitation with a fraction of inspired oxygen (FiO2)
value of 0.21 or 1.0.
Study design Blinded assessments were conducted at 2-3 years corrected age with the Bayley Scales of Infant
and Toddler Development, Third Edition or the Ages and Stages Questionnaire by intention to treat.
Results Of the 290 children enrolled, 40 could not be contacted and 10 failed to attend appointments. Among
the 240 children for whom outcomes at age 2 years were available, 1 child had a lethal congenital anomaly, 1 child
had consent for follow-up withdrawn, and 23 children died. The primary outcome, which was available in 238 (82%)
of those randomized, occurred in 47 of the 117 (40%) children assigned to initial FiO2 0.21 and in 38 of the 121
(31%) assigned to initial FiO2 1.0 (OR, 1.47; 95% CI, 0.86-2.5; P = .16). No difference in NDI was found in 215
survivors randomized to FiO2 0.21 vs 1.0 (OR, 1.26; 95% CI, 0.70-2.28; P = .11). In post hoc exploratory analyses
in the whole cohort, children with a 5-minute blood oxygen saturation (SpO2) <80% were more likely to die or to
have NDI (OR, 1.85; 95% CI, 1.07-3.2; P = .03).
Conclusions Initial resuscitation of infants <32 weeks’ gestation with initial FiO2 0.21 had no significant effect
on death or NDI compared with initial FiO2 1.0. Further evaluation of optimum initial FiO2, including SpO2 target-
ing, in a large randomized controlled trial is needed.
Trial Registration Australian and New Zealand Clinical Trials Network Registry ACTRN 12610001059055 and
the National Malaysian Research Registry NMRR-07-685-957 (J Pediatr 2018;■■:■■-■■).

he equipoise for using higher oxygen strategies for newborn infant resuscitation has been lost.1 Concerns of oxidative

T stress have led to widespread use of lower levels of oxygen (eg, fraction
of inspired oxygen [FiO2] 0.21-0.3) to initiate respiratory support at birth,
especially for preterm infants.2 Between January 1, 2008, and June 14, 2014, the
To2rpido study (Targeted Oxygen in the Resuscitation of Preterm Infants and Their From the 1School of Women’s and Children’s Health, The
Developmental Outcome; ACTRN 12610001059055) recruited 292 infants of <32 University of New South Wales, Sydney, NSW, Australia;
2Department of Pediatric Research, The University of

weeks’ gestation in Australia, Malaysia, and Qatar to determine whether initial re- Oslo, Oslo, Norway; 3NHMRC Clinical Trials Centre, The
University of Sydney, Sydney; 4Westmead International
spiratory stabilization with FiO2 0.21 would decrease the risk of death and dis- Network for Neonatal Education and Research, Sydney;
5
Faculty of Health, University of Technology Sydney,
ability compared with FiO2 1.0. Owing to the paradigm shift away from using higher Sydney; 6Department of Newborn Care, The Royal
Hospital for Women, Sydney; 7Illawarra Health and
oxygen resuscitation strategies, the study failed to achieve the target sample size Medical Research Institute and Graduate School of
of 1976 infants (988 in each arm), and recruitment was discontinued on advice Medicine, The University of Wollongong, Wollongong;
8 Hunter Medical Research Institute, University of
Newcastle, Newcastle; 9Department of Neonatology, The
John Hunter Hospital, Newcastle; 10Department of
Neonatology, Royal Prince Alfred Hospital, Sydney, NSW,
ASQ Ages and Stages Questionnaire Australia; 11Department of Neonatology, Sungai Buloh
Hospital, Selangor, Malaysia; 12Department of
BPD Bronchopulmonary dysplasia Neonatology, Hospital Kuala Lumpur, Kuala Lumpur;
13Department of Neonatology, University of Malaya
BSID-III Bayley Scales of Infant Development, Third Edition
FiO2 Fraction of inspired oxygen Medical Centre, Kuala Lumpur; 14Department of
Paediatrics, Universiti Kebangsaan Malaysia Medical
IVH Intraventricular hemorrhage Centre, Kuala Lumpur, Malaysia; and 15Hamad Medical
NDI Neurodevelopmental injury Corporation, Doha, Qatar
RCT Randomized controlled trial Funded by the Thrasher Research Fund for Children and
ROP Retinopathy of prematurity the Leslie Stevens Fund for Newborn Research. The
authors declare no conflicts of interest.
RR Risk ratio
SpO2 Blood oxygen saturation 0022-3476/$ - see front matter. © 2018 Elsevier Inc. All rights
WPPSI Wechsler Preschool and Primary Scale of Intelligence reserved.
https://doi.org10.1016/j.jpeds.2018.05.053

FLA 5.5.0 DTD ■ YMPD10038_proof ■ July 6, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com
by the Data and Safety Monitoring Committee at 292 pa-
tients. However, a post hoc exploratory analysis of random-
ized subgroups found that infants <28 weeks’ gestation who
were initially resuscitated with FiO2 0.21 were at increased risk
of mortality compared with those resuscitated with FiO2 1.0
(22% [10 of 46] vs 6% [3 of 54]; risk ratio [RR], 3.9; 95% CI,
1.1-13.4; P = .01).3
Because the To2rpido study did not achieve the target sample
size, and given the wide 95% CIs around the estimates of effect
on the primary outcome, these results remain in the realm of
hypothesis-generating rather than demonstrating evidence of
no effect,4 and questions about optimum oxygen strategies in
preterm infant resuscitation remain. To date, 12 published
studies have examined higher oxygen vs lower oxygen strat-
egies for preterm infant resuscitation,3,5-15 but a meta-analysis
of 8 of these studies3,6-12 found no difference in the risk of major
morbidities, such as bronchopulmonary dysplasia (BPD) (RR,
0.88; 95% CI, 0.68-1.14), intraventricular hemorrhage
(IVH) grade >3 (RR, 0.81; 95% CI, 0.52-1.27), or death (RR,
0.99; 95% CI, 0.52-1.91) for infants at <29 weeks’ gestation.16
Initial FiO2 is not the sole important variable in delivery room
resuscitation. In 2010, expert committees2,17 recommended ad-
justing FiO2 to meet preductal blood oxygen saturation (SpO2)
targets derived from healthy, spontaneously breathing term and
preterm infants during the first 10 minutes of life.18,19 These
recommendations were published after recruitment for the
To2rpido study began. The study methodology did not change,
because international SpO2 recommendations vary widely and
are acknowledged to be weakly evidence-based with an un-
certain impact on long-term infant outcomes.20 In an indi-
vidual patient analysis of the 8 previously mentioned studies,3,6-12
infants who did not reach an SpO2 of 80% at 5 minutes,
whether because of inherent differences in pathology or in-
sufficient oxygen supplementation, were at increased risk for
severe IVH (OR, 2.04; 95% CI, 1.01-4.11) and death (OR, 4.57;
95% CI, 1.62-13.98).21
The impact of higher or lower oxygen strategies on long-
term infant outcomes, including neurodevelopmental impair-
ment (NDI), is uncertain. A retrospective cohort study of
preterm (<29 weeks) Canadian children found no difference
in the risk of death or severe NDI after Canadian resuscita-
tion policy was changed from FiO2 1.0 (n = 581) to FiO2 0.21
(n = 445) or intermediate fractions (FiO2 0.22-0.99; n = 483).
However, resuscitation with FiO2 1.0 was associated with an
increased risk of NDI compared with FiO2 0.21 (aOR, 1.57;
95% CI, 1.05-2.35), although this may have been reflective of
type I error, being 1 of 14 comparisons.22 Recently, Boronat
et al found no difference in rates of cerebral palsy (OR, 1.12;
95% CI, 0.26-4.84) or NDI (OR, 1.17; 95% CI, 0.60-2.30)
among 206 children enrolled in 3 multicenter European ran-
domized controlled trials (RCTs) that randomized infants <32
weeks’ gestation to initial resuscitation with FiO2 0.3 or 0.6.23
However, the confidence intervals were wide and thus incon-
sistent with substantial benefit or harm, confirming the need
for larger studies.24 In this study, we present post-hospital dis-
charge outcomes of children randomly assigned to resuscita-
tion with initial FiO2 0.21 or 1.0 from the To2rpido study.3 The
2 Thamrin et al
FLA 5.5.0 DTD ■ YMPD10038_proof ■ July 6, 2018
Volume ■■ • ■■ 2018
total number of patients recruited represented 15% of the pre-
determined sample size required to demonstrate a 20% de-
crease in relative risk (from 30% to 24%) in death or NDI by
24 months corrected age.
Methods
The To2rpido study was an international multicenter,
nonblinded, RCT conducted in 6 centers (2 in Australia, 3 in
Malaysia, and 1 in Qatar) in which 292 preterm infants re-
ceived respiratory support initiated with FiO2 0.21 or 1.0, which
was then increased by 10% each minute if SpO2 was <65%
before 5 minutes or <80% after 5 minutes, and decreased by
10% each minute if SpO2 was >95% at any time, with the aim
of reaching a minimum SpO2 of ≥80% at 5 minutes of life.
The primary outcome was death or major disability at 2 years
corrected age. The study design and secondary short-term out-
comes, including organ injury, prolonged ventilator support,
and death before hospital discharge, have been reported pre-
viously. Two infants were withdrawn before randomization,
leaving 145 infants in each arm to follow-up.3
Children were assessed as close as possible to 24 months of
age by pediatricians blinded to initial FiO2 allocation. Assess-
ments included clinical and neurologic examinations, visual
assessment, hearing assessment, and developmental assess-
ments using the Bayley Scales of Infant and Toddler Devel-
opment, Third Edition (Bayley-III),25,26 the Griffith Scale of
Infant Development,27 and the Wechsler Preschool and Primary
Scale of Intelligence (WPPSI)28 test in children aged >3 years.
The Ages and Stages Questionnaire (ASQ) was sent to fami-
lies who could not attend physical assessments.29,30 Infants with
incomplete Bayley-III tests were categorized as missing.
In nonrandomized, post hoc, exploratory analyses, associa-
tions between initial FiO2 (0.21 vs 1.0) and attainment of
5-minute SpO2 ≥80% with neurodevelopmental outcomes and
death were assessed in the whole cohort.
Children were classified as having a major disability who had
any of the following: (1) Bayley-III assessment score ≥1 SD
below the sample mean in any of the subscales (cognitive, lan-
guage, or motor)25; (2) ASQ assessment score ≥1 SD below the
sample mean in any of the subscales (communication, gross
and fine motor, problem solving, or personal social)29; (3)
Griffith27 or WPSSI28 assessment score >1 SD below the popu-
lation mean; (4) cerebral palsy at a Gross Motor Function Clas-
sification System score ≥2 (range, 1-5); (5) any abnormality
on neurological examination; (6) any hearing impairment; or
(7) bilateral blindness. Hearing loss with or without a hearing
aid was considered abnormal, but correctable visual impair-
ment such as myopia was considered normal.
Statistical Analyses
Infants were stratified at randomization by gestation (<28 and
28-31 + 6 weeks).3 Continuous data were compared using the
Student t test; nonparametric data, using the Mann-Whitney
U test. Categorical variables were compared using the c2 test
(or Fisher exact test if any cell had <5 patients). In post hoc
exploratory analyses within the whole cohort, the associa-
■■ 2018
tions between death or measures of major disability and FiO2,
SpO2, gestational age, birth weight, sex, severe IVH (grade >3),31
and bronchopulmonary dysplasia (BPD; defined as need for
oxygen and/or respiratory support at 36 weeks corrected
age),32,33 were assessed by logistic regression analyses. A P value
of <.05 was considered statistically significant. Bonferroni cor-
rection was not used for multiple comparisons.34 Analyses were
performed with SPSS version 22 (IBM, Armonk, New York).35
Results
A total of 290 infants were randomized to initial resuscita-
tion with FiO2 0.21 (n = 145) or FiO2 1.0 (n = 145). Forty chil-
dren, including 9 from Qatar and 13 from Malaysia, could not
be contacted despite multiple attempts. Families of 10 chil-
dren were contacted but did not attend follow-up appoint-
ments (Figure). Children who were followed up had higher
5-minute Apgar scores, higher SpO2 at 5 minutes, and older
mothers were more likely to have BPD. None of the children
lost to follow-up had IVH grade ≥3 (Table I; available at
www.jpeds.com).
Twenty children died before hospital discharge, and 3 died
after discharge, 1 each from sudden infant death syndrome at
3 months, pertussis at 6 months, and effects of a house fire at
12 months. Neurodevelopmental data were not obtained from
2 of the surviving children, including a child with a postna-
tal diagnosis of Trisomy 21 and a child for whom consent for
follow-up was withdrawn (Figure). Neurodevelopmental as-
sessments were conducted in 215 of the 240 survivors (91%).
Of these, 69 were <28 weeks’ gestation and 110 were between
28-31 + 6 weeks’ gestation; 102 were randomized to FiO2 0.21,
Figure. Patient flow chart.
Preterm Infant Outcomes after Randomization to Initial Resuscitation with FiO2 0.21 or 1.0
FLA 5.5.0 DTD ■ YMPD10038_proof ■ July 6, 2018
ORIGINAL ARTICLES
and 113 were randomized to FiO2 1.0. Bayley-III25 assess-
ments were conducted in 185 children. The WPPSI28 was used
in 2 children who fell outside the age range of Bayley-III testing.
The Griffiths test27 was performed on one child because of local
expertise. The ASQ29 was administered to 11 children who could
not attend the Bayley-III test at the scheduled time. Three chil-
dren were unable to complete the Bayley-III due to fatigue or
behavioral problems. Sixteen children were assessed with a neu-
rological examination only.
There was no difference in the risk of the primary outcome
of survival without disability or death or disability between
the children resuscitated with FiO2 0.21 and those resusci-
tated with FiO2 1.0 (Table II), regardless of gestational age
(Table III). SpO2 ≥80% was achieved by a significantly higher
proportion of infants in the FiO2 1.0 group compared with the
FiO2 0.21 group (84% [114 of 136] vs 42% [59 of 140]; OR,
7.11; 95% CI, 4.04-12.54; P < .001). At 5 minutes, SpO2 was
undetectable in 13 infants, resulting in 104 with SpO2 ≥80%
and 173 with SpO2 <80%. Overall, children with SpO2 < 80%
at 5 minutes were more likely than those with SpO2 ≥80% to
die or be disabled (OR, 1.85; 95% CI, 1.07-3.21; P = .03). More
mature infants (28-31 + 6 weeks’ gestation) had lower cogni-
tive Bayley-III scores when the 5-minute SpO2 was <80% (mean
[SD], 100.8 [12.5] vs 95.4 [12.4]; P = .02) (Table IV).
After adjustment, initial FiO2, 5-minute SpO2, male sex, ges-
tational age <29 weeks, IVH, and BPD were not associated with
risk of death or disability. However, 5-minute SpO2 ≥80% (aOR,
0.50; 95% CI, 0.26-0.98; P = .04) and gestational maturity (28-
31 + 6 weeks; aOR, 0.44; 95% CI, 0.23-0.86; P = .01) were as-
sociated with decreased risk, and IVH grade ≥3 was associated
with increased risk (aOR, 2.84; 95% CI, 1.32-6.12; P = .008)
of death/disability (Table III).
3
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Table II. Rates of primary outcome of death or major disability at 2 years corrected age for infants resuscitated with
initial FiO2 0.21 vs 1.0
Infants randomized Infants randomized
Outcomes to initial FiO2 0.21 to initial FiO2 1.0 OR (95% CI) P value
Deceased or major disability (primary outcome), n/N (%)
All patients 47/117 (40) 38/121 (31) 1.47 (0.86-2.5) .16
<28 wk gestation 22/40 (55) 20/46 (23) 1.59 (0.68-3.73) .29
28-31 + 6 wk gestation 25/77 (33) 18/75 (24) 1.52 (0.75-3.11) .25
Deceased (excludes children unable to be followed up at 2 y), n/N (%)
All patients 15/117 (13) 8/121 (7) 2.08 (0.85-5.10) .11
<28 wk gestation 11/40 (28) 6/26 (13) 1.26 (0.40-3.98) .69
28-31 + 6 wk gestation 4/77 (5) 2/75 (3) 2.00 (0.36-11.26) .43
Major disability (survivors), n/N (%)
All patients 32/102 (31) 30/113 (27) 1.26 (0.70-2.28) .44
<28 wk gestation 11/29 (38) 14/40 (35) 1.13 (0.42-3.06) .80
28-31 + 6 wk gestation 21/73 (29) 16/73 (22) 1.44 (0.68-3.05) .34
Cerebral palsy (GMFCS >2), n/N (%)
All patients 1/102 (1) 5/113 (4) 0.21 (.02-1.86) .16
<28 wk gestation 0 1/40 (1) — —
28-31 + 6 wk gestation 1/73 (1) 4/73 (5) 0.24 (.03-2.20) .20
Blind (uncorrectable visual impairment), n/N (%)
All patients 0 1/113 (1) — —
<28 wk gestation 0 1/40 (1) — —
28-31 + 6 wk gestation 0 0 — —
Hearing impairment (any), n/N (%)
All patients 2/102 (2) 3/113 (3) 0.73 (0.12-4.48) .74
<28 wk gestation 0 1/40 (1) — —
28-31 + 6 wk gestation 2/73 (3) 2/73 (3) 1.00 (0.14-7.29) 1.00
Bayley-III score <85 at 2 y corrected age, n/N (%)
All patients
Cognitive score <85 13/81 (16) 8/98 (8) 2.15 (0.84-5.48) .11
Language score <85 18/81 (22) 19/95 (20) 1.14 (0.55-2.36) .72
Motor score <85 13/80 (16) 17/94 (18) 0.88 (0.40-1.94) .75
<28 wk gestation
Cognitive score <85 5/24 (21) 4/35 (11) 2.04 (0.49-8.56) .33
Language score <85 6/24 (10) 9/34 (15) 0.93 (0.28-3.07) .90
Motor score <85 5/23 (9) 9/33 (16) 0.74 (0.21-2.59) .64
28-31 + 6 wk gestation
Cognitive score <85 8/57 (14) 4/63 (6) 2.40 (0.68-8.47) .17
Language score <85 12/57 (21) 10/61 (16) 1.36 (0.54-3.44) .52
Motor score <85 8/57 (14) 8/61 (13) 1.08 (0.37-3.10) .88
Bayley-III score, survivors, mean (SD)
All patients
Cognitive 97.0 (14.0) 98.1 (13.0) — .60
Language 95.1 (16.1) 94.3 (15.0) — .68
Motor 95.0 (15.7) 94.1 (14.8) — .74
<28 wk gestation
Cognitive 95.0 (16.4) 95.4 (14.2) — .91
Language 92.9 (15.1) 92.6 (18.1) — .94
Motor 91.6 (22.3) 91.7 (18.3) — .99
28-31 + 6 wk gestation
Cognitive 97.9 (12.9) 99.6 (12.2) — .46
Language 96.0 (16.5) 95.3 (13.1) — .78
Motor 96.4 (12.2) 95.4 (12.4) — .65

GMFCS, Gross Motor Function Classification System.

Any disability is defined as any composite BSID score <85, any delay on any neurologic test, any hearing loss, blindness, or cerebral palsy with GCFMS score >2. Includes BSID, n = 185; WIPPSI,
n = 2; Griffiths, n = 1; ASQ, n = 11; neurologic examination, n = 16.

Table III. Association of clinical variables with death or

disability (primary outcome) in all patients Discussion
Variables aOR 95% CI P value
Initiating respiratory support for infants <32 weeks’ gesta-
FiO2 1.0 0.71 0.36-1.37 .31 tion with either FiO2 0.21 or 1.0 was not associated with any
5-min SpO2 ≥80% 0.50 0.26-0.98 .04*
Male sex 1.69 0.93-3.09 .08
difference in the primary composite outcome of death or dis-
Gestational age 28-31 + 6 wk 0.44 0.23-0.86 .01* ability. This failure to show a difference may reflect the study’s
IVH grade ≥3 2.84 1.32-6.12 .008* lack of power to address the primary hypothesis (ie, type II
BPD 1.08 0.53-2.12 .84
error), because it randomized only 290 of the planned sample
Significant values are in bold type. of 1976 infants.3,35 Thus, our results should be viewed as
4 Thamrin et al

FLA 5.5.0 DTD ■ YMPD10038_proof ■ July 6, 2018

■■ 2018 ORIGINAL ARTICLES

Table IV. Rates of death and/or major disability in relation to SpO2 at 5 minutes after birth: nonrandomized compari-
sons within the whole trial cohort
Outcomes 5-min SpO2 <80% 5-min SpO2 ≥80% OR (95% CI) P value
Death or major disability, n/N (%)
All patients 41/91 (45) 42/137 (30) 1.85 (1.07-3.21) .03
<28 wk gestation 20/37 (54) 21/45 (47) 1.34 (0.56-3.22) .51
28-31 + 6 wk gestation 21/54 (39) 21/92 (22) 2.15 (1.03-4.47) .04
Deceased (excludes children unable to be followed up at 2 y), n/N (%)
All patients 13/91 (14) 10/137 (7) 2.12 (0.89-5.06) .09
<28 wk gestation 9/37 (24) 8/45 (18) 1.49 (0.51-4.34) .47
28-31 + 6 wk gestation 4/54 (7) 2/92 (2) 3.60 (0.64-20.35) .15
Major disability (survivors), n/N (%)
All patients 28/78 (36) 32/127 (25) 1.66 (0.90-3.07) .10
<28 wk gestation 11/28 (39) 13/37 (35) 1.19 (0.43-3.30) .73
28-31 + 6 wk gestation 17/50 (34) 19/90 (21) 1.93 (0.89-4.17) .10
Cerebral palsy (GMFCS >2), n/N (%)
All patients 1/71 (1) 3/105 (3) 0.49 (.05-4.77) .54
<28 wk gestation 1/26 (4) 0 — —
28-31 + 6 wk gestation 0 3/74 (4) — —
Blind (uncorrectable visual impairment), n/N (%)
All patients 0 0 — —
<28 wk gestation 0 0 — —
28-31 + 6 wk gestation 0 0 — —
Hearing impairment (any), n/N (%)
All patients 2/71 (3) 2/105 (2) 1.49 (0.21-10.85) .69
<28 wk gestation 0 1/31 — —
28-31 + 6 wk gestation 2/45 (4) 1/74 (1) 3.40 (0.30-38.56) .32
Bayley-III score <85 at 2 y corrected age, n/N (%)
All patients
Cognitive score <85 13/70 (19) 8/102 (8) 2.68 (1.05-6.86) .04
Language score <85 16/70 (23) 20/99 (20) 1.17 (0.56-2.46) .67
Motor score <85 13/69 (19) 16/98 (16) 1.19 (0.53-2.67) .67
<28 wk gestation
Cognitive score <85 5/26 (19) 4/31 (13) 1.61 (0.38-6.74) .52
Language score <85 6/26 (23) 8/30 (27) 0.83 (0.24-2.79) .76
Motor score <85 6/25 (24) 7/29 (24) 0.99 (0.28-3.47) .99
28-31 + 6 wk gestation
Cognitive score <85 8/44 (18) 4/71 (6) 3.72 (1.05-13.21) .04
Language score <85 10/44 (23) 12/69 (17) 1.40 (0.55-3.58) .49
Motor score <85 7/44 (16) 9/69 (13) 1.26 (0.43-3.67) .67
Bayley-III score, survivors, mean (SD)
All patients
Cognitive 95.4 (13.7) 99.2 (13.4) .07
Language 93.0 (15.7) 95.7 (15.4) .26
Motor 94.5 (12.9) 94.3 (17.0) .93
<28 weeks' gestation
Cognitive 95.3 (15.9) 95.6 (14.8) .94
Language 92.7 (16.6) 93.5 (17.6) .86
Motor 94.8 (15.9) 89.1 (23.1) .31
28-31 + 6 weeks' gestation
Cognitive 95.4 (12.4) 100.8 (12.5) .02
Language 93.2 (15.4) 96.7 (14.5) .22
Motor 94.3 (11.0) 96.4 (13.3) .38

Major disability is defined as any composite BSID score <85, any delay on any neurologic tests, any hearing loss, blindness, or cerebral palsy with GCFMS score >2. Significant values are in bold
type. Infants without a 5-minute SpO2 reading were excluded from the analysis.

hypothesis-generating and may help inform the design of future
appropriately-powered studies.
Recruitment failure was due primarily to a paradigm shift
against the use of higher FiO2 for infant delivery room
resuscitation.1 The recruitment period for the To2rpido
study spanned major changes in expert committee
recommendations, including SpO2 targets for the first 10
minutes of life and recommendations for the use of lower
(FiO2 ≤0.3) instead of higher (FiO2 >0.65) oxygen levels to
initiate preterm infant resuscitation.2,17 We did not change
study methodology because even now, SpO2 targets vary
Preterm Infant Outcomes after Randomization to Initial Resuscitation with FiO2 0.21 or 1.0
widely,20 and the implications of SpO2 targeting remain
unclear.
In post hoc exploratory analyses of the whole cohort, not
reaching SpO2 80% by 5 minutes was associated with an overall
increased risk of death and/or major disability in all infants
and of lower cognitive scores in more mature infants. Whether
this was due to insufficient provision of oxygen therapy or
because adequate oxygenation was impossible owing to lung
disease is unclear and merits further study. However, at least
60 multiple comparisons were reported in addition to the
primary outcome, and it is possible that this was a chance
5

FLA 5.5.0 DTD ■ YMPD10038_proof ■ July 6, 2018

THE JOURNAL OF PEDIATRICS • www.jpeds.com
finding. Thus, we acknowledge that our results should be viewed
with caution.35
There is emerging information suggesting that SpO2 in the
first 10 minutes of life is important in influencing preterm
infant outcomes. An individual participant meta-analysis of
706 infants <32 weeks’ gestation from 8 RCTs resuscitated with
lower (FiO2 <0.3) vs higher (FiO2 ≥0.6) initial oxygen showed
a significantly increased risk of IVH grade ≥3, bradycardia (heart
rate <100 bpm), and death if SpO2 80% was not attained by
5 minutes.21 Severe IVH increases the risk of NDI,32 but only
5 surviving infants in our study had severe IVH. This could
reflect a causal association between saturation and outcome,
may indicate that low saturation is a marker of the severity of
illness, or may be an iatrogenic response to insufficient oxygen
therapy. In future studies, novel monitoring techniques, such
as near-infrared spectroscopy, may be beneficial to guide oxygen
delivery in the first few minutes of life. Near-infrared spec-
troscopy data show that early cerebral hypoxia, but not
hyperoxia,36 during the immediate transition (eg, 72 hours after
birth) may increase the risk of brain hemorrhage and injury.37
We made every attempt to locate the patients after hospi-
tal discharge. Most of the patients lost to follow-up were from
Malaysia and Qatar, which have more itinerant populations
compared with Australia. Boronat et al and Saugstad et al found
follow-up more challenging in patients from lower-resourced
countries.23,38 Whether patients from Malaysia and Qatar would
have had different neurodevelopmental outcomes to patients
from Australia will remain speculative, but certainly children
who were lost to follow-up were significantly less likely to have
BPD or severe IVH before hospital discharge, which are both
independent risk factors for NDI.32 Children lost to follow-
up also had younger mothers than those who were followed
up, an established risk for poorer neurodevelopmental
outcomes in high-risk populations.39 Nonetheless, we recom-
mend that trials continue to recruit patients from lower-
resourced countries. The consequences of poor resuscitation
practice may be even more devastating in children from de-
veloping countries than in those from higher-resourced coun-
tries, where facilities and infrastructure are available to address
and assist disability.40
To ensure optimum follow-up, several forms of develop-
mental assessments were used. The validity of these studies has
been assessed in previous studies, and most have been found
to overestimate rather than underestimate disability com-
pared with the Bayley-III,25-27 including the Griffiths,27 which
was used in 1 child because of a lack of local expertise in per-
forming the Bayley-III in a rural setting. The ASQ was used
in 11 children who were unable to attend a recruitment center
for a physical assessment. Most of these patients lived some
distance away from the 2 Australian hospitals. Age-appropriate
ASQs have been shown to improve patient compliance and
reduce the cost of follow-up.41,42 The ASQ may also predict
certain aspects of the Bayley-III that are conducted at older
ages, thereby obviating the need for more extensive, time-
consuming, and costly follow-up when the child is older. For
example, in a prospective evaluation of 141 very low birth
weight (>1250 g) infants, Agarwal et al noted negative pre-
6 Thamrin et al
FLA 5.5.0 DTD ■ YMPD10038_proof ■ July 6, 2018
Volume ■■
dictive values of >90% for major disability at 24 months and
>98% for motor disability from 9 months between the ASQ
and Bayley-III.43
We acknowledge that a lack of uniformity in developmen-
tal assessments (using Bayley-III, ASQ, and Griffith test results)
is a limitation of our study. One of our aims was to obtain the
most developmental data we could from the initial 292 infants
recruited. Ideally, in future studies, one developmental assess-
ment should be used to yield a more accurate result. Consid-
eration may be given to the use of lower-cost methods, such
as the ASQ, at earlier ages to improve follow-up rates and de-
crease costs of future studies.43
Finally, the major limitation of this study was the limited
sample size. Oxygenation strategies in preterm infants during
the first minutes of life remain an open question. There is in-
creasing evidence indicating that initial FiO2 is not the sole
factor influencing preterm infant outcomes and other physi-
ological measures, such as SpO221 and heart rate,44 may be as
or even more important. Even though our study showed no
difference in the primary outcome of the risk of death and/
or major disability in infants <32 weeks’ gestation after initial
resuscitation with FiO2 0.21 or 1.0, comparison analysis of SpO2
demonstrated increased death and disability at a 5-minute SpO2
<80%. This was particularly the case in more mature infants,
who were less likely to die before hospital discharge. The impact
of this hypothesis-generating finding should be examined in
large RCTs designed to incorporate relevant current practice,
including FiO2 titration and SpO2 targeting. Large samples of
thousands rather than hundreds of infants will be needed to
reliably demonstrate reductions in mortality, major morbid-
ity, or disability.24 ■
We thank the Data and Safety Monitoring Committee (Nicholas Evans
[chair], Val Gebski, and Wendy Hague) for their support and advice and
research officers Andreja Fujek and Joanna Michalowski for their as-
sistance with data collection.
Submitted for publication Dec 11, 2017; last revision received Apr 28, 2018;
accepted May 31, 2018
Reprint requests: Ju Lee Oei, MBBS, FRACP, MD, Department of Newborn
Care, The Royal Hospital for Women, Barker St, Randwick, NSW 2031,
Australia. E-mail: j.oei@unsw.edu.au.
References
1. Oei JL, Ghadge A, Coates E, Wright IM, Saugstad OD, Vento M, et al. Cli-
nicians in 25 countries prefer to use lower levels of oxygen to resuscitate
preterm infants at birth. Acta Paediatr 2016;105:1061-6.
2. Perlman JM, Wyllie J, Kattwinkel J, Atkins DL, Chameides L, Goldsmith
JP, et al. Part 11: Neonatal resuscitation: 2010 International Consensus
on Cardiopulmonary Resuscitation and Emergency Cardiovascular Care
Science with treatment recommendations. Circulation 2010;122(16 Suppl
2):S516-38.
3. Oei JL, Saugstad OD, Lui K, Wright IM, Smyth JP, Craven P, et al. Tar-
geted oxygen in the resuscitation of preterm infants, a randomized clini-
cal trial. Pediatrics 2017;139:pii: e20161452.
4. Tarnow-Mordi WO, Duley L, Field D, Marlow N, Morris J, Newnham J,
et al. Timing of cord clamping in very preterm infants: more evidence is
needed. Am J Obstet Gynecol 2014;211:118-23.
5. Lundstrøm KE, Pryds O, Greisen G. Oxygen at birth and prolonged ce-
rebral vasoconstriction in preterm infants. Arch Dis Child Fetal Neona-
tal Ed 1995;73:F81-6.
■■ 2018
6. Wang CL, Anderson C, Leone TA, Rich W, Govindaswami B, Finer NN.
Resuscitation of preterm neonates by using room air or 100% oxygen.
Pediatrics 2008;121:1083-9.
7. Escrig R, Arruza L, Izquierdo I, Villar G, Sáenz P, Gimeno A, et al. Achieve-
ment of targeted saturation values in extremely low gestational age neo-
nates resuscitated with low or high oxygen concentrations: a prospective,
randomized trial. Pediatrics 2008;121:875-81.
8. Vento M, Moro M, Escrig R, Arruza L, Villar G, Izquierdo I, et al. Preterm
resuscitation with low oxygen causes less oxidative stress, inflammation,
and chronic lung disease. Pediatrics 2009;124:e439-49.
9. Rabi Y, Singhal N, Nettel-Aguirre A. Room-air versus oxygen adminis-
tration for resuscitation of preterm infants: the ROAR study. Pediatrics
2011;128:e374-81.
10. Kapadia VS, Chalak LF, Sparks JE, Allen JR, Savani RC, Wyckoff MH. Re-
suscitation of preterm neonates with limited versus high oxygen strat-
egy. Pediatrics 2013;132:1488-96.
11. Aguar M, Cubells E, Escobar J, Kuligowski J, Iriondo M, Izquierdo M, et al.
Preterm babies randomly assigned to be blindly resuscitated with higher
(60%) vs lower (30%) initial FIO2: effects on oxidative stress and mor-
tality. Pediatric Academic Societies/Asian Society for Pediatric Research
Joint Meeting. May 3-6, 2014; Vancouver, British Columbia, Canada;
3843.540.
12. Rook D, Schierbeek H, Vento M, Vlaardingerbroek H, van der Eijk AC,
Longini M, et al. Resuscitation of preterm infants with different in-
spired oxygen fractions. J Pediatr 2014;164:1322-6.e3.
13. Kumar VH, Carrion V, Wynn KA, Nielsen L, Reynolds AM, Ryan RM.
Oxygen resuscitation and oxidative stress biomarkers in premature infants.
Res Rep Neonatol 2014;4:91-9.
14. Ezaki S, Suzuki K, Kurishima C, Miura M, Weilin W, Hoshi R, et al. Re-
suscitation of preterm infants with reduced oxygen results in less oxida-
tive stress than resuscitation with 100% oxygen. J Clin Biochem Nutr
2009;44:111-8.
15. Harling AE, Beresford MW, Vince GS, Bates M, Yoxall CW. Does the use
of 50% oxygen at birth in preterm infants reduce lung injury? Arch Dis
Child Fetal Neonatal Ed 2005;90:F401-5.
16. Oei JL, Vento M, Rabi Y, Wright I, Finer N, Rich W, et al. Higher or lower
oxygen for delivery room resuscitation of preterm infants below 28 com-
pleted weeks gestation: a meta-analysis. Arch Dis Child Fetal Neonatal
Ed 2017;102:F24-30.
17. American Heart Association and the American Academy of Pediatrics.
Neonatal Resuscitation Program. Available at: https://www.aap.org/en-
us/continuing-medical-education/life-support/NRP/Pages/NRP.aspx. Ac-
cessed October 16, 2017.
18. Mariani G, Dik PB, Ezquer A, Aguirre A, Esteban ML, Perez C, et al. Pre-
ductal and post-ductal O2 saturation in healthy term neonates after birth.
J Pediatr 2007;150:418-21.
19. Dawson JA, Kamlin CO, Vento M, Wong C, Cole TJ, Donath SM, et al.
Defining the reference range for oxygen saturation for infants after birth.
Pediatrics 2010;125:e1340-7.
20. Wilson A, Vento M, Shah PS, Saugstad O, Finer N, Rich W, et al. A review
of international clinical practice guidelines for the use of oxygen in the
delivery room resuscitation of preterm infants. Acta Paediatr 2017;107:20-
27.
21. Oei JL, Finer NN, Saugstad OD, Wright IM, Rabi Y, Tarnow-Mordi W,
et al. Outcomes of oxygen saturation targeting during delivery room
stabilisation of preterm infants. Arch Dis Child Fetal Neonatal Ed 2017.
doi:10.1136/archdischild-2016-312366. [Epub ahead of print].
22. Soraisham AS, Rabi Y, Shah PS, Singhal N, Synnes A, Yang J, et al.
Neurodevelopmental outcomes of preterm infants resuscitated with dif-
ferent oxygen concentrations at birth. J Perinatol 2017;37:1141-7.
23. Boronat N, Aguar M, Rook D, Iriondo M, Brugada M, Cernada M, et al.
Survival and neurodevelopmental outcomes of preterms resuscitated with
different oxygen fractions. Pediatrics 2016;138:e20161405.
24. Tarnow-Mordi W, Cruz M, Morris J. Design and conduct of a large ob-
stetric or neonatal randomized controlled trial. Semin Fetal Neonatal Med
2015;20:389-402.
Preterm Infant Outcomes after Randomization to Initial Resuscitation with FiO2 0.21 or 1.0
FLA 5.5.0 DTD ■ YMPD10038_proof ■ July 6, 2018
ORIGINAL ARTICLES
25. Bayley N. Bayley Scales of Infant and Toddler Development, 3rd ed (Bayley-
III). San Antonio (TX): Pearson; 2005.
26. Goh SKY, Tham EKH, Magiati I, Sim L, Sanmugam S, Qiu A, et al. Analy-
sis of item-level bias in the Bayley-III language subscales: the validity and
utility of standardized language assessment in a multilingual setting. J
Speech Lang Hear Res 2017;60:2663-71.
27. Picciolini O, Squarza C, Fontana C, Giannì ML, Cortinovis I, Gangi S,
et al. Neurodevelopmental outcome of extremely low birth weight infants
at 24 months corrected age: a comparison between Griffiths and Bayley
Scales. BMC Pediatr 2015;15:139.
28. Bode MM, D’Eugenio DB, Mettelman BB, Gross SJ. Predictive validity
of the Bayley, Third Edition at 2 years for intelligence quotient at 4 years
in preterm infants. J Dev Behav Pediatr 2014;35:570-5.
29. Twombly E, Nickel R, Clifford J, Murphy K, Hoselton R, Potter L, et al.
Ages and Stages Questionnaires, 3rd ed (ASQ®-3). Baltimore (MD): Brookes
Publishing; 2009.
30. Rubio-Codina M, Araujo MC, Attanasio O, Muñoz P, Grantham-
McGregor S. Concurrent validity and feasibility of short tests currently
used to measure early childhood development in large scale studies. PLoS
One 2016;11:e0160962.
31. Thome UH, Genzel-Boroviczeny O, Bohnhorst B, Schmid M, Fuchs H,
Rohde O, et al. Neurodevelopmental outcomes of extremely low
birthweight infants randomised to different PCO2 targets: the PHELBI
follow-up study. Arch Dis Child Fetal Neonatal Ed 2017;102:F376-82.
32. Wilson-Costello D, Walsh MC, Langer JC, Guillet R, Laptook AR, Stoll
BJ, et al. Impact of postnatal corticosteroid (PNS) use on
neurodevelopment at 18-22 months adjusted age: effects of dose, timing,
and risk of bronchopulmonary dysplasia in extremely low birth weight
infants (ELBW). Pediatrics 2009;123:e430-7.
33. Shennan AT, Dunn MS, Ohlsson A, Lennox K, Hoskins EM. Abnormal
pulmonary outcomes in premature infants: prediction from oxygen re-
quirement in the neonatal period. Pediatrics 1988;82:527-32.
34. Perneger TV. What’s wrong with Bonferroni adjustments. BMJ
1998;316:1236-8.
35. Wang R, Lagakos SW, Ware JH, Hunter DJ, Drazen JM. Statistics in
medicine–reporting of subgroup analyses in clinical trials. N Engl J Med
2007;357:2189-94.
36. Plomgaard AM, Alderliesten T, Austin T, van Bel F, Benders M, Claris
O, et al. Early biomarkers of brain injury and cerebral hypo- and hyperoxia
in the SafeBoosC II trial. PLoS One 2017;12:e0173440.
37. Pichler G, Schmölzer GM, Urlesberger B. Cerebral tissue oxygenation
during immediate neonatal transition and resuscitation. Front Pediatr
2017;5:29.
38. Saugstad OD, Ramji S, Irani SF, El-Meneza S, Hernandez EA, Vento M,
et al. Resuscitation of newborn infants with 21% or 100% oxygen: follow-
up at 18 to 24 months. Pediatrics 2008;112:296-300.
39. Sutcliffe AG, Barnes J, Belsky J, Gardiner J, Melhuish E. The health and
development of children born to older mothers in the United Kingdom:
observational study using longitudinal cohort data. BMJ 2012;345:e5116.
40. Wilkinson DJ, Singh M, Wyatt J. Ethical challenges in the use of thera-
peutic hypothermia in Indian neonatal units. Indian Pediatr 2010;47:387-
93.
41. Bian X, Xie H, Squires J, Chen CY. Adapting a parent-completed,
socioemotional questionnaire in China: the Ages and Stages Question-
naires: social-emotional. Infant Ment Health J 2017;38:258-66.
42. Mackin R, Ben Fadel N, Feberova J, Murray L, Nair A, Kuehn S, et al. ASQ3
and/or the Bayley-III to support clinicians’ decision making. PLoS One
2017;12:e0170171.
43. Agarwal PK, Shi L, Daniel LM, Yang PH, Khoo PC, Quek BH, et al.
Prospective evaluation of the Ages and Stages Questionnaire 3rd
Edition in very-low-birthweight infants. Dev Med Child Neurol
2017;59:484-9.
44. Kapadia V, Oei J, Saugstad OD, Rabi Y, Finer NN, Wright IM, et al.
BradyPrem study: Is heart rate the most vital of all vital signs in preterm
resuscitation? Abstract. Proceedings of the Joint European Neonatal So-
cieties meeting. October 2017; Venice, Italy.
7
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume ■■

Table I. Characteristics of surviving patients eligible for follow-up

Follow-up assessed Not assessed
Characteristics (n = 215) (n = 50) OR (95% CI), P value
Gestation, wk, mean (SD) 28.4 (1.9) 28.9 (1.9) P = .10
Birth weight, g, mean (SD) 1150 (330) 1246 (336) P = .07
FiO2 0.21, n (%) 117 (54) 26 (52) 1.10 (0.60-2.04), P = .87
5-min SpO2 <80%, n (%) 91 (42) 12 (25) 2.32 (1.15-4.69), P = .02
Male sex, n (%) 109 (51) 33 (66) 0.53 (0.28-1.01), P = .05
5-min Apgar score, median (range) 9 (5-10) 9 (2-10) P = .04
Maternal age, y, mean (SD) 31.5 (6.6) 28.5 (5.9) P = .002
Cesarean delivery, n (%) 150 (70) 36 (72) 0.90 (0.45-1.78), P = .76
BPD, n (%) 64 (30) 7 (14) 2.60 (1.11-6.09), P = .03

Descriptive data for the infants in each randomized group have been reported previously. Those withdrawn from follow-up are excluded. See the Figure for more data. Significant values are in
bold type.

7.e1 Thamrin et al

FLA 5.5.0 DTD ■ YMPD10038_proof ■ July 6, 2018