Journal of Tropical Pediatrics, 2016, 0, 1–12

doi: 10.1093/tropej/fmw100
Clinical Review

CLINICAL REVIEW

Evidence to Support Oxygen Guidelines for

Children with Emergency Signs in Developing
Countries: A Systematic Review and
Physiological and Mechanistic Analysis
by Shidan Tosif and Trevor Duke
Centre for International Child Health, University of Melbourne, Murdoch Children’s Research Institute,
Royal Children’s Hospital, Melbourne, Australia
Correspondence: Shidan Tosif, Centre for International Child Health, University of Melbourne, 50 Flemington Rd, Parkville
VIC 3052, Melbourne, Australia. E-mail <shidan.tosif@rch.org.au>.

ABSTRACT
There are currently no evidence-based oxygen saturation targets for treating children with life-
threatening conditions. We reviewed evidence of SpO2 targets for oxygen therapy in children with
emergency signs as per WHO Emergency Triage Assessment and Treatment guidelines. We system-
atically searched for physiological data and international guidelines that would inform a safe
approach. Our findings suggest that in children with acute lung disease who do not require resuscita-
tion, a threshold SpO2 for commencing oxygen of 90% will provide adequate oxygen delivery.
Although there is no empirical evidence regarding oxygen saturation to target in children with emer-
gency signs from developing countries, a SpO2 of  94% during resuscitation may help compensate
for common situations of reduced oxygen delivery. In children who do not require resuscitation or
are stable post resuscitation with only lung disease, a lower limit of SpO2 for commencing oxygen of
90% will provide adequate oxygen delivery and save resources.

K E Y W O R D S : resuscitation, post-resuscitation care, guidelines, oxygen therapy, oxygen saturation

targets, low- and middle-income countries

INTRODUCTION Triage Assessment and Treatment (ETAT) guide-

Previously, apart from oxygen therapy based
on clinical signs, the WHO has recommended a
single threshold of pulse oximetry saturation
(SpO2) of <90% for giving oxygen to children with
pneumonia. Although this is unchanged for stable
patients with pneumonia or other respiratory in-
fections, the recently updated WHO Emergency
lines now recommend a target SpO2 of 94%
during resuscitation for children with emergency
signs [1].
WHO ETAT guidelines aim to identify chil-
dren with immediately life-threatening conditions in
developing countries, analogous to Advanced
Paediatric Life Support Guidelines [2]. WHO

C The Author [2017]. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
V  1
2  Oxygen therapy for children with emergency signs
TEXTBOX 1: WHO EMERGENCY TRIAGE (ETAT) CLINICAL SIGNS
Obstructed or absent breathing
Severe respiratory distress
Central cyanosis
Signs of shock (cold hand, capillary refill >3 seconds, rapid and weak pulse, and low or un-measurable blood
pressure)
Coma
Convulsions
Signs of severe dehydration in a child with diarrhea
emergency signs relate to severe airway, breathing, cir-
culation or conscious state problems (see Textbox 1).
Oxygen therapy for hypoxemia is widely accepted
and reflected in international guidelines [3–7].
Hypoxemia is identified by arterial blood oxygen satur-
ation (SaO2) below normal ranges and is a complica-
tion of pneumonia and a risk factor for death [8–12].
Hypoxia refers to a relative deficiency of oxygen that is
unable to meet needs at the tissue level [13] and is a
complication of hypoxemia, or reduced oxygen deliv-
ery, or inability to use oxygen at a tissue level.
Children with acute respiratory infection in developing
countries are at an increased risk of hypoxemia and
death due to comorbidities and late presentation.
Non-respiratory causes of hypoxemia or tissue hypoxia
include severe sepsis [11, 12, 14], seizures, coma and
severe anemia [12]. Malnourished children are more
susceptible to infections and have poorer outcomes
with intercurrent disease [15, 16]. There is a high
prevalence of anemia in preschool-age children in de-
veloping countries [17], and anemia commonly
complicates acute infections. Combinations of lung
diseases, sepsis, nutritional cardiac impairment and ef-
fect of anemia on oxygen delivery carry high risks.
This review presents the data on the SpO2 that
should be targeted in sick children. We searched for
international guidelines and papers on physiological
data, which would provide a framework for a safe
approach.
METHODS
Search methods and sources
A systematic review protocol was used. We
searched MEDLINE (inception–2015), EMBASE
(inception–2015), physiology/scientific journals and
information in related textbook chapters for studies
comparing oxygen saturation targets in children with
emergency signs, using terms in Table 1. PubMed
was searched using keywords only to retrieve e-pubs
and items not indexed in Medline. Gray literature
was searched using the British Library, WHO, The
New York Academy of Science and Google Scholar.
Past related reviews and reference lists of relevant
articles were identified. Broad search terms were
used to maximize sensitivity.
Inclusion and exclusion criteria
Studies that addressed the PICOT question ‘In chil-
dren presenting with emergency signs, at which oxy-
gen saturation targets should oxygen therapy be
commenced, or ceased, to prevent short and long
term morbidity/mortality?’ were included for system-
atic review. Guidelines and physiological and mechan-
istic data comparing oxygen saturation targets and
physiology in children with critical illness were sys-
tematically searched and summarized to form a narra-
tive. Studies not relating to infants or children were
excluded, as well as non-English-language articles.
Data synthesis
Following query of databases, titles and abstracts
of retrieved studies were screened. For articles that
could not be excluded based on the title and abstract,
or had some possible relevance, the full text was re-
viewed. Studies from developing countries were pre-
ferred, but all relevant studies were included.
Included studies were then assessed according to
GRADE methodology, and a summary of the
 Oxygen therapy for children with emergency signs  3

Table 1. Medline search strategy

1. exp Airway Management/ or respiration/ or exp Airway Resistance/ or exp Airway Obstruction/ or exp Alkalosis,
Respiratory/ or alkalosis/ or hyperventilation/ or exp Respiratory Therapy/ or exp Acidosis, Respiratory/ or
acidosis/ or exp Respiratory System/ or exp “work of breathing”/ or exp Respiratory Tract Infections/ or exp
Respiratory Tract Diseases/ or Shock, Hemorrhagic/ or exp Shock, Traumatic/ or Shock/ or Shock, Septic/ or
Shock, Cardiogenic/ or exp unconsciousness/ or brain death/ or death/ or exp Hypoglycemia/ or exp Seizures/
or epilepsy/ or exp cyanosis/ or Dehydration/ or exp Anemia/ or exp Hypotension/ or exp diarrhea/
2. ((obstruct* adj3 (breathing or airway or respiration)) or shock or cyanosis or cyanoses or cyanotic or convul-
sion* or convulsive or coma or alkalosis or alkaloses or alkalotic or hyperventilate* or hyper-ventilate* or acidosis
or acidoses or acidotic or (respiratory adj disease*) or unconsciousness or death* or hypoglyc?emi* or seizure*
or epilep* or diarrhoea* or diarrhea* or hypotensi* or anaemi* or anemi*).tw,kf.
3. th.fs.
4. exp anoxia/
5. (anox* or hypox*).tw,kf.
6. oxygen/bl
7. exp oximetry/
8. oximetr*.tw,kf.
9. exp Oxygen Inhalation Therapy/
10. ((oxygen adj3 therap*) or (oxygen adj delivery) or (oxygen adj administration) or (oxygen adj3
supplement*)).tw,kf.
11. treatment outcome/ or exp treatment failure/
12. morbidity/ or mortality/ or child mortality/ or infant mortality/ or survival rate/
13. prognosis/
14. (mortalit* or morbidit* or survival or outcome* or death* or died).tw,kf.
15. (newborn* or neonat* or infan* or pre-schooler* or preschooler* or child* or adolescen* or pediatric* or paedi-
atric* or youth*1 or teen*).af.
16. (1 or 2 or 3) and (4 or 5 or 6 or 7 or 8) and (9 or 10) and (11 or 12 or 13 or 14) and 15
17. exp animals/ not human*.sh.
18. 16 not 17
19. limit 18 to english

Similar search terms were used for other databases with adaptations as needed

findings table was compiled. Physiological and mech-
anistic data and guidelines were compiled in a narra-
tive synthesis with findings summarized in subject
themes.
RESULTS
A total of 1824 articles were scanned from
MEDLINE, EMBASE and PubMed (see Annex 1
[18]). One randomized trial and two observational
studies were identified that described outcomes at
different oxygen saturation targets in children, and
these are described below and summarized in the
GRADE summary (Table 2).
Cunningham et al. [19] performed a double-
blind, randomized, equivalence trial of 615 patients
with bronchiolitis in the UK. Diagnosis of bronchio-
litis was made according to clinical criteria (SIGN 91
bronchiolitis [20]). In total, 308 patients were
randomized to receive oxygen if SpO2 was <94%,
compared with 307 for SpO2 <90%, using modified
oximeters. Infants admitted directly to intensive care
or high dependency units or those with underlying
lung or cardiac disease were excluded. Median time
to resolution of cough was 15.0 days in both groups,
with no difference between the two groups.
Recorded adverse events were not significantly
 4 

Table 2. GRADE Summary of Findings Table

Summary of findings:
Outcomes for SpO2 targets in children with emergency signs
Patient or population: children with emergency signs
Setting: primary care, clinics, hospitals
Intervention: oxygen
Comparison: standard therapy
Outcomes Anticipated absolute Relative effect @ of Quality of the Comments
effectsa (95% CI) (95% CI) participants evidence
Risk with standard Risk with (studies) (GRADE)
therapy oxygen
Oxygen saturation target in No difference between – 615 (1 RCT) 丣丣丣HIGHb Equivalence trial,
Oxygen therapy for children with emergency signs

bronchiolitis (comparing SpO2 standard care and no increased risk

90% vs. 94%) Assessed with: modified group for from using 90% vs.
Duration of cough median time to cough 94% SpO2 target,
resolution of 15.0 days recorded adverse
events
did not differ
significantly
Mortality with 85% SpO2 101 per 1000 66 per 1000 RR 0.65 961 丣丣LOWc
targetAssessed with: (41 to 103) (0.41 to 1.02) (1 observational
Case fatality rates in study)
pneumonia
Mortality with 90% SpO2 50 per 1000 32 per 1000 RR 0.65 11291 丣丣LOWd
targetAssessed with: (26 to 39) (0.52 to 0.78) (1 observational
Case fatality rates in study)
pneumonia
(continued)
 Oxygen therapy for children with emergency signs  5

different between both groups. The authors con-

cluded that a lower saturation target of 90% was safe
and as clinically effective as 94%.
Duke et al. [9] compared a prospectively studied
group of 703 children with severe pneumonia treated
using a pulse oximetry protocol with a retrospective
control group of 258. Diagnosis was according to crite-
ria for severe pneumonia in Papua New Guinea, consist-
ent with WHO classification for very severe pneumonia.
In the historical control group, oxygen therapy was
The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). based on clinical signs, particularly cyanosis. Using a
SpO2 lower limit of <85% for oxygen therapy in the
prospective protocol, case fatality rates were reduced
from 10 to 6.5%, and the mortality risk ratio was 0.65
(95% CI 0.41–1.02, two-sided Fisher’s exact test,
P ¼ 0.07). This study was done at 1600 m altitude.
High quality: We are very confident that the true effect lies close to that of the estimate of the effect

Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be
Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be

Duke et al. [21] demonstrated a reduction in mor-

Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially

tality by installing an improved oxygen system in five

hospitals in Papua New Guinea. Diagnoses were
close to the estimate of the effect, but there is a possibility that it is substantially different

made according to WHO criteria. This included the

introduction of pulse oximetry-guided oxygen therapy
for SpO2 < 90%. Before introduction of the system,
356 of 7161 children admitted in the five hospitals for
pneumonia died (case-fatality rate 4.97% [95% CI
Single disease population in high-income country, infants with severe disease excluded (-2 indirectness)

4.5–5.5]), whereas 133 of 4130 children died in the

Single disease cohort, high case fatality rate, one facility population, at high altitude (-2 indirectness)

27 months after the introduction of the system

(322% [2.7–3.8]). This study was done in three high-
land hospitals at an altitude of 1600-1800 m and in
two coastal hospitals at sea level.
Single disease cohort, high case fatality rate, at high altitude (-2 indirectness)

Physiological mechanisms
substantially different from the estimate of effect

These clinical studies provide some evidence for ap-

propriate oxygen therapy targets, but there is hetero-
geneity in disease etiology and setting. There were
different from the estimate of the effect
GRADE Working Group grades of evidence

no empirical data on oxygen delivery thresholds in

multi-system critically ill states; this may be different
than in children with single-organ disease such as
pneumonia or bronchiolitis. We summarize physio-
CI: Confidence interval; RR: Risk ratio.

logical mechanisms in the following sections, as a

way of understanding an approach to provision of
oxygen for children with emergency signs such as
those in ETAT guidelines.

Oxygen–hemoglobin dissociation curve

Explanations:

The relationship between oxygen saturation and the

partial pressure of oxygen in blood is not linear, illus-
trated by the sigmoid shape of the oxygen–
•
•

d
b
a

c
6  Oxygen therapy for children with emergency signs
Fig. 1. Oxygen–hemoglobin dissociation curve.
hemoglobin dissociation curve (Fig. 1). Target oxy-
gen saturations and definitions of hypoxemia are
often directed toward the ‘safe’ section of the curve,
corresponding to PaO2 above 60 mmHg and SaO2
greater than 90%. At pressures above PaO2
60 mmHg, oxygen saturation (SaO2) of blood does
not change significantly even with large changes in
oxygen partial pressure. However, below a PaO2 of
60 mmHg, small reductions in PaO2 greatly reduce
SaO2 and blood oxygen content.
Oxygen content
Oxygen content is the sum of oxygen bound to
hemoglobin (98% of all transported oxygen) and
oxygen dissolved in the plasma (2%). Oxygen con-
tent is therefore highly dependent on hemoglobin
saturation. It can be calculated using the following
equation:
Oxygen content ¼ ðHb  k1  SaO2 Þ
þ ðk2  PaO2 Þ
Where Hb is the hemoglobin concentration; k1 is
a constant to define maximum oxygen binding cap-
acity of hemoglobin (k1 is known as Hüfner’s con-
stant; the milliliter of oxygen that can be bound by 1
gram of hemoglobin at standard temperature and
pressure. The value for k1 varies between authors but
1.39 ml/g will be used for the purposes of this re-
view.); SaO2 is the percent of hemoglobin saturated
with oxygen; PaO2 is the partial pressure of oxygen
in arterial blood; k2 is a constant to define the solu-
bility of oxygen in the plasma (Solubility coefficient
of oxygen in plasma ¼ 0.0031).
This equation helps to illustrate the effect of vari-
ous physiological abnormalities on arterial oxygen
content and subsequent oxygen delivery. For example,
a healthy child may have an arterial oxygen content of
approximately 16.4 mL/dL (Hb 12 g/dL, 100% SaO2,
PaO2 105 mmHg). In contrast, in a child with acute
respiratory distress, hypoxemia and mild anemia
(Hb 9 g/dL, SaO2 88%, PaO2 60 mmHg), the arterial
oxygen content may be significantly lower, at approxi-
mately 10.8 mL/dL [22]. Whether this is sufficient to
match metabolic demands without resulting in anaer-
obic metabolism, lactic acidosis and cell metabolic
dysfunction depends on the adequacy of the cardiac
output, regional tissue perfusion and whole body and
regional tissue oxygen consumption. A cycle can be
established where the increased cardiac work that is
required to supply metabolic demands requires more
oxygen.
Oxygen delivery
Physiological homeostasis requires that oxygen deliv-
ery meet the demands of tissue oxygen consumption.
Oxygen delivery is the product of cardiac output and
blood oxygen content, which as shown above depends
on hemoglobin concentration and hemoglobin–
oxygen saturation (SaO2). Hypoxia can be caused by
inadequate transfer of oxygen across the lungs (hypo-
xemic hypoxia), decreased arterial oxygen content
(anemic hypoxia), inadequate blood flow (ischemic
hypoxia) or abnormal cellular oxygen utilization
(cytotoxic hypoxia) [13]. Conditions presenting with
WHO’s emergency signs may be associated with each
of these pathophysiological processes.
Critical oxygen saturation levels
The oxygen tension below which cellular oxidative
metabolism fails (a state called dysoxia) varies, and
multiple factors, including the specific organ or tis-
sue, cellular metabolic requirements, pH and others,
determine this. There is a strong diffusion gradient
from arteries to cells: within cells, the lower limit for
dysoxia is a PO2 between 3-15 mmHg, depending
on cell type and activity, and within mitochondria,
 Oxygen therapy for children with emergency signs  7

this threshold may be as low as 1 mmHg, and even

SaO2

98
68
50

98
93
97
(%)
0.1 mmHg within muscle cells[23].
Lumb [24] described a theoretical model for

(mmHg)
Arterial blood

36
27

67
92
100

112
determining the critical venous oxygen tension that

PaO2
Table 3. Lowest arterial oxygen levels compatible with a cerebral venous PO2 of 2.7 kPa (20 mmHg) under various conditions
is associated with dysoxia, using venous PO2 that ap-
proximates to end-capillary PO2. In this model,

O2 Content
physiological factors are adjusted to reflect oxygen

20.0
13.8
10.1

19.9
11.2
14.8
(ml/dl)
saturation at a venous PO2 of 20 mmHg and oxygen
content of 6.4 ml/dl, below which oxidative cellular
metabolism fails. These calculations suggested that

content difference
under conditions where the cardiac function and
hemoglobin are normal, the estimated arterial

Art./Ven. O2

13.5
7.4
7.4
3.7

7.4
oxygen content required to prevent dysoxia is 13.8

10
ml/dl, PaO2 36 mmHg, and this corresponds to a
SaO2 of 68%. However, in uncompensated anemia,
an arterial oxygen saturation (equivalent to SpO2) of

O2 Content
at least 93% is needed to maintain cerebral metabolic

6.4
6.4

6.4
3.8
4.8
12.6
(ml/dl)
Cerebral venous blood
demand [24]. For uncompensated cerebral ischemia,
an even higher SpO2 may be necessary. Physiological
compensatory mechanisms such as increased cere-

SvO2

63
32
32

32
32
32
bral blood flow that help to maintain cerebral oxy-

(%)
genation and other protective factors have not been
taken into consideration in these calculations. (mmHg)

33
20
20

20
20
20
PvO2

However, these examples (see Table 3) illustrate the

nature of oxygen requirement in disease states with
multiple comorbidities, such as might be seen in
blood flow
Cerebral

ml.min1

1240
620
620

340
620
460
children with co-existent lung and brain disease
(e.g. pneumonia and meningitis), or lung and circu-
latory impairment (e.g. pneumonia and sepsis), or
consumption

CNS disease (such as cerebral malaria) and severe

ml.min1
Brain O2

anemia.
46
46
46

46
46
46
Blood O2

International pediatric resuscitation guidelines

capacity
ml.dl1

20
20
20

20
12
15

Most national and international advanced pediatric

Adapted from Nunn’s Applied Physiology, 7th Edition23.

resuscitation guidelines recommend delivering high

Uncompensated arterial hypoxemia

concentration of oxygen in the first stages of resusci-

Uncompensated cerebral ischemia
Arterial hypoxemia with increased

Combined anemia and ischemia

tation, for treatment of circulatory and respiratory

failure, even prior to determination of oxygen satur-
ation [3–7], summarized in Table 4.
Uncompensated anemia

Several guidelines recommend higher oxygen sat-

cerebral blood flow

uration targets in children who are more severely un-

well. The WHO publication Oxygen Therapy for
Normal values

Children recommends higher targets (>94%) for se-

vere disease states where oxygen delivery from the
lungs to body tissues is impaired, or where vital
organs may be susceptible to low oxygen levels [25].
8  Oxygen therapy for children with emergency signs
Table 4. International guidelines describing oxygen use in resuscitation
Guideline
Resuscitation Council (UK) [6]
European Resuscitation Council [7]
American Heart Association Guidelines [4]
Australian and New Zealand Committee on
Resuscitation [5]
NICE: Assessment and initial management
of feverish illness in children younger
than 5 years [3]
DISCUSSION
There is limited evidence to determine oxygen satur-
ation targets to reduce morbidity and mortality in
children with emergency signs. Two pre–post obser-
vational studies in a developing country demonstrated
benefit from oxygen therapy, for a target SpO2
of > 85 and >90% in the treatment of children with
WHO-defined severe pneumonia. In a randomized
controlled trial for oxygen therapy for bronchiolitis in
a developed country, there was no disadvantage with
a saturation target of >90% compared with 94%.
Oxygen saturation target description
‘100% oxygen should be used for initial resuscitation. After
ROSC, titrate the inspired oxygen, using pulse oximetry,
to achieve an oxygen saturation of 94–98%’
‘Give oxygen at the highest concentration (i.e. 100%) during
initial resuscitation. . .Once the child is stabilized and/or
there is ROSC, titrate the fraction of inspired oxygen
(FiO2) to achieve normoxemia, or at least (if arterial
blood gas is not available), maintain SpO2 in the range of
94–98%.’
‘It is reasonable to ventilate with 100% oxygen during CPR
because there is insufficient information on the optimal
inspired oxygen concentration. Once the circulation is
restored, monitor systemic oxygen saturation. It may be
reasonable, when the appropriate equipment is available,
to titrate oxygen administration to maintain the oxyhemo-
globin saturation 94%.’
‘A high concentration of oxygen should be administered
during resuscitation regardless of any preceding condition.
There is insufficient evidence for choosing any concentra-
tion of oxygen during acute resuscitation. It is reasonable
to use 100% oxygen initially for resuscitation. After
ROSC, the concentration of inspired oxygen should be
reduced to a level which yields a satisfactory level of oxy-
gen in arterial blood measured by arterial blood gas ana-
lysis (PaO2 80-100 mmHg) or by percutaneous oximetry
(SpO2 94-98%).’
Oxygen should be given to children with fever who have
signs of shock or oxygen saturation (SpO2) of less than
92% when breathing air. Treatment with oxygen should
also be considered for children with an SpO2 of greater
than 92%, as clinically indicated.
In the absence of controlled trials, principles of
oxygenation are helpful in determining a safe ap-
proach by demonstrating the impact of different
physiological derangements for children with emer-
gency conditions commonly presenting in low- and
middle-income countries. Maintaining SpO2 above
90% corresponds to the safe section of the oxygen–
hemoglobin dissociation curve, and in the absence of
co-morbid derangements in oxygen will ensure ad-
equate cerebral oxygenation and avoid dysoxia.
Lumb’s calculations also indicate there is sufficient

luxury in a target SpO2 of 90%, well above critical
levels associated with dysoxia.
Most international guidelines for the management
of shock say to give 100% oxygen by face mask or
nasal prongs, and to have a target SpO2 of 94-98%.
The UK National Institute of Clinical Excellence
(NICE) guideline is the only one to specify a lower
SpO2 saturation limit in the setting of fever and
shock, with the provision that oxygen therapy should
be considered at higher SpO2 if clinically indicated
(Table 4). Most of these guidelines are used in high-
and middle-income countries, where oxygen therapy
is more available, but where underlying comorbid-
ities and late presentation are less common than in
developing countries.
Other guidelines for oxygen therapy outside the
resuscitation phase of treatment are based on eti-
ology and focus on single-organ lung disease. In the
treatment of pneumonia, for example, oxygen ther-
apy is recommended for oxygen saturations of 92%
by several guidelines [3, 26]. For bronchiolitis, the
American Academy of Pediatrics recommends oxy-
gen therapy if SpO2 < 90% [27].
Although pulse oximetry is the most reliable, non-
invasive assessment for measuring hemoglobin oxy-
gen saturation to assess hypoxemia, it also has limita-
tions. In anemic states, SpO2 may be in the normal
range despite a reduced PaO2, making the assess-
ment of anemia an important component of any as-
sessment for oxygen therapy. Depending on the
oximeter model and application, there is a variation
in accuracy of readings [28, 29], and many oximeters
are affected by movement artifact or impaired per-
ipheral perfusion. Particularly at lower saturations,
the error margin is greater [30], and this needs to be
taken into account when producing guidelines.
Studies characterizing etiology and prevalence of
hypoxemia in children presenting with WHO’s emer-
gency signs are needed. More research is required to
describe disease states with multiple comorbidities,
as those children who present with a combination of
features are most at risk of adverse cerebral oxygen-
ation. Although randomized control studies are not
possible, future implementation studies demonstrat-
ing benefit may be helpful, as part of quality im-
provement studies. As the provision of oximeters
and availability of oxygen therapy improves, clear
Oxygen therapy for children with emergency signs  9
guidelines are needed to direct this resource in de-
veloping countries.
A target saturation of  94% will compensate for
the potential of reduced oxygen delivery, which may
be more common in children with WHO’s emer-
gency signs arising from severe pneumonia, septic
shock, severe anemia, CNS infection or heart failure.
A target saturation of  94% during the resuscitation
phase may also help to compensate for the error of
the test inherent with the use of some pulse
oximeters.
There are several limitations to this review. There
were few studies identified, and in those included,
there is heterogeneity in setting and patient selec-
tion. Whilst broad search parameters were used, it is
possible that some relevant studies were not cap-
tured, and studies where the full text was not avail-
able in English were excluded. Aspects of physiologic
and mechanistic data are affected in part by assump-
tions, calculation or theories, which are not absolute
and should be interpreted in context. The quality of
evidence identified in this review ranges from low to
high, depending on the study, according to GRADE
assessment.
Controlled trials of different oxygen targets are
difficult in low-income settings, and likely to be
confounded by marked population heterogeneity,
co-morbidities and the many other determinants of
outcomes besides the SpO2 level at which oxygen
therapy is commenced. Performing randomized con-
trol trials, where oxygen is withheld for a control
group or where a much lower target is given to a
study group, would be unethical, as it is clear that
oxygen is effective in treatment of hypoxemia and
that improved oxygen systems improve quality of
care in developing countries [21, 31]. Conducting
studies to identify differences between smaller incre-
ments of target oxygen saturations would be difficult,
as these would be confounded by heterogeneity of
the populations, pulse oximeter systematic error and
prohibitively large sample sizes to measure clinically
relevant outcomes.
CONCLUSION
For management of stable patients with uncompli-
cated respiratory disease, such as severe pneumonia
10  Oxygen therapy for children with emergency signs
or bronchiolitis, a target such as SpO2  90% is ap-
propriate and consistent with WHO and other inter-
national guidelines. The recommendation of giving
oxygen to all children with WHO’s emergency signs
if SpO2 < 95% needs to be weighed against the
increased demand that is placed on resources in de-
veloping countries where oxygen supplies may be
scarce. More research is needed to support the im-
plementation of oxygen guidelines, particularly in de-
veloping countries. Greater efforts should be put
into making sufficient oxygen sources available
where all seriously ill children present.
FUNDING
This systematic review was funded by the Department of
Maternal, Newborn, Child and Adolescent Health, WHO,
Geneva.
ACKNOWLEDGEMENTS
The authors would like to acknowledge Dr Shamim Qazi and
Dr Wilson Were, Department of Maternal, Newborn, Child
and Adolescent Health, WHO, for supporting this review,
and Ms Poh Chua, medical librarian, Royal Children’s
Hospital Melbourne, for assistance with search protocols.
REFERENCES
1 WHO. Guideline: Updates on Paediatric Emergency
Triage, Assessment and Treatment. Geneva: World Health
Organization, 2015.
2 Samuels M; Advanced Life Support Group (Manchester
England). Advanced Paediatric Life Support: The Practical
Approach. 5th edn. Chichester, West Sussex, UK: Wiley-
Blackwell, 2012.
3 Fields E, Chard J, Murphy MS, et al. Assessment and initial
management of feverish illness in children younger than 5
years: summary of updated NICE guidance. Bmj
2013;346:f2866.
4 American Heart Association. 2010 & 2015 American Heart
Association Guidelines for Cardiopulmonary Resuscitation and
Emergency Cardiovascular Care. Part 12: Pediatric Advanced
Life Support, Guideline 4.1.6, 2015. Available online: https://
eccguidelines.heart.org/index.php/circulation/cpr-ecc-guide
lines-2/part-12-pediatric-advanced-life-support/ [last accessed
2nd Jan 2017]
5 ANZCOR, Introduction to Paediatric Advanced Life Support
Techniques in Paediatric Advanced Life Support. ANZCOR
Guideline 12.6, January 2016: p. 2. Available online:
https://resus.org.au/guidelines/ [last accessed 2nd Jan
2017].
6 Maconochie I, Bingham B, Skellett S. Paediatric advanced life
support guidelines. Resuscitation Council (UK), 2015.
Available online: https://www.resus.org.uk/resuscitation-
guidelines/paediatric-advanced-life-support/ [last accessed
2nd January 2017].
7 Maconochie IK, Bingham R, Eich C, et al. European
Resuscitation Council guidelines for resuscitation 2015: sec-
tion 6. Paediatric life support. Resuscitation 2015;95: 223–48.
8 Djelantik IG, Gessner BD, Sutanto A, et al. Case fatality
proportions and predictive factors for mortality among chil-
dren hospitalized with severe pneumonia in a rural develop-
ing country setting. J Trop Pediatr 2003;49:327–32.
9 Duke T, Mgone J, Frank D. Hypoxaemia in children with
severe pneumonia in Papua New Guinea. Int J Tuberc
Lung Dis 2001;5:511–19.
10 Onyango FE, Steinhoff MC, Wafula EM, et al. Hypoxaemia
in young Kenyan children with acute lower respiratory in-
fection. Bmj 1993;306:612–15.
11 Subhi R, Adamson M, Campbell H, et al. The prevalence of
hypoxaemia among ill children in developing countries: a
systematic review. Lancet Infect Dis 2009;9:219–27.
12 Chisti MJ, Duke T, Robertson CF, et al. Clinical predictors
and outcome of hypoxaemia among under-five diarrhoeal
children with or without pneumonia in an urban hospital,
Dhaka, Bangladesh. Trop Med Int Health 2012;17:106–11.
13 McLellan SA, Walsh TS. Oxygen delivery and haemoglo-
bin. Contin Edu Anaesth Crit Care Pain 2004;4:123–6.
14 DeBruin W, Notterman DA, Magid M, et al. Acute hypoxe-
mic respiratory failure in infants and children: clinical and
pathologic characteristics. Crit Care Med 1992;20:
1223–34.
15 Chisti MJ, Salam MA, Bardhan PK, et al. Treatment failure
and mortality amongst children with severe acute malnutri-
tion presenting with cough or respiratory difficulty and
radiological pneumonia. PloS One 2015;10:e0140327.
16 Man WD, Weber M, Palmer A, et al. Nutritional status of
children admitted to hospital with different diseases and its
relationship to outcome in the Gambia, West Africa. Trop
Med Int Health 1998;3:678–86.
17 McLean E, Cogswell M, Egli I, et al. Worldwide prevalence
of anaemia, WHO vitamin and mineral nutrition informa-
tion system, 1993-2005. Public Health Nutr 2009;12:
444–54.
18 Moher D, Liberati A, Tetzlaff J, et al. Group P. Preferred re-
porting items for systematic reviews and meta-analyses: the
PRISMA statement. PLoS Med 2009;6:e1000097.
19 Cunningham S, Rodriguez A, Adams T, et al. Oxygen satur-
ation targets in infants with bronchiolitis (BIDS): a double-
blind, randomised, equivalence trial. Lancet 2015; 386:1041–8.
20 Scottish Intercollegiate Guidelines. Bronchiolitis in
Children (SIGN 91). 2006.
21 Duke T, Wandi F, Jonathan M, et al. Improved oxygen sys-
tems for childhood pneumonia: a multihospital effectiveness
study in Papua New Guinea. Lancet 2008;372:1328–33.
22 Lucking SE. Pediatric Critical Care Study Guide: Text And
Review. London; New York: Springer, 2012.

23 Connett RJ, Honig CR, Gayeski TE, Brooks GA. Defining
hypoxia: a systems view of VO2, glycolysis, energetics, and
intracellular PO2. J Appl Physiol 1990;68:833–42.
24 Lumb AB. Nunn’s Applied Respiratory Physiology.
7th edn. Edinburgh: Churchill Livingstone/Elsevier, 2010.
25 World Health Organization. Oxygen Therapy in Children.
Geneva: Switzerland, 2015.
26 Harris M, Clark J, Coote N, et al. British Thoracic Society
guidelines for the management of community acquired
pneumonia in children: update 2011. Thorax 2011;
66(Suppl. 2):ii1–ii23.
27 Ralston SL, Lieberthal AS, Meissner HC, et al.
Clinical practice guideline: the diagnosis, management, and
Oxygen therapy for children with emergency signs  11
prevention of bronchiolitis. Pediatrics 2014;134:
e1474–502.
28 Van de Louw A, Cracco C, Cerf C, et al. Accuracy of pulse
oximetry in the intensive care unit. Intensive Care Med
2001;27:1606–13.
29 Seguin P, Le Rouzo A, Tanguy M, et al. Evidence for the
need of bedside accuracy of pulse oximetry in an intensive
care unit. Crit Care Med 2000;28:703–6.
30 Ross PA, Newth CJ, Khemani RG. Accuracy of pulse oxim-
etry in children. Pediatrics 2014;133:22–9.
31 Matai S, Peel D, Wandi F, et al. Implementing an oxygen
programme in hospitals in Papua New Guinea. Ann Trop
Paediatr 2008;28:71–8.
12  Oxygen therapy for children with emergency signs

Annex 1. PRISMA flow diagram [18]