Original Research

Journal of Intensive Care Medicine

1-9
Mortality in Pediatric Acute Respiratory ª The Author(s) 2017
Reprints and permission:
sagepub.com/journalsPermissions.nav
Distress Syndrome: A Systematic Review DOI: 10.1177/0885066617705109
journals.sagepub.com/home/jic
and Meta-Analysis

Judith Ju-Ming Wong, MBBCh, BAO, MRCPCH1,2, Mark Jit, BSc, PhD, MPH3,4,
Rehena Sultana, MSc5, Yee Hui Mok, MBBS, MRCPCH2,6,
Joo Guan Yeo, MBBS, MRCPCH2,6, Jia Wen Janine Cynthia Koh, GCE A-Levels7,
Tsee Foong Loh, MBBS, MRCPCH2,6, and Jan Hau Lee, MBBS, MRCPCH, MCI2,6

Abstract
Objective: Sparse and conflicting evidence exists regarding mortality risk from pediatric acute respiratory distress syndrome
(ARDS). We aimed to determine the pooled mortality in pediatric ARDS and to describe its trend over time.
Data Sources and Study Selection: MEDLINE, EMBASE, and Web of Science were searched from 1960 to August 2015.
Keywords or medical subject headings (MESH) terms used included “respiratory distress syndrome, adult,” “acute lung injury,” “acute
respiratory insufficiency,” “acute hypoxemic respiratory failure,” “pediatrics,” and “child.” Study inclusion criteria were (1) pediatric
patients aged 0 days to 18 years, (2) sufficient baseline data described in the pediatric ARDS group, and (3) mortality data. Randomized
controlled trials (RCTs) and prospective observational studies were eligible. Data Extraction and Synthesis: Data on study
characteristics, patient demographics, measures of oxygenation, and mortality were extracted using a standard data extraction
form. Independent authors conducted the search, applied the selection criteria, and extracted the data. Methodological quality of
studies was assessed. Meta-analysis using a random-effects model was performed to obtain pooled estimates of mortality. Meta-
regression was performed to analyze variables contributing to change in mortality over time. Eight RCTs and 21 observational studies
(n ¼ 2274 patients) were included. Pooled mortality rate was 24% (95% confidence interval [CI]: 19-31). There was a decrease in
mortality rates over 3 epochs (2000, 2001-2009, and 2010: 40% [95% CI: 24-59], 35% [95% CI: 21-51], and 18% [95% CI: 12-26],
respectively, P < .001). Observational studies reported a higher mortality rate than RCTs (27% [95% CI: 24-29] versus 16% [95% CI: 12-
20], P < .001). Earlier year of publication was an independent factor associated with mortality. Conclusion: Overall mortality rate in
pediatric ARDS is approximately 24%. Studies conducted and published later were associated with better survival.

Keywords
systematic review, meta-analysis, acute respiratory distress syndrome, acute lung injury, mortality, pediatric

1
Department of Pediatrics, KK Women’s and Children’s Hospital, Singapore,
Introduction Singapore
2
Duke-NUS Medical School, Singapore, Singapore
Traditionally, acute respiratory distress syndrome (ARDS) in 3
Department of Infectious Disease Epidemiology, London School of Hygiene &
children is diagnosed using the American-European Consensus Tropical Medicine, London, United Kingdom
4
Conference (AECC) criteria.1 More recently, the Berlin and the Modelling and Economics Unit, Public Health England, London, United
Kingdom
Pediatric Acute Lung Injury Consensus Conference (PALICC) 5
Centre for Quantitative Medicine, Duke-NUS Medical School, The Academia,
definitions have been proposed.2,3 Studies on pediatric ARDS Singapore, Singapore
report an incidence of approximately 1% to 4% of all pediatric 6
Children’s Intensive Care Unit, Department of Pediatric Subspecialties, KK
intensive care unit (PICU) admissions.4-6 Compared to those Women’s and Children’s Hospital, Singapore, Singapore
7
without lung injury, critically ill children with ARDS have Yong Loo Lin School of Medicine, National University of Singapore, Singa-
pore, Singapore
significantly increased morbidity and mortality.7 Short-term
consequences include increased duration of mechanical venti- Received December 10, 2016. Received revised March 7, 2017. Accepted
lation, PICU, and hospital stay.8 Long-term consequences for publication March 27, 2017.
include diminished lung function and exercise tolerance,
Corresponding Author:
reduced quality of life, and diminished neurocognitive func- Judith Ju-Ming Wong, Department of Pediatrics, KK Women’s and Children’s
tion.9-11 These patients also incur a significant financial burden Hospital, 100 Bukit Timah Road, 229899 Singapore, Singapore.
due to hospitalization costs and pediatric intensive care Email: [email protected]
2 Journal of Intensive Care Medicine XX(X)

therapies.12 Moreover, mortality in pediatric ARDS is high, J.H.L.): definition used, demographics, baseline measures of
with previous estimates varying widely between 22% and oxygenation, ventilator settings, and mortality. We adapted the
65%.6,13,14 data collection form from one that was previously utilized in a
Conflicting data exist regarding the mortality trend of pedia- prior systematic review performed in adult ARDS.18 Disagree-
tric ARDS over time.14-17 A recent systematic review and ments were resolved by consensus.
meta-analysis in pediatric ARDS reported a pooled mortality A priori, we defined eligible studies to be those that meet
of 33.7%.17 In contrast to adult data, this systematic review did the following criteria: (1) pediatric patients aged 0 days to
not demonstrate declining mortality rate in children with 18 years, (2) sufficient baseline data described in the ARDS
ARDS over time.18 In addition, this pediatric systematic review group, and (3) survival data in the ARDS group. Any definition
found that geographical location had an impact on mortality of of ARDS was accepted as long as the label of ARDS was used.
children with ARDS. This systematic review was performed We deemed a study to have “sufficient baseline data” if age,
using a search strategy that included retrospective study severity scores, oxygenation measures (eg, partial pressure of
designs. Because previous investigators have demonstrated that arterial oxygen:fraction of inspired oxygen ratio [P/F ratio] and
data from randomized controlled trials (RCTs) usually report oxygenation index [OI]), and risk factors for ARDS were
lower mortality due to stringent selection criteria, we postulate reported. We included all RCTs and prospective observational
that the choice of included study designs is an important influ- studies. Studies involving adults, specific pediatric populations
ence on the mortality outcome.18 (eg, bronchiolitis), post hoc analysis of another existing RCT
In view of limitations in existing medical literature, we (if there were no duplicate data), database/registry studies, and
conducted a systematic review using a comprehensive search studies that involved a retrospective arm and a prospective arm
strategy and performed a meta-analysis to (1) determine the were included if they fulfilled the previously mentioned inclu-
current overall mortality of ARDS in critically ill children, sion criteria. Studies involving neonatal/perinatal respiratory
(2) evaluate whether mortality has changed over time, and distress, prematurity, retrospective studies, abstracts, studies
(3) determine whether study design was independently associ- with less than 20 patients, and studies on follow-up of ARDS
ated with mortality. survivors were excluded. The primary outcome was the pro-
portion of death among patients with ARDS. For primary out-
come, we also planned subgroup analysis a priori: year of
Materials and Methods publication and study design.
This systematic review and meta-analysis was conducted in Methodological quality of RCTs was assessed using the
close accordance with the Preferred Reporting Items for Sys- Cochrane Collaboration’s tool for assessing risk of bias.20 The
tematic Reviews and Meta-Analyses guidelines.19 We identi- quality of observational studies was assessed using the RTI
fied the Patient, Intervention, Control, Outcome, Study design item bank that addresses the risk of bias, confounding, and
as critically ill children with ARDS, with the primary outcome precision at the study level.21
of interest being mortality. With the help of an experienced
librarian, the search strategy was developed using the appro-
priate keywords and MESH terms. MESH terms used include Statistical Analysis
“respiratory distress syndrome, adult,” “acute lung injury,” We performed a meta-analysis using random-effects model to
“acute respiratory insufficiency,” “acute hypoxemic respira- obtain DerSimonian-Laird pooled estimates of mortality and
tory failure,” “pediatrics,” “child,” “infant,” and “adolescent.” associated exact binomial 95% confidence intervals (95% CIs).
We did not use “mortality” as a search term because many When pooling the estimates of mortality, we excluded the treat-
studies that report clinical outcomes were not indexed with ment arm of the RCTs because all randomized treatments were
“mortality” as a keyword or MESH term. A complete descrip- not standard of care.18 A continuity correction of 0.5 was added
tion of the search strategy for PubMed/MEDLINE is outlined to zero counts. The I2 statistics was used to quantify hetero-
in Supplemental Appendix 1. Language was not restricted. geneity across studies, and an I2 statistics of 50% or more
Studies up to August 15, 2015, were included. indicates a considerable amount of heterogeneity. Separate
Two authors (J.J-M.W., J.H.L.) independently carried out an pooled estimates were then obtained for studies classified
online search in MEDLINE, EMBASE, and Web of Science according to (1) year of publication (2000, 2001-2009,
using the predetermined keywords and MESH terms from 1960 2010) and (2) study design (observational vs RCT). We also
to August 2015. The authors screened potential studies by obtained a cumulative forest plot–based year of publication.
examining the title and abstract. The full-text articles of the The simultaneous effect of both variables was explored using
shortlisted studies were then assessed for eligibility and data a random-effects logistic meta-regression. Univariate and mul-
extracted independently. Reference lists of obtained articles tivariate random-effects meta-regressions were performed to
and relevant review articles were also hand searched. We con- assess for other potential covariates that could predict the het-
tacted authors of non-English publications and publications erogeneity and summarized as odd ratio (OR) with correspond-
with insufficient baseline data to obtain missing data that were ing 95% CI. The included covariates were year of publication,
not reported. A standardized data collection form was used to study design, and geographical location. Using the PF ratio as a
extract the following data by 3 authors (J.J-M.W., J.C.J.K., marker of severity of lung disease, we also conducted a
Wong et al 3

immunodeficiency virus exposure, patients with ARDS who

had the complication of pulmonary barotrauma, and children
with immunosuppression.28,35,40,47 Of note, 1 study utilized
very high tidal volumes (22 + 8 mL/kg) and ventilator pres-
sures (peak inspiratory pressures up to 73 + 15 cm water and
positive end-expiratory pressures 14 + 4 cm water).34
Most studies (24 [83%] of 29) used the AECC definition for
ARDS (Table 1 and Supplemental eTable2). The remaining
(5 [17%] of 29) used the following to define ARDS: the
description of “diffuse alveolar disease with hypoxemia,” lung
injury score >2.5, and OI >12; 2 studies did not provide any
specific definition except the label of acute lung injury (ALI)/
ARDS.27-29,34,41 The degree of hypoxemia as described by the
P/F ratio and OI across studies was mostly in the moderate to
severe category.
All included RCTs had moderate to low risk of bias (Sup-
Figure 1. Preferred Reporting Items for Systematic Reviews and plemental Digital Content, eTable 3). Two studies did not state
Meta-Analyses (PRISMA) flowchart of the number of studies in each
clear primary and secondary outcome measures.28,29 In addi-
stage of search.
tion, these 2 studies used no specific definition except the label
of ALI. Three studies were stopped early. Two RCTs studying
stratified analysis to assess the effect of time on studies with
the role of surfactant in pediatric ARDS was stopped early due
mean/median PF ratio 100 and >100.
to slow patient recruitment.22,23 One RCT studying prone posi-
Bubble plots were performed based on year of publication
tioning was stopped early due to futility.24 Only 1 RCT was
stratified by study design. Bubble sizes in those plots were
powered to detect a statistically significant outcome.25
proportional to the total patient recruited in the study. Linear
Included observational studies had a moderate to high risk of
regression lines were fit to estimate the mortality rate over the
bias (Supplemental Digital Content, eTable 4). Only 2 studies
year of publication stratified by study design and geographical
reported both standard ventilation and weaning strategies.45,48
location separately. We performed sensitivity analyses exclud-
Nine studies carried out some form of stratification or multi-
ing smaller studies with <50 patients. Statistical analysis was
variate analysis.8,12-14,35,37-39,47 Follow-up was complete or near
performed using Statistical Analysis System (SAS version 9.3;
complete in most studies except in Yapicioğlu et al, which cen-
SAS Institute, Cary, North Carolina) and Comprehensive
sored data after 14 days.42 Two studies by Kong et al were not
Meta-Analysis software (version 3). A P value of <.05 was
explicit in the duration of follow-up and the type of
considered to indicate statistical significance.
mortality.36,37

Results Mortality in Pediatric ARDS

The search from the 3 major databases yielded a total of 4026 Overall pooled mortality including the control arm of RCTs
studies and 1 more was found from hand searching reference and all observational studies was 24% (95% CI: 19-31; Figure
lists. Twenty-nine studies were finally included (Figure 1). 2). Mortality reported in observational studies was higher than
Excluded studies with reasons were summarized in Supple- in RCTs (27% [95% CI: 24-29] vs 16% [12-20], P < .001).
mental Table 1. There were 8 RCTs and 21 observational stud- Mortality in studies published in years 2000, 2001 to 2009,
ies with a total number of 2274 patients with ARDS. and 2010 onward showed stepwise improvement: 40% (95%
Four RCTs included in this systematic review studied the CI: 24-59), 35% (95% CI: 21-51), and 18% (95% CI: 12-26);
efficacy of surfactant and 1 RCT each studied the efficacy of P < .001, respectively (Figure 3).
prone positioning, inhaled nitric oxide, immunomodulatory In the multivariate meta-regression model, a later year of
feeds, and a lung recruitment maneuver, respectively study was significantly associated with survival (adjusted odds
(Table 1).22-29 There were 2 RCTs that studied the same patient ratio [OR]: 0.94; 95% CI: 0.94-0.95; P < .001). When stratified
cohort. One study reported the use of bovine surfactant (cal- according to the type of study design, a later year of study for
factant) and the other reported the impact of fluid balance on RCTs (OR: 0.88; 95% CI: 0.86-0.91; P < .001) and observa-
outcomes; we have included only the former in this meta- tional studies (OR: 0.95; 95% CI: 0.94-0.95; P < .001) were
analysis due to overlapping patient population.26,30 still significantly associated with survival (Figure 4). Com-
There were 21 observational studies (Supplemental Table 2). pared to RCT, observational study design itself was not asso-
There was 1 non-randomized controlled study, 1 population- ciated with increased mortality (adjusted OR: 0.91; 95% CI:
based study, 12 cohort studies, and 7 descriptive studies on 0.83-1.01; P ¼ .062). There were insufficient randomized stud-
ARDS.8,13,14,16,31-46 Four studies included only specific popu- ies conducted in each continent to include geographical loca-
lations: postcardiopulmonary bypass, infants with human tion into the meta-regression. Instead, a summary of studies
4 Journal of Intensive Care Medicine XX(X)

Table 1. Characteristics of Included Randomized Controlled Trials.

Study Definition of ARDS Risk Factors Intervention Control Primary Outcome

Moller et al22 AECC Mixed Bovine surfactant (Alveofact; Standard of care (n ¼ 16) P/F ratio at 48 hours
n ¼ 22) Mean P/F ratio: (reported 30-day
Mean P/F ratio: 71.3 + 13.7 64.3 + 16.2 mortality)
Willson et al23 AECC Predominantly Bovine surfactant Air placebo (n ¼ 75) 28 VFD (reported
pulmonary (Calfactant; n ¼ 77) Mean P/F ratio: hospital mortality)
Mean P/F ratio: 128 + 54 126 + 73
Mean OI: 20.0 + 12.9 Mean OI: 20.5 + 14.7
Curley et al24 AECC Predominantly Prone nursing (n ¼ 51) Supine nursing (n ¼ 51) 28 VFD (reported
pulmonary Mean P/F ratio: 153 + 65 Mean P/F ratio: hospital mortality)
Mean OI: 18 + 18 147 + 60
Mean OI: 15 + 12
Cui et al28 No specific Cardiopulmonary Lung recruitment maneuver Standard of care (n ¼ 32) Unclear
definition except bypass only (n ¼ 32) Mean P/F ratio:
the label of ALI Mean P/F ratio: 109 + 5.4 112.5 + 10.2
Thomas et al25 AECC; S/F ratio Predominantly Synthetic surfactant Air placebo (n ¼ 81) Duration of mechanical
250 pulmonary (Lucinactant; n ¼ 84) Mean P/F ratio: ventilation (reported
Mean P/F ratio: 211 + 157 178 + 79 14-day mortality)
Mean OI: 9.2 + 5.0 Mean OI: 9.8 + 6.4
Willson et al26 AECC Predominantly Bovine surfactant Air placebo (n ¼ 53) 90-day mortality
pulmonary (Calfactant; n ¼ 56) Median P/F ratio
Median P/F ratio: 120 (IQR 125 (IQR: 115-140)
115-135)
Jacobs et al29 No specific Mixed EPA þ GLA feeds (n ¼ 19) Pediasure (n ¼ 18) Unclear
definition except Mean P/F ratio: 176.9 + 19.4 Mean P/F ratio:
the label of ALI (SEM) 202.3 + 22.7 (SEM)
Mean OI: 14.5 + 2.7 (SEM) Mean OI: 11.5 + 2.1
(SEM)
Bronicki et al27 OI  12 Mixed Inhaled nitric oxide (n ¼ 26) Air placebo (n ¼ 29) 28 VFD (reported
Mean OI: 22.0 + 8.4 Mean OI: 25.6 + 14.9 28-day mortality)
Abbreviations: AECC, America-European Consensus Conference; ARDS, acute respiratory distress syndrome; EPA, eicosapentaenoic acid; GLA, gamma-linolenic
acid; IQR, interquartile range; OI, oxygenation index (mean airway pressure  fraction of inspired oxygen)/partial pressure of inspired oxygen; P/F ratio, partial
pressure of inspired oxygen:fraction of inspired oxygen ratio; S/F ratio, oxygen saturation:fraction of inspired oxygen ratio; SEM, standard error of mean;
VFD, ventilator-free days.

according to geographical location without taking into account review highlighted the paucity of good-quality data in pediatric
study design showed improving mortality trend over time in ARDS. The majority of included studies was observational and
North America, Europe, and Asia (Supplemental Figure 1). had small sample sizes.
In studies with PF ratio >100, a later year of study had an Our pooled mortality rate of pediatric ARDS (24%; 95% CI:
OR of 0.94 (0.94-0.95), P < .001, and in studies with PF ratio 19-31) is lower than previously reported in another meta-
100, a later year of study had an OR of 0.98 (0.97-0.99), analysis (34%; 95% CI: 29-40). 17 In addition, although
P < .001. There were no RCTs with a mean/median PF ratio Schouten et al demonstrated lack of improvement of mortality
100, and hence, we could not examine the impact of type of over time, we found a declining mortality rate over time in
study design with this stratification. We conducted a sensitivity children with ARDS.17 We postulate that the differences in
analysis including only studies with 50 patients (n ¼ 12). There findings may be due to methodological differences. In the
remained a decrease in the odds of mortality with a later year of search strategy of the prior systematic review, investigators
study (adjusted OR: 0.98; 95% CI: 0.97-0.98; P < .001) and included the search term “mortality.” We made an assumption
RCT design (adjusted OR: 0.73; 95% CI: 0.67-0.81; P < .001). that some studies that reported mortality outcomes may not
include “mortality” as a keyword. In this current systematic
review that did not use “mortality” as a keyword, we managed
Discussion to identify an additional 8 RCTs that were not included in the
To date, our meta-analysis involved the largest number of prior published systematic review.
patients with pediatric ARDS. It confirmed previous findings Another methodological difference is the inclusion of retro-
of high mortality risk (pooled mortality of 24%) in children spective studies (18 of the 32 studies were retrospective) and
with ARDS. In addition, we demonstrated that mortality rates studies with less than 20 patients in the report by Schouten et al.
declined over time regardless of study design. Our systematic A subgroup analysis by Schouten et al found no significant
Wong et al 5

Figure 2. Cumulative forest plot of mortality in all included studies.

difference in mortality between retrospective and prospective We demonstrated that there is an improving mortality trend
designs.17 However, this finding must be interpreted in the in pediatric ARDS. There are several plausible explanations for
context that retrospective studies are inherently exposed to this. In the past, the incidence of pediatric ARDS was under-
patient and treatment selection biases.49 We excluded studies estimated as most patients were diagnosed according to their
with less than 20 patients because we assumed that these cen- underlying diseases.52 With increasing awareness, there is pos-
ters see a low volume of patients with ARDS and may affect the sibly an increase in prompt recognition/diagnosis of pediatric
pooled mortality adversely.50,51 A sensitivity analysis with a ARDS enabling earlier treatment. Moreover, since the advent
more stringent threshold of more than 50 patients showed con- of peripheral capillary oxygen saturation (SpO2)-based indices,
sistent results demonstrating decreasing mortality over time clinicians are able to diagnose ARDS without the need for
and now RCT design as well. However, this could have created measurement of arterial oxygenation.53,54 As a result, studies
a bias in the assessment of ARDS mortality as the number of over time may have included a patient population with lower
PICUs across the world that does not see much ARDS is not disease severity and thus lower mortality. We attempted to
known. We postulated, a priori, that there would be a difference address this confounder by stratifying studies with a mean/
in mortality between RCTs and prospective observational stud- median PF ratio 100 and >100. However, in our stratified
ies, but we did not find any difference. Conduct and reporting analysis based on the PF ratio, consistent decrease in mortality
in RCTs are tightly governed by checklists and guidelines; risk was demonstrated in later studies. The improving mortality
hence, we assume information generated is more robust.49 Ran- trend may also signify the development of effective ARDS treat-
domized controlled trials apply more stringent inclusion/exclu- ments. For example, the use of lower tidal volumes (6-8 mL/kg)
sion criteria and study protocols to improve safety and has been shown to reduce mortality in adult ARDS and is now
maximize therapeutic effects.18,50 This may inadvertently a standard of care in pediatric patients.3,55 Lastly, improve-
exclude certain subgroups of patients (eg, bone marrow trans- ments in mortality may reflect general improvements in
plant patients or extracorporeal membrane oxygenation) that pediatric critical care support.56
may affect reported mortality rate. We decided to report the One of the main strengths of this systematic review is the
overall mortality in pediatric ARDS without including the inclusion of a comprehensive search conducted without the
treatment arm of the RCTs because these treatments were col- search term “mortality,” which identified many relevant studies
lectively not the current standard of care in pediatric ARDS.3,18 that would have otherwise been missed. For example, the
6 Journal of Intensive Care Medicine XX(X)

Figure 3. Forest plot of mortality by time periods.

Figure 4. Bubble plot showing mortality rate against the year of study publication stratified by study design. Bubble sizes in those plots were
proportional to the total patient recruited in the study.
Wong et al 7

Conclusion
The overall pooled mortality rate in pediatric ARDS is approx-
imately 24% and is lower than previously published reports.
Earlier year of publication was an independent factor adversely
affecting mortality in pediatric ARDS regardless of study
design.

Acknowledgments
The authors thank Ms Peggy Fong Bih Yuh, head librarian of KK
Women’s and Children’s Hospital, Singapore, for assisting with the
formulation of the search strategy. The authors also thank Dr Jason
Phua from the National University Hospital, Singapore, for sharing the
data collection form used for a previous adult ARDS review that was
adapted for use in this systematic review.18

Figure 5. World map showing the distribution of studies across Declaration of Conflicting Interests
continents. Red stars signify the number of randomized controlled The author(s) declared no potential conflicts of interest with respect to
trials. Blue pentagons signify the number of prospective observational the research, authorship, and/or publication of this article.
studies. All countries involved in multicenter studies were counted
separately to reflect the research energy in each continent. Funding
The author(s) received no financial support for the research, author-
MEDLINE search generated 2558 articles with the current ship, and/or publication of this article.
search strategy and this subsequently yielded 23 relevant stud-
ies that were included. When search terms such as “mortality,” Supplemental Material
“child mortality,” “hospital mortality,” or “infant mortality” The online [appendices/data supplements/etc] are available at http://
were added, only 801 articles were generated and this yielded journals.sagepub.com/doi/suppl/10.1177/0885066617705109.
only 8 relevant studies.
It is also timely to determine the pooled mortality following References
the publication of the PALICC group recommendations 1. Bernard GR, Artigas A, Brigham KL, et al. The American-
that promote optimization and consistency of care for children European Consensus Conference on ARDS. Definitions, mechan-
with pediatric ARDS so as to provide a baseline for compar- isms, relevant outcomes, and clinical trial coordination. Am J
ison for future trials.3 However, the findings of our review Respir Crit Care Med. 1994;149(3):818-824. PubMed PMID:
should be interpreted in the context of the limitations. Firstly, 7509706.
many studies did not report important details like ventilation 2. De Luca D, Piastra M, Chidini G, et al. The use of the Berlin
strategies (eg, tidal volumes, prone positioning) and nonpul- definition for acute respiratory distress syndrome during infancy
monary treatments (eg, sedation, paralysis, nutrition, fluid and early childhood: multicenter evaluation and expert consensus.
management). Additionally, due to the small number of eli- Intensive Care Med. 2013;39(12):2083-2091. PubMed PMID:
gible studies, we were unable to enter these important clinical WOS:000326694900003.
variables into the meta-regression model as well as other vari- 3. Pediatric Acute Lung Injury Consensus Conference Group. Pedia-
ables such as geographical location and case mix. The studies tric acute respiratory distress syndrome: consensus recommenda-
also reported various mortality rates (eg, PICU mortality, tions from the Pediatric Acute Lung Injury Consensus
hospital mortality, and 60-day mortality) that could have cre- Conference. Pediatr Crit Care Med. 2015;16(5):428-439.
ated bias in our pooled analysis. Secondly, studies in this PubMed PMID: 25647235.
meta-analysis were highly heterogeneous. Some studies 4. Erickson S, Schibler A, Numa A, et al. Acute lung injury in
included enrolled patients with specific risk factors (eg, car- pediatric intensive care in Australia and New Zealand: a prospec-
diopulmonary bypass), while other studies included all tive, multicenter, observational study. Pediatr Crit Care Med.
patients. Thirdly, there is possible publication bias since the 2007;8(4):317-323.
funnel plot generated is mildly skewed (Supplemental Digital 5. Dahlem P, van Aalderen WMC, Hamaker ME, Dijkgraaf MGW,
Content, eFigure 2). This may be contributed by heterogeneity Bos AP. Incidence and short-term outcome of acute lung injury in
mentioned earlier and poor methodological quality of the mechanically ventilated children. Eur Respir J. 2003;22(6):
smaller studies; 2 studies did not state a clear primary and 980-985. PubMed PMID: WOS:000187009900022.
secondary outcome measure, and 5 studies used unconven- 6. Wong JJ, Loh TF, Testoni D, Yeo JG, Mok YH, Lee JH. Epide-
tional criteria to define ARDS. Lastly, there is a paucity of miology of pediatric acute respiratory distress syndrome in singa-
good-quality data in pediatric ARDS, with the majority of pore: risk factors and predictive respiratory indices for mortality.
studies originating from North America (Figure 5). This can Front Pediatr. 2014;2:78. PubMed PMID: 25121078. Pubmed
potentially limit generalizability of the results. Central PMCID: PMC4110624.
8 Journal of Intensive Care Medicine XX(X)

7. Costil J, Cloup M, Leclerc F, et al. Acute respiratory distress Evidence-based Practice Center under Contract No. 290-2007-
syndrome (ARDS) in children: multicenter Collaborative Study 10056-I) Rockville, MD: Agency for Healthcare Research and
of the French Group of Pediatric Intensive Care. Pediatr Pulmo- Quality; August, 2013.
nol Suppl. 1995;11:106-107. PubMed PMID: 7547318. 22. Moller JC, Schaible T, Roll C, et al. Treatment with bovine sur-
8. Barreira ER, Munoz GO, Cavalheiro PO, et al. Epidemiology and factant in severe acute respiratory distress syndrome in children: a
outcomes of acute respiratory distress syndrome in children randomized multicenter study. Intensive Care Med. 2003;29(3):
according to the Berlin definition: a multicenter prospective 437-446. PubMed PMID: WOS:000182095500017.
study. Crit Care Med. 2015;43(5):947-953. PubMed PMID: 23. Willson DF, Thomas NJ, Markovitz BP, et al. Effect of exogenous
25654173. surfactant (calfactant) in pediatric acute lung injury: a randomized
9. Quasney MW, Lopez-Fernandez YM, Santschi M, Watson RS. controlled trial. JAMA. 2005;293(4):470-476.
The outcomes of children with pediatric acute respiratory distress 24. Curley MAQ, Hibberd PL, Fineman LD, et al. Effect of prone
syndrome: proceedings from the Pediatric Acute Lung Injury positioning on clinical outcomes in children with acute lung
Consensus Conference. Pediatr Crit Care Med. 2015;16(5 suppl injury: a randomized controlled trial. JAMA. 2005;294(2):
1):S118-S131. PubMed PMID: 26035362. 229-237. PubMed PMID: WOS:000230406100027.
10. Heyland DK, Groll D, Caeser M. Survivors of acute respiratory 25. Thomas NJ, Guardia CG, Moya FR, et al. A pilot, randomized,
distress syndrome: relationship between pulmonary dysfunction controlled clinical trial of lucinactant, a peptide-containing syn-
and long-term health-related quality of life. Crit Care Med. 2005; thetic surfactant, in infants with acute hypoxemic respiratory fail-
33(7):1549-1556. PubMed PMID: 16003061. ure. Pediatr Crit Care Med. 2012;13(6):646-653.
11. Herridge MS, Cheung AM, Tansey CM, et al. One-year outcomes 26. Willson DF, Thomas NJ, Tamburro R, et al. Pediatric calfactant in
in survivors of the acute respiratory distress syndrome. N Engl J acute respiratory distress syndrome trial. Pediatr Crit Care Med.
Med. 2003;348(8):683-693. PubMed PMID: 12594312. 2013;14(7):657-665.
12. Hu X, Qian S, Xu F, et al. Incidence, management and mortality 27. Bronicki RA, Fortenberry J, Schreiber M, Checchia PA, Anas
of acute hypoxemic respiratory failure and acute respiratory dis- NG. Multicenter randomized controlled trial of inhaled nitric
tress syndrome from a prospective study of Chinese paediatric oxide for pediatric acute respiratory distress syndrome. J Pediatr.
intensive care network. Acta Paediatr. 2010;99(5):715-721. 2015;166(2):365-369.e1.
13. Flori HR, Glidden DV, Rutherford GW, Matthay MA. Pediatric 28. Cui Q, Zhou H, Zhao R, et al. The effects of open lung ventilation
acute lung injury: prospective evaluation of risk factors associated on respiratory mechanics and haemodynamics in atelectatic
with mortality. Am J Respir Crit Care Med. 2005;171(9): infants after cardiopulmonary bypass. J Int Med Res. 2009;
995-1001. PubMed PMID: WOS:000228816000011. 37(1):113-120.
14. Yu WL, Lu ZJ, Wang Y, et al; Collaborative Study Group of 29. Jacobs BR, Nadkarni V, Goldstein B, et al. Nutritional immuno-
Pediatric Respiratory Failure. The epidemiology of acute respira- modulation in critically ill children with acute lung injury: feasi-
tory distress syndrome in pediatric intensive care units in China. bility and impact on circulating biomarkers. Pediatr Crit Care
Intensive Care Med. 2009;35(1):136-143. PubMed PMID: WOS: Med. 2013;14(1):e45-e56.
000262549100018. 30. Willson DF, Thomas NJ, Tamburro R, et al. The relationship
15. Zambon M, Vincent JL. Mortality rates for patients with acute of fluid administration to outcome in the pediatric calfactant in
lung injury/ARDS have decreased over time. Chest. 2008;133(5): acute respiratory distress syndrome trial. Pediatr Crit Care
1120-1127. PubMed PMID: 18263687. Med. 2013;14(7):666-672. PubMed PMID: WOS:000325705
16. Zimmerman JJ, Akhtar SR, Caldwell E, Rubenfeld GD. Incidence 100008.
and outcomes of pediatric acute lung injury. Pediatrics. 2009; 31. Casado-Flores J, Martinez de Azagra A, Ruiz-Lopez MJ, Ruiz M,
124(1):87-95. Serrano A. Pediatric ARDS: effect of supine-prone postural
17. Schouten LR, Veltkamp F, Bos AP, et al. Incidence and mortality of changes on oxygenation. Intensive Care Med. 2002;28(12):
acute respiratory distress syndrome in children: a systematic review 1792-1796. PubMed PMID: 12447525.
and meta-analysis. Crit Care Med. 2016;44(4):819-829. 32. Cruces P, Dı́az F, Puga A, et al. Angiotensin-converting enzyme
18. Phua J, Badia JR, Adhikari NK, et al. Has mortality from acute insertion/deletion polymorphism is associated with severe hypox-
respiratory distress syndrome decreased over time? A systematic emia in pediatric ARDS. Intensive Care Med. 2012;38(1):
review. Am J Respir Crit Care Med. 2009;179(3):220-227. 113-119.
19. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group 33. Cruces P, Donoso A, Valenzuela J, Diaz F. Respiratory and hemo-
Preferred reporting items for systematic reviews and meta- dynamic effects of a stepwise lung recruitment maneuver in
analyses: the PRISMA statement. BMJ. 2009;339:b2535. pediatric ARDS: a feasibility study. Pediatr Pulmonol. 2013;
20. Higgins JPT, Altman DG, Gøtzsche PC, et al. The Cochrane 48(11):1135-1143. PubMed PMID: 23255291.
Collaboration’s tool for assessing risk of bias in randomised trials. 34. Davis SL, Furman DP, Costarino AT Jr. Adult respiratory distress
BMJ. 2011;343:d5928. syndrome in children: associated disease, clinical course, and
21. Viswanathan M, Berkman ND, Dryden DM, Hartling L. Asses- predictors of death. J Pediatr. 1993;123(1):35-45. PubMed
sing Risk of Bias and Confounding in Observational Studies of PMID: 8320623.
Interventions or Exposures: Further Development of the RTI Item 35. Kitchin OP, Masekela R, Becker P, Moodley T, Risenga SM,
Bank. Methods Research Report. (Prepared by RTI–UNC Green RJ. Outcome of human immunodeficiency virus-exposed
Wong et al 9

and -infected children admitted to a pediatric intensive care unit 46. Lopez-Herce J, de Lucas N, Carrillo A, Bustinza A, Moral R.
for respiratory failure. Pediatr Crit Care Med. 2012;13(5): Surfactant treatment for acute respiratory distress syndrome. Arch
516-519. Dis Child. 1999;80(3):248-252. PubMed PMID: 10325705.
36. Kong MYF, Gaggar A, Li Y, Winkler M, Blalock JE, Clancy JP. Pubmed Central PMCID: PMC1717873.
Matrix metalloproteinase activity in pediatric acute lung injury. 47. Farias JA, Fernandez A, Monteverde E, et al. Critically ill infants
Int J Med Sci. 2009;6(1):9-17. and children with influenza A (H1N1) in pediatric intensive care
37. Kong MYF, Li Y, Oster R, Gaggar A, Clancy JP. Early elevation units in Argentina. Intensive Care Med. 2010;36(6):1015-1022.
of matrix metalloproteinase-8 and-9 in pediatric ARDS is associ- PubMed PMID: 20237757.
ated with an increased risk of prolonged mechanical ventilation. 48. Farias JA, Fernandez A, Monteverde E, et al. Mechanical ventila-
Plos One. 2011;6(8):e22596. PubMed PMID: WOS:0002935 tion in pediatric intensive care units during the season for acute
58900019. lower respiratory infection: a multicenter study. Pediatr Crit Care
38. Li Y, Wang Q, Chen H, Gao HM, Zhou T, Qian SY. Epidemio- Med. 2012;13(2):158-164. PubMed PMID: 21725275.
logical features and risk factor analysis of children with acute lung 49. Moher D, Hopewell S, Schulz KF, et al. CONSORT 2010 expla-
injury. World J Pediatr. 2012;8(1):43-46. nation and elaboration: updated guidelines for reporting parallel
39. López-Fernández Y, Azagra AMD, De La Oliva P, et al. Pediatric group randomised trials. Int J Surg. 2012;10(1):28-55. PubMed
acute lung injury epidemiology and natural history study: inci- PMID: 22036893.
dence and outcome of the acute respiratory distress syndrome in 50. Van Spall HG, Toren A, Kiss A, Fowler RA. Eligibility criteria of
children. Crit Care Med. 2012;40(12):3238-3245. randomized controlled trials published in high-impact general
40. Piastra M, De Luca D, Pietrini D, et al. Noninvasive pressure- medical journals: a systematic sampling review. JAMA. 2007;
support ventilation in immunocompromised children with 297(11):1233-1240. PubMed PMID: 17374817.
ARDS: a feasibility study. Intensive Care Med. 2009;35(8): 51. Hess DR. Retrospective studies and chart reviews. Respir Care.
1420-1427. 2004;49(10):1171-1174. PubMed PMID: 15447798.
41. Smith DW, Frankel LR, Derish MT, et al. High-frequency jet 52. Kneyber MC, Brouwers AG, Caris JA, Chedamni S, Plotz FB.
ventilation in children with the adult respiratory distress syn- Acute respiratory distress syndrome: is it underrecognized in the
drome complicated by pulmonary barotrauma. Pediatr Pulmonol. pediatric intensive care unit? Intensive Care Med. 2008;34(4):
1993;15(5):279-286. PubMed PMID: 8327286. 751-754. PubMed PMID: 18288473.
42. Yapicioğlu H, Yildizdaş D, Bayram I, Sertdemir Y, Yilmaz HL. 53. Khemani RG, Thomas NJ, Venkatachalam V, et al. Comparison
The use of surfactant in children with acute respiratory distress of SpO2 to PaO2 based markers of lung disease severity for
syndrome: efficacy in terms of oxygenation, ventilation and mor- children with acute lung injury. Crit Care Med. 2012;40(4):
tality. Pulm Pharmacol Ther. 2003;16(6):327-333. 1309-1316. PubMed PMID: 22202709.
43. Yehya N, Servaes S, Thomas NJ. Characterizing degree of lung 54. Thomas NJ, Shaffer ML, Willson DF, Shih MC, Curley MA. Defin-
injury in pediatric acute respiratory distress syndrome. Crit Care ing acute lung disease in children with the oxygenation saturation
Med. 2015;43(5):937-946. index. Pediatr Crit Care Med. 2010;11(1):12-17. PubMed PMID:
44. Samransamruajkit R, Tongnuan C, Sosothikul D, Deelodejana- 19561556. Pubmed Central PMCID: PMC2936504.
wong J, Sritippayawan S, Prapphal N. Pulmonary coagulopathy 55. Ventilation with lower tidal volumes as compared with traditional
in pediatric acute lung injury/acute respiratory distress syndrome. tidal volumes for acute lung injury and the Acute Respiratory
Crit Care Shock. 2013;16(3):68-73. 6p. Distress Syndrome. The Acute Respiratory Distress Syndrome
45. De Luca D, Lopez-Rodriguez E, Minucci A, et al. Clinical and Network. N Engl J Med. 2000;342(18):1301-1308. PubMed
biological role of secretory phospholipase A2 in acute respiratory PMID: 10793162.
distress syndrome infants. Crit Care. 2013;17(4):R163. PubMed 56. Bersten AD, Cooper DJ. Better supportive care, less ARDS. just
PMID: 23883784. Pubmed Central PMCID: PMC4057254. do it? Am J Respir Crit Care Med. 2011;183(1):6-7.