To evaluate the association between hypoxaemia and mortality from acute lower respiratory infections (ALRI) in children in low- and middle-income countries (LMIC).
Observational studies reporting on the association between hypoxaemia and death from ALRI in children below five years in LMIC.
Risk of Bias Assessment
Quality In Prognosis Studies tool with minor adaptations to assess the risk of bias; funnel plots and Egger’s test to evaluate publication bias.
Out of 11,627 papers retrieved, 18 studies from 13 countries on 20,224 children met the inclusion criteria. Twelve (66.6%) studies had either low or moderate risk of bias. Hypoxaemia defined as oxygen saturation rate (SpO2) <90% associated with significantly increased odds of death from ALRI (OR 5.47, 95% CI 3.93 to 7.63) in 12 studies on 13,936 children. An Sp02 <92% associated with a similar increased risk of mortality (OR 3.66, 95% CI 1.42 to 9.47) in 3 studies on 673 children. Sensitivity analyses (excluding studies with high risk of bias and using adjusted OR) and subgroup analyses (by: altitude, definition of ALRI, country income, HIV prevalence) did not affect results. Only one study was performed on children living at high altitude.
The results of this review support the routine evaluation of SpO2 for identifying children with ALRI at increased risk of death. Both a Sp02 value of 92% and 90% equally identify children at increased risk of mortality. More research is needed on children living at high altitude. Policy makers in LMIC should aim at improving the regular use of pulse oximetry and the availability of oxygen in order to decrease mortality from ALRI.
Citation: Lazzerini M, Sonego M, Pellegrin MC (2015) Hypoxaemia as a Mortality Risk Factor in Acute Lower Respiratory Infections in Children in Low and Middle-Income Countries: Systematic Review and Meta-Analysis. PLoS ONE 10(9): e0136166. https://doi.org/10.1371/journal.pone.0136166
Editor: Zulfiqar A. Bhutta, The Hospital for Sick Children, CANADA
Received: February 27, 2015; Accepted: July 31, 2015; Published: September 15, 2015
Copyright: © 2015 Lazzerini et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Acute lower respiratory infections (ALRI), such as pneumonia and bronchiolitis, are the leading cause of morbidity and mortality in children under five years of age. According to recent estimates, every year about 120–156 million cases of ALRI occur globally with approximately 1.4 million resulting in death. More than 95% of these deaths occur in low and middle income countries (LMIC)[1–3].
Currently there is a significant debate on how to improve the identification of cases of “severe ALRI“, and on which are the best prognostic factors that can be used in routine care to identify children with a higher risk of death [4–6].
According to a systematic review, hypoxaemia as detected with pulse oximetry has been observed in 13% of children with WHO-defined pneumonia requiring hospitalisation (severe and very severe classifications). This corresponds to at least 1.5 to 2.7 million annual cases of hypoxaemic pneumonia presenting to health-care facilities .
Although several research papers have suggested that hypoxia may be associated with increased odds of death in children with ALRI, no recent systematic review has synthetised the evidence in this regard. A previous systematic review dates back to year 2001, while a significant amount of literature has been published after that date . The aim of this work was to systematically review and meta-analyse the evidence on the association between hypoxia, defined using different cut-offs, and ALRI mortality in children in LMIC.
A better understanding of this association may inform both policy makers, health staff directly involved with clinical management of ALRI in children, and researchers.
Search strategy and eligibility criteria
In conducting this review we followed the guidelines reported in the PRISMA (Preferred Reporting Items for systematic reviews and meta-analyses) . and the MOOSE (Meta-analysis of Observational Studies) . A protocol including detailed methods of the review was developed before starting the review.
We searched up to January 2015 the following databases: MEDLINE through Pubmed (from 1956); Embase through OVID (from 1974); Global Health Library (WHO website, no date restrictions), LILACS through the Virtual Health Library (no date restrictions); Science Citation Index Expanded (SCI-EXPANDED) through Web of Science (from 1992); Social Sciences Citation Index (SSCI) through Web of Science (from 1992). The search strategy is reported in Table 1. Manual searches of reference lists were also performed. We did not apply any language restrictions.
Observational studies were eligible for inclusion if they reported the association between death from ALRI and hypoxaemia, in children under 5 years of age in LMIC, as defined by the World Bank . Both studies at hospital level and in the community were included. Studies reporting selectively on children with very specific co-morbidities, such as studies on children with cancer, organ transplant, burns, ventilator-acquired pneumonia, SARS or avian flu were excluded. Studies with less than five events (deaths) were also excluded.
Studies were selected for inclusion by two independent authors (ML and MS). Any disagreement was resolved through discussion. The full text of all eligible citations was examined in detail. Seven authors were contacted for additional information and four provided additional data, and/or clarifications on the published data.
Two authors (MS and MCP) extracted data from included studies, using a pre-piloted data-extraction form. Disagreements were resolved by discussion between the two authors and consensus with a third author (ML). We extracted information regarding: country, setting and altitude where the study was performed (altitude, if not specified in the study was attributed based on the altitude of the location were the study was performed); period of the study; definition of hypoxaemia; definition for ALRI; characteristics of the population; study design; sample size; type of analysis performed (univariate or multivariate); confounders.
We extracted both unadjusted and adjusted OR or risk ratios (RR), and crude numbers. When possible, we converted RR to OR, using a formula for computing OR from RR . To avoid mistakes due to data manipulation, data were first extracted as in the original paper, and then converted as needed.
In studies also including children over five years of age, only data on children under five were extracted. If sorting was not possible, we only included the study if at least 80% of the children were under five years old.
Single studies that reported on cut-offs of oxygen saturation rate (SpO2) that could not be grouped with other studies were grouped as more appropriate to improve the informativeness of results (Table 2, notes).
Assessment of risk of bias
Two review authors (MS and MCP) independently assessed the risk of bias using the Quality In Prognosis Studies (QUIPS)  tool 15 with minor adaptations (Table A in S1 File). The tool includes 32 key items divided into 6 domains: 1) Study participation; 2) Study attrition; 3) Prognostic factor measurement; 4) Outcome measurement; 5) Study Confounding; 6) Statistical Analysis and reporting. For each study, each individual key item was assessed, and each domain was graded in one of the following categories of risk of bias, based on whether the domain fully complied, partly complied, did not comply at all, or did not report in respect of the characteristic expressed by the items: 1) Low risk of bias; 2) Moderate risk of bias; 3) High risk of bias; 4) Unknown. The overall risk of bias for each single study was rated as follows (Table B in S1 File): 1) Low—if all the six domains were scored as low, or if not more than two moderate or unknown risks of bias were identified with no high risk of bias; 2) Moderate—if moderate or unknown risk of bias was identified in three domains without any high risk of bias, or if a high risk of bias was identified in association with not more than one moderate or unknown risk of bias; 3) High—if a high risk of bias was identified in two or more domains independently from the grading in the other domains, or if moderate or unknown risk of bias was identified in four or more domains independently from grading in the other domains.
When meta-analysis was possible and appropriate we generated a pooled OR using the inverse-variance weighting method. As we expected high heterogeneity among studies (in the population, in the definition of ALRI, in the definition of hypoxaemia, and the overall methodology), we selected a priori the DerSimonian and Laird random effect model , which accounts for intra- and inter- study variability.
Pooled data were presented in a forest plot, where studies are ordered chronologically; data from one study that could not be meta-analyzed were described in text.
We tested the null hypothesis that all studies evaluate the same true effect by the Cochran’s Q test, with p<0.05 considered statistically significant. The degree of heterogeneity between studies was assessed by visual inspection of the forest plots and I-squared (I2) statistic with its 95% confidence intervals. Heterogeneity was considered low for I2 values between 25%-50%, moderate for 50%-75%, and high for ≥75% .
Sensitivity analyses and exploration of heterogeneity
We performed sensitivity analyses to examine the effect of removing the studies with high risk of bias and of using adjusted OR rather than the crude OR, when the former was available.
Using subgroup and metaregression analysis, we planned to explore the effect of the following study-level factors: a) Altitude (studies in populations living at 0–1,499 meters above sea level vs 1,500–2,499 meters vs > 2,500 meters); b) different definitions of ALRI used (WHO definition versus other definitions); c) income (low income countries versus lower middle income countries, as defined by the World Bank13 at the moment of the study start—no study was available for the category of upper middle income countries); d) HIV country prevalence (high HIV prevalence versus non-high HIV prevalence, where a high HIV prevalence was defined as a prevalence of HIV in the population of 15–49 year-olds at the time of the study > 5%.
We assessed potential publication bias and small study effects with funnel plots and the Egger test .
All statistical analysis was performed using Stata version 12 .
Characteristics of the studies
The systematic search yielded overall 11,627 records (Fig 1). Overall 18 studies [20–37] from 13 countries on 20,224 children met the inclusion criteria. Characteristics of the studies are summarized in Table 2.
All except three studies were published during the last 15 years. Twelve studies were performed in Africa (two studies in Gambia, Kenya, Uganda and South Africa; one each in Central African Republic, Malawi, Mozambique, Zambia), three in South East Asia (two in Bangladesh, one in Indonesia), two in Western Pacific Region (one in Papua New Guinea, one in China) and one in South America (Colombia). All studies were hospital-based, 12 were conducted in an urban setting setting, five in a rural setting, and one in both. Three studies were multi-centered. Fourteen studies were performed at an altitude of between 0–1,499 meters above sea level, three studies at an altitude between 1,500–2,499 meters, and one study at an altitude of > 2,500 meters above sea level (this study used as a cut-off of Sp02 a value of 90%).
Twelve studies contrubuted on the evaluation of hypoxaemia defined as a cut-off oxygen of SpO2 of 90%, four studies on a cut-off of 92%; one study on a cut-off of 85%, and one study on cut-offs of 70% versus 70–84%.
Overall, 12 studies used the WHO definition of pneumonia, while the other used other clinical criteria (five studies), radiological classification (one study) and ICD10 classification (one study). Eleven studies also included newborns. Eleven studies enrolled children with either severe or very severe pneumonia. Two studies in Bangladesh enrolled children who all presented either diarrhea (one study) or malnutrition (the other study) as a co-morbidity. In two studies in Malawi and South Africa respectively over half of enrolled children were HIV infected. Mortality rate in the included studies ranged from 3.4% to 15.3%.
One study  reported data separatly on HIV-positive and HIV-negative children, and was included in the meta-analysis as two sub-studies.
Risk of bias for included studies is reported in Fig 2. Six studies were considered at high risk of bias, with lack of adjustment for confounding and other aspects of the statistical analysis being the most frequent risk of bias. The other 12 studies were considered either at low (eight studies) or at moderate (four studies) risk of bias.
Results of the meta-analysis
Results of the meta-analysis are reported in Fig 3. Seventeen studies were included in the meta-analysis.
Notes: seventeen studies were able to be pooled in the meta-analysis. The remaining study that could not be meta-analysed used an SpO2 cut-off of 92% and found an HR of 12.2 (95% CI 1.6–92.0).29.
Hypoxaemia defined with a cut-off for oxygen saturation rate (SpO2) below 90% was associated with significant increased odds of death from ALRI (OR 5.47, 95% CI 3.93 to 7.63) in 13 studies on 13,928 children.
An Sp02 below 92% compared to over 92% was associated with an increased risk of mortality of 3.66 (95% CI 1.42 to 9.47) in 3 studies on 673 children. Using the same cut-off of 92%, one study on 614 children found an HR of 12.2 (1.6–92.0) .
One study on 4,306 Indonesian children using a cut-off of SpO2 of 85% identified an OR of 5.61(95% CI 4.48 to 7.03).
A study in Papua New Guinea on 6,703 children comparing an SpO2 of < 70% to values 70–84% resulted in an OR of 2.46 (95% CI 1.30 to 4.65).
Heterogeneity of results was moderate in both the group of studies using a SpO2 cut-off value of 92% (3 studies, I2 = 61.4%), and in the group of studies using a cut-off value of 90% (13 studies, 65.3%).
Results of sensitivity analyses are reported in Table 3. When only studies with either low or moderate risk of bias were included in the analysis (11 studies), hypoxaemia was overall still significantly associated with increased odds of mortalty (OR 5.71, 95% CI 4.34 to 7.50).
Both when all studies were retained in the meta-analysis, but adjusted OR were used for studies providing them, and when only studies prividing an adjusted ORs were analysed (four studies), hypoxaemia was still significantly associated with increased odds of mortalty (OR = 4.66, 95% CI 3.50 to 6.19; OR = 4.69, 95% CI 2.55 to 8.61).
Subgroup analyses and metaregression
Results of the subgroup analysis and metaregression are reported in Table 4. Overall, subgroup analysis did not affect results. When studies were sub-grouped based on different altitudes where the studies were performed, the overall association between hypoxaemia and mortality persisted both in the subgroup studies in populations living at 0–1,499 meters above sea level (OR 5.23, 95%CI 3.99 to 6.86, 13 studies) and at 1,500–2,499 meters above sea level (OR 5.34, 95%CI 2.74 to 10.60, 3 studies, 4 comparisons). Fig 4 and Fig 5 report the subgroup analysis in detail: overall in both subgroups all explored SpO2 cut-offs were significantly associated with increased odds of mortality, without significant differences between different cut-offs, although in the subgroup of studies at 0–1,499 meters (Fig 5) there was a trend for higher risk of mortality with decreasing Sp02 cut-offs (92%: OR 3.66, 95% CI 1.42 to 9.47; 90%: OR 5.80, 95% CI 3.98 to 8.46). One only small study (255 children) reporting on a population in Colombia living at an altitude > 2,500 meters, and evaluating an Sp02 of 90%, did not identify a significant association with increased mortalty risk (OR 1.51, 95%CI 0.75 to 3.04).
When studies where subgrouped based on different diagnostic criteria for ALRI (WHO criteria vs other criteria), hypoxaemia was significanty associated with mortality in both subgroups (OR 5.06, 95% CI 3.79 to 6.76; OR 4.84, 95% CI 2.82 to 8.30, p = 0.811).
When studies where subgrouped based on incomes in the country (low vs lower-middle medium) hypoxaemia was significanty associated with mortality in both subgroups (OR 4.92, 95% CI 3.78 to 6.40; OR 5.19, 95% CI 2.64 to 10.21, p = 0.982)
When studies where subgrouped based on HIV prevalence in the country, hypoxaemia was significanty associated with mortality in both subgroups (OR 4.43, 95% CI 2.57 to 7.64; OR 4.33, 95% CI 3.02 to 6.02, p = 0.971).
Interpretation of the results
This systematic review synthesises the available evidence on the association between hypoxaemia and ALRI mortality in children in LMIC. Results from this review confirm the importance of measuring oxygen saturation in children with ALRI in order to identify children with higher risk of mortality. Overall, the association with hypoxaemia is well documented: the pooled analysis of 12 studies showed that children with SpO2 below 90% have a 5.4 fold increase in the risk of death, while 3 studies showed that children with SpO2 below 92% have a 3.6 fold increase in odds of death. Sensitivity analyses (including only studies with either low or moderate risk of bias and using adjusted OR) and subgroup analyses (by altitude, definition of ALRI, country income, HIV prevalence) did not change the results. Tests for publication bias did not suggest their existence.
A previous review published several years ago  identified only three studies reporting on hypoxaemia and mortality in children with ALRI. This review updates the current evidence-synthesis on this subject.
Strengths and limitations
In performing this review we used a comprehensive search strategy without language restrictions; quality of retrieved studies was overall fair, without suggestion of publications bias. We were aware of the possibility of introducing bias at every stage of the review process and tried to minimise this aspect by following strict methods and standards suggested for systematic reviews [11,12,15].
Moderate heterogeneity of results among studies may be explained by heterogeneity in the characteristics of the studies, such as design, setting, type of population included, definition of ALRI, and power of the study (which may explain large confidence intervals in some instances). Subgroup analyses reduced part of the heterogeneity in some sub-groups, however heterogeneity may be also explained by other confounding factors affecting mortality rates, such as the different quality of care provided in the studies, and different algorithms used to treat hypoxaemia.
This review identifies a lack of data on the prognostic value of Sp02 in populations living at very high altitudes (> 2,500 meters above sea level).
Other limitations in this review are related to limitations in the designs of the included studies and in the availability of individual patient data: few studies in the analysis provided had adjusted OR after correction for other related risk factors, and no individual patient database was availble to generate pooled-adjusted OR. However, sensitivity analysis using adjusted ORs confirmed the primary results.
Implication for policies and research
The results of this review support the routine evaluation of oxygen saturation rate for identifying children with ALRI at higher risk of death. This is in line with current WHO guidelines, which recommend determining the presence of hypoxaemia with pulse-oximetry in all children with ALRI [38–39]. This indication recognises the inaccuracy and unreliability of clinical signs to detect cases who need oxygen, together with the role of oxygen as an essential treatment for pneumonia . Two systematic reviews highlighted that neither single nor combined symptoms and signs are sufficiently effective for predicting hypoxaemia among young children with ALRI [39–41].
Currently, there are different types of indications regarding at which Sp02 value to start oxygen therapy [38–47]. WHO recommends starting oxygen therapy when the saturation rate is below 90%, when at an altitude < 2,500 meters above sea level [38–39]. Other guidelines on acute respiratory conditions indicate as a threshold to start oxygen therapy either a value of SpO2 < 90% [42–44] or < 92% [45–46], or between 90–92% . Direct evidence in support of any specific Sp02 threshold for starting supplementation with oxygen is lacking: a systematic review could not identify any study comparing outcomes of children receiving oxygen at different cut-offs of hypoxaemia . Results of our review indicate that both the category of children with an SpO2 < 90% (OR 5.47, 95% CI 3.93 to 7.63) and children with SpO2 < 92% (3.66, 95% CI 1.42 to 9.47) have an increased risk of mortality, without significant difference in the ORs between the two Sp02 thresholds, and with similar sesults both at low altituides and at moderate altitudes (<2,500 meters above sea level). This suggests that even children with SpO2 below 92% may benefit from oxygen therapy.
For children living at high altitude (> 2,500 meters above sea level), WHO recommends a lower cut-off for given oxygen, such as SpO2 < 87% [38–39]. This review identifies only a small study (255 children) reporting on a population in Colombia living at an altitude > 2,500 meters, and evaluating an Sp02 of 90%, without any significant association with increased mortalty risk (OR 1.51, 95%CI 0.75 to 3.04). This may suggest that the use of Sp02 thresholds lower than 90% is appropriate, although there is no evidence in support of any precise cut-off. Further studies are therefore needed to explore the prognistic value of different cut-offs of oxygen saturation for populations living at very high altitudes (> 2,500 meters above sea level).
Effective and cheap systems for delivering oxygen have been developed and tested in low-resource countries [49–51].Policy makers should aim at improving the availability of such oxygen delivery systems together with the availability of pulse oximetry in LMIC. Alongside the provision of equipment, training and monitoring to promote effective use of available resouces is also needed. New relatively simple technologies, utilising for example smartphone devices such as pulse oxymeter, may facilitate the implementation of such practices .
This review adds some information to the current knowledge of prognostic factors for ALRI mortality in children in LMIC. A recent systematic review  highlighted that a broad range of factors are associated with increased odds of death for ALRI in children, including both child-factors (age, female sex, prematurity, low birth weight, malnutrition, inadequate breastfeeding, infections with HIV/AIDS and other co-morbidities, chronic diseases), parental factors (socio-economical status, maternal education), envoromental factors (indoor air pollution, second hand smoke exposure). The current knowledge should now be integrated in order to develop and field test comprehensive prognostic indexes–such as scoring systems including both hypoxaemia and other child, parental and environmental risk factors- in order to identify children at higher risk of ALRI mortality who should be targeted by more intensive interventions and follow up care.
The results of this review support the routine evaluation of oxygen saturation rate for identifying children with ALRI at higher risk of death. Despite the lack of direct evidence in support of any specific Sp02 threshold for starting supplementation with oxygen, this review shows that both an Sp02 value of 92% and 90% equally identify children at increased risk of mortality. Further studies should focus on children living at high altitudes. Policy makers should aim at improving the availability of pulse oximetry and oxygen in LMIC.
We thank Victoria Lutje, of the Cochrane Infectious Disease Group, for having searched for relevant papers in Embase.
The PRISMA Checklist is reported as Table C in in S1 File.
Conceived and designed the experiments: ML. Performed the experiments: ML MS MCP. Analyzed the data: MS. Contributed reagents/materials/analysis tools: ML MS MCP. Wrote the paper: ML MS MCP.
- 1. Nair H, Simoes EA, Rudan I, Gessner BD, Azziz-Baumgartner E, Zhang JS et al. (2013) Global and regional burden of hospital admissions for severe acute lower respiratory infections in young children in 2010: a systematic analysis. Lancet 381: 1380–90. pmid:23369797
- 2. Walker CL, Rudan I, Liu L, Nair H, Theodoratou E, Bhutta ZA et al. (2013) Global burden of childhood pneumonia and diarrhoea. Lancet 381: 1405–16. pmid:23582727
- 3. Liu L, Johnson HL, Cousens S, Perin J, Scott S, Lawn JE et al. (2012) Global, regional, and national causes of child mortality: An updated systematic analysis for 2010 with time trends since 2000. Lancet 379: 2151–61. pmid:22579125
- 4. Lozano R, Naghavi M, Foreman K, Lim S, Shibuya K, Aboyans V et al. (2012) Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380: 2095–128. pmid:23245604
- 5. Jackson S, Mathews KH, Pulanic D, Falconer R, Rudan I, Campbell H et al. (2013) Risk factors for severe acute lower respiratory infections in children: a systematic review and meta-analysis. Croat Med J 54: 110–21. pmid:23630139
- 6. Wonodi CB, Deloria-Knoll M, Feikin DR, DeLuca AN, Driscoll AJ, Moisi JC et al. (2012) Evaluation of risk factors for severe pneumonia in children: The pneumonia etiology research for child health study. Clinical Infectious Diseases 54 Suppl 2: S124–S131). pmid:22403226
- 7. Duke T, Subhi R, Peel D, Frey B. Pulse oximetry: technology to reduce child mortality in developing countries. Ann Trop Paediatr. 2009;29:165–75. pmid:19689857
- 8. Jensen LA, Onyskiw JE, Prasad NG. Meta-analysis of arterial oxygen saturation monitoring by pulse oximetry in adults. Heart Lung. 1998;27:387–408. pmid:9835670
- 9. Subhi R, Adamson M, Campbell H, Weber M, Smith K, Duke T; Hypoxaemia in Developing Countries Study Group. The prevalence of hypoxaemia among ill children in developing countries: a systematic review. Lancet Infect Dis. 2009;9:219–27. pmid:19324294
- 10. Lozano JM. (2001) Epidemiology of hypoxaemia in children with acute lower respiratory infection. Int J Tuberc Lung Dis 5: 496–504. pmid:11409574
- 11. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP et al. (2009) The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med 6: e1000100. pmid:19621070
- 12. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D et al. (2000) Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 283: 2008–12. pmid:10789670
- 13. The World Bank, Country and Lending Groups. (2014) Historical classification. Available: http://dataworldbank.org/about/country-classifications/a-short-history (Accessed 21 January 2015).
- 14. Chinn S. (2002) Comparing and combining studies of bronchial responsiveness. Thorax 57: 393–5. pmid:11978913
- 15. Hayden JA, Cote P, Bombardier C. (2006) Evaluation of the quality of prognosis studies in systematic reviews. Ann Intern Med 144: 427–37. pmid:16549855
- 16. DerSimonian R, Laird N. Meta-analysis in clinical trials. (1986) Control Clin Trials 7: 177–88. pmid:3802833
- 17. Higgins JP, Thompson SG, Deeks JJ, Altman DG. (2003) Measuring inconsistency in meta-analyses. BMJ 327: 557–60. pmid:12958120
- 18. Egger M, Davey SG, Schneider M, Minder C. (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315: 629–34. pmid:9310563
- 19. StataCorp. (2011) Stata Statistical Software: Release 12. College Station, Texas 77845 USA: StataCorp LP.
- 20. Chisti MJ, Duke T, Robertson CF, Ahmed T, Faruque AS, Bardhan PK et al. Co-morbidity: exploring the clinical overlap between pneumonia and diarrhoea in a hospital in Dhaka, Bangladesh. Ann Trop Paediatr 2011; 31: 311–9. pmid:22041465
- 21. Chisti MJ, Salam MA, Ashraf H, Faruque AS, Bardhan PK, Hossain MI et al. Clinical risk factors of death from pneumonia in children with severe acute malnutrition in an urban critical care ward of Bangladesh. Plos One 2013; 8: e73728. pmid:24040043
- 22. Demers AM, Morency P, Mberyo-Yaah F, Jaffar S, Blais C, Somsé P et al. Risk factors for mortality among children hospitalized because of acute respiratory infections in Bangui, Central African Republic. Pediatr Infect Dis J 2000; 19: 424–32. pmid:10819338
- 23. Djelantik IG, Gessner BD, Sutanto A, Steinhoff M, Linehan M, Moulton LH et al. Case fatality proportions and predictive factors for mortality among children hospitalized with severe pneumonia in a rural developing country setting. J Trop Pediatr 2003; 49: 327–32. pmid:14725409
- 24. Duke T, Mgone J, Frank D. Hypoxaemia in children with severe pneumonia in Papua New Guinea. Int J Tuberc Lung Dis 2001; 5: 511–9. pmid:11409576
- 25. Graham SM, Mankhambo L, Phiri A, Kaunda S, Chikaonda T, Mukaka M et al. Impact of human immunodeficiency virus infection on the etiology and outcome of severe pneumonia in Malawian children. Pediatr Infect Dis J 2011; 30: 33–8. pmid:21173674
- 26. Junge S, Palmer A, Greenwood BM, Kim Mulholland E, Weber MW. The spectrum of hypoxaemia in children admitted to hospital in The Gambia, West Africa. Trop Med Int Health. 2006 Mar;11(3):367–72. pmid:16553917
- 27. Mwaniki MK, Nokes DJ, Ignas J, Munywoki P, Ngama M, Newton CR, et al. Emergency triage assessment for hypoxaemia in neonates and young children in a Kenyan hospital: an observational study. Bull World Health Organ.2009;87:263–70. pmid:19551234
- 28. McNally LM, Jeena PM, Gajee K, Thula SA, Sturm AW, Cassol S et al. Effect of age, polymicrobial disease, and maternal HIV status on treatment response and cause of severe pneumonia in South African children: a prospective descriptive study. Lancet 2007; 369: 1440–51. pmid:17467514
- 29. Nantanda R, Hildenwall H, Peterson S, Kaddu-Mulindwa D, Kalyesubula I, Tumwine JK. Bacterial aetiology and outcome in children with severe pneumonia in Uganda. Ann Trop Paediatr 2008; 28: 253–60. pmid:19021940
- 30. Nantanda R, Ostergaard MS, Ndeezi G, Tumwine JK. Clinical outcome of children with acute asthma and pneumonia in Mulago hospital, Uganda: a prospective study. BMC Pediatris 2014; 14: 285.
- 31. Onyango FE, Steinhoff MC, Wafula EM, Wariua S, Musia J, Kitonyi J. Hypoxaemia in young Kenyan children with acute lower respiratory infection. BMJ 1993; 306: 612–5. pmid:8369033
- 32. Reed C, Madhi SA, Klugman KP, Kuwanda L, Ortiz JR, Finelli L, et al. Development of the Respiratory Index of Severity in Children (RISC) score among young children with respiratory infections in South Africa. PLoS One.2012;7(1):e27793. pmid:22238570
- 33. Rodríguez CE, Hernández LJ, Aristizábal R, Guzmán MC, Castillo C, Aristizábal G et al. Factors affecting mortality from acute respiratory infections in a population of children under five years living in the city of Bogota. Investig segur soc salud 2010; 12: 21–38.
- 34. Sigauque B, Roca A, Bassat Q, Morais L, Quinto L, Berenguera A et al. Severe pneumonia in Mozambican young children: clinical and radiological characteristics and risk factors. J Trop Pediatr 2009; 55: 379–87. pmid:19401405
- 35. Smyth A, Carty H, Hart CA. Clinical predictors of hypoxaemia in children with pneumonia. Ann Trop Paediatr 1998; 18: 31–40. pmid:9691999
- 36. Usen S, Weber M, Mulholland K, Jaffar S, Oparaugo A, Omosigho C et al. Clinical predictors of hypoxaemia in Gambian children with acute lower respiratory tract infection: prospective cohort study. BMJ 1999; 318: 86–91. pmid:9880280
- 37. Zhang Q, Zhongqin G, Zhenjiang B, MacDonald NE. A 4 years prospective study to determine severe community acquired pneumonia in children in Southern China Pediatric Pneumology 2013; 48: 390–397.
- 38. World Health Organization (2013) Pocket book of hospital care for children: guidelines for the management of common childhood illnesses. Second edition. Available: http://www.whoint/maternal_child_adolescent/documents/child_hospital_care/en/ 2013 Accessed: 21 January 2015.
- 39. World Health Organization (2013)Technical Recommendations for management of common childhood conditions.Evidence for technical update of pocket book recommendations, Available: http://apps.who.int/iris/bitstream/10665/44774/1/9789241502825_eng.pdf?ua=1&ua=1 Accessed: 21 January 2015.
- 40. Zhang L, Mendoza-Sassi R, Santos JC, Lau J. Accuracy of symptoms and signs in predicting hypoxaemia among young children with acute respiratory infection: a meta-analysis. Int J Tuberc Lung Dis. 2011;15:317–25. pmid:21333097
- 41. Rigau D, Rojas MX, Alonso P. Summary of Findings tables for oxygen therapy in infants, children and adults with lower respiratory tract infections, 2010 (WHO, unpublished).
- 42. Luna Paredes MC, Asensio de la Cruz O, Cortell Aznar I, Martínez Carrasco MC, Barrio Gómez de Agüero MI, et al; Grupo de Técnicas de la Sociedad Española de Neumología Pediátrica. [Oxygen therapy in acute and chronic conditions: Indications, oxygen systems, assessement and follow-up]. An Pediatr (Barc). 2009; 71:161–74.
- 43. Camargo CA Jr, Rachelefsky G, Schatz M. Managing asthma exacerbations in the emergency department: summary of the National Asthma Education and Prevention Program Expert Panel Report 3 guidelines for the management of asthma exacerbations. J Emerg Med. 2009;37(2 Suppl):S6–S17. pmid:19683665
- 44. American Academy of Pediatrics Subcommittee on Diagnosis and Management of Bronchiolitis. Diagnosis and management of bronchiolitis. Pediatrics. 2006;118:1774–93. pmid:17015575
- 45. Harris M, Clark J, Coote N, Fletcher P, Harnden A, McKean M, et al. British Thoracic Society Standards of Care Committee. British Thoracic Society guidelines for the management of community acquired pneumonia in children: update 2011. Thorax. 2011;66 Suppl 2:ii1–23. pmid:21903691
- 46. González de Dios J, Ochoa Sangrador C; Grupo de revisión y panel de expertos de la Conferencia de Consenso del Proyecto aBREVIADo (BRonquiolitis-Estudio de Variabilidad, Idoneidad y ADecuación). [Consensus conference on acute bronchiolitis (I): methodology and recommendations]. An Pediatr (Barc). 2010;72:221.e1-221.e33.
- 47. Zar HJ, Jeena P, Argent A, Gie R, Madhi SA; Working Groups of the Paediatric Assembly of the South African Thoracic Society. Diagnosis and management of community-acquired pneumonia in childhood—South African Thoracic Society Guidelines. S Afr Med J. 2005; 95):977–81, 984–90. pmid:16482985
- 48. Subhi R, In sick children, what are the best cut-off oxygen saturations at which oxygen would be indicated at sea level and high altitude to prevent death or sequelae? (WHO unpublished, 2010).
- 49. Duke T, Graham SM, Cherian MN, Ginsburg AS, English M, Howie S, et al Union Oxygen Systems Working Group. Oxygen is an essential medicine: a call for international action. Int J Tuberc Lung Dis. 2010;14:1362–8. pmid:20937173
- 50. Duke T, Wandi F, Jonathan M, et al. Improved oxygen systems for childhood pneumonia: a multihospital effectiveness study in Papua New Guinea. Lancet. 2008; 372:1328–1333. pmid:18708248
- 51. Catto AG, Zgaga L, Theodoratou E, Huda T, Nair H, El Arifeen S, et al An evaluation of oxygen systems for treatment of childhood pneumonia. BMC Public Health. 2011;11 Suppl 3:S28. pmid:21501446
- 52. Petersen CL, Chen TP, Ansermino JM, Dumont GA. Design and evaluation of alow-cost smartphone pulse oximeter. Sensors (Basel). 2013;13:16882–93.
- 53. Sonego M, Pellegrin MC, Becker G, Lazzerini M.Risk factors for mortality from acute lower respiratory infections (ALRI) in children under five years of age in low and middle-income countries: a systematic review and meta-analysis of observational studies. Plos One 2015 Jan 30;10(1):e0116380. pmid:25635911