Acute lower respiratory infection (ALRI), such as pneumonia and bronchiolitis, constitute a major cause of respiratory failure in children (1, 2), of whom 15% may need intensive care unit (ICU) care and respiratory support (3). The global mortality annually is estimated as 6,50,000 for ALRI in children younger than 5 years (4). Standard oxygen therapy (SOT) at a flow rate of 1–4 L/min for children with ALRI and hypoxemia is recommended by the World Health Organization (WHO) (5, 6). However, despite the provision of SOT and supportive care, more effective respiratory support systems are still needed for some critically ill children because of worsening respiratory distress (7).

To avoid airway invasivity related to the intubation, various non-invasive ventilation (NIV) strategies, such as continuous positive airway pressure (CPAP), bilevel positive airway pressure (BIPAP) ventilation, and high flow nasal cannula (HFNC), have been proposed among pediatric patients (8, 9). However, the effect of NIV in children with ALRI remains inconclusive. A previous pairwise meta-analysis (10) that compared two of the three respiratory support devices (SOT, CPAP, and HFNC) revealed that in terms of reducing treatment failure, HFNC was better than SOT but inferior to CPAP. However, there were no significant differences between the therapies in terms of intubation and mortality rates. However, this conventional meta-analysis only focused on head-to-head trials of comparison of two interventions without assessing multiple interventions as a whole. In addition, since the publication of this meta-analysis, several new randomized controlled trials (RCTs) have been published (11–18). More importantly, there are no systematic reviews that have assessed the efficacy of BIPAP in pediatric patients.

A network meta-analysis (NMA) enables comparison of multiple interventions and improved precision through a combination of direct and indirect estimates of effects (19, 20). Therefore, we aimed to perform a rigorous and comprehensive Bayesian network meta-analysis to update current clinical study data and evaluate the effect of multiple NIV strategies in children with ALRI.

This systematic review was registered on PROSPERO with registration number: CRD42020172156. This report complies with the recommendations of Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) Extension statement for NMA (21).

In this first network meta-analysis comparing different NIV modalities in children with ALRI, we found that CPAP reduced the risk of intubation when compared to standard oxygen therapy. What is more, a lower incidence of treatment failure rate was found for both CPAP and HFNC when compared to standard oxygen therapy. However, there was no evidence of differences for all modalities concerning in-hospital mortality. Overall, the certainty of evidence was low or very low for all the outcomes, not only because of the small number of trials in each node, but also due to the variability of methodology and participants among studies.

A pairwise meta-analysis (10) of different NIV modalities in children with ALRI by Luo and colleagues showed that HFNC reduced treatment failure than SOT, despite no reduction in intubation rate. They also reported that HFNC had a higher risk of treatment failure when compared with CPAP. Our primary and sensitivity analyses are consistent with these previous findings. However, we found that CPAP use was associated with lower risks of treatment failure and intubation compared to SOT. This contrasts with the previous study by Luo et al. (10) where no significant differences between these two interventions were observed. The differences between their results and ours may be explained by the following reasons. First, this NMA had included more recently published studies. Second, this was an NMA where apart from the direct synthesis, the indirect evidence also contributed to the overall effect estimate.

Compared with low-flow oxygen therapy, CPAP works by delivering continuous distending pressure (43). Application of CPAP prevents collapse of alveoli and small airways during expiration, thus increases functional residual capacity (43, 44). HFNC has been shown to improve alveolar ventilation and carbon dioxide elimination (45) by decreasing the dead space through establishing washout in the nasopharyngeal space (46). High flow rates can also provide variable end-expiratory distending pressure based on flow rate (47, 48). Such mechanisms may explain the greater benefit of CPAP and HFNC over standard oxygen therapy in some patients. However, the airway pressure generated by HFNC is determined not only by the HFNC flow, but also by the degree of air leak from both mouth and nose (44). This may partially explain why HFNC displayed a less favorable clinical response than CPAP.

In comparison to CPAP, BIPAP delivers positive airway pressure at two different levels during inspiration and expiration, and could more efficiently decrease inspiratory work of breathing than CPAP (49). However, our pooled results suggested that when compared with SOT, BIPAP did not decrease risk of intubation. It should be noted that most of the evidence that contributed to this comparison was indirect. There was only one trial included in this meta-analysis that directly assessed the effect of BIPAP by comparing it with SOT, and this study did not report treatment failure or mortality. Therefore, conclusions regarding the relative effectiveness of BIPAP are limited, and further research is required.

In our study, we were unable to show benefits among different interventions on in-hospital mortality. Several possible explanations may account for these results. First, among the enrolled studies, 12 (57%) were focused on patients with acute bronchiolitis. Given that very low incidence of mortality has been reported since the introduction of NIV for acute bronchiolitis (33), this would make it difficult to detect differences among various methods of respiratory support. Second, in several trials (32, 33, 36–38), once treatment failure or deterioration occurred, crossover to an alternative respiratory support as rescue therapy was allowed, and the crossover may have avoided exacerbation of respiratory distress. The dilution effect of crossover may partially explain the absence of differences between any of the interventions on mortality.

Among the studies we included, 10 (47.6%) were conducted in high-income countries, where 60%−81% of children improved with standard oxygen therapy alone. Considering that non-invasive oxygen approaches are more costly and complex than standard oxygen therapy, these strategies might not be needed as preferred forms of treatment for mild cases. In resource-rich regions, more research is needed to identify children who are likely to deteriorate on standard oxygen therapy, and therefore who would benefit from the early initiation of non-invasive ventilation. However, at the other end of the spectrum of clinical severity, several included studies (11, 18, 31) from low- and middle-income countries (LMICs) showed case fatality rates of 7–14%. Meta-regression and subgroup analyses of country income (LMICs vs. high-income) found that there were no significant interactions between subgroups with regard to treatment failure and intubation rates. However, due to the low number of studies that reported mortality, subgroup analysis for in-hospital mortality was not conducted. Existing data are scarce on the effect of non-invasive ventilation on mortality of children in LMICs, and the results of previous studies remain controversial (11, 18, 31, 50). More studies will be necessary to define the role of various non-invasive ventilation modalities in resource-limited settings.

Some limitations of this review should be noted. First, there was limited statistical power for some comparisons, such as intubation rate and in-hospital mortality, because of the small numbers of trials forming several of the nodes in this network of the meta-analysis. Notably, only one of the included studies examined the efficacy of BIPAP, indicating that further studies are necessary to obtain more precise effects estimates. Second, this study involved a heterogeneous population with a wide age range (0–18 years) and disease spectrum (bronchiolitis and pneumonia). However, the heterogeneity of the study population may enhance the generalizability of the results. Moreover, to address this problem, multiple subgroup analyses were performed. Multiple subgroup results remained consistent with the overall findings. Hence, we believed that the results of our study were credible. Third, the included trials used variable definitions of treatment failure, which might produce heterogeneity. Finally, when eliminating the McCollum study (18) in the sensitivity analysis, some non-significant findings became significant, which encourages further studies to validate these results.

Compared to standard oxygen therapy, the use of CPAP reduced intubation rate in children with ALRI and the need for supplemental oxygen. Additionally, both CPAP and HFNC decreased the risk of treatment failure when compared with standard oxygen therapy. However, evidence is still lacking to show benefits concerning mortality between different interventions. Further large-scale, multicenter studies are needed to confirm our results.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

ZW is responsible for the conception, data search, data analysis, statistical analyses, data interpretation, writing, and revising the manuscript. YH is responsible for inclusion and exclusion of studies, assessment of methodological quality, data extraction, writing, and revising the manuscript. XZ is responsible for inclusion and exclusion of studies, assessment of methodological quality, data extraction, and statistical analyses. ZL is responsible for the supervision and reviewed the manuscript for important intellectual content. All authors approved the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.