Introduction

High-flow nasal cannula (HFNC) therapy has rapidly gained popularity as respiratory support. HFNC therapy has been proved effective in various indications in neonatal care and acute bronchiolitis in infants [1,2,3]. In acute bronchiolitis the HFNC has reduced treatment failure rate compared to conventional oxygen treatment (COT), but it has had similar effectiveness as continuous positive airway pressure (CPAP) [4, 5]. In adults, previous systematic reviews have found HFNC beneficial in preventing escalation to intubation in acute hypoxemic respiratory failure, in preventing extubation failure, and in improving procedural oxygenation [6,7,8,9]. Because of the favoring evidence in these patient groups, the use of HFNC has expanded beyond neonatal respiratory support and bronchiolitis treatment in pediatrics. Simultaneously, there is ongoing effort to reduce the overuse of HFNC in acute bronchiolitis [10, 11]. There are no previous systematic summaries about HFNC use as primary respiratory support for other indications in the pediatric population. A recent systematic review found that HFNC use was associated to higher likelihood of extubation failure in young children [12]. Expanding HFNC use to new patient groups without evidence could also have negative effects such as increased costs and length of hospitalization, prolonged exposure to supplementary oxygen, and delayed escalation of respiratory support.

Previous randomized studies in children have typically compared HFNC to conventional oxygen therapy (COT), and continuous positive airway pressure (CPAP) [13]. The main hypothesis has been that HFNC would be more effective and provide benefits over COT, but be non-inferior to CPAP and better tolerated [14, 15]. Intervention tolerability is especially important in younger children.

To provide better knowledge on current evidence and to guide future studies, we aim to systematically evaluate for which indications HFNC has been studied in randomized controlled trials in pediatric patients.

Methods

Study design and search process

We conducted a systematic scoping review. We searched Pubmed, Scopus, and Web of Science databases in February, 2023. The complete search strategy is provided in Supplementary Materials. Two authors independently screened each abstract and full texts. Cases with conflicting decisions were decided either by mutual consensus or third-party opinion. All authors participated in the screening process.

We have reported our scoping review according to the Scoping review extension for Preferred Reporting Items in Systematic Reviews and Meta-analyses (PRISMA-ScR) [16].

Inclusion and exclusion criteria

We used following PICOS (patients, interventions, comparator, outcome, and study design) as our inclusion criteria. Patients had to be pediatric patients, and we defined pediatric patients as children younger than 18 years. Intervention was high-flow nasal cannula therapy. HFNC was defined by the authors in the included studies. Control intervention or comparator could either be standard low flow oxygen therapy or noninvasive continuous positive airway pressure therapy or other support mode (for example, laryngeal mask airway). We did not specify any pre-selected outcome as either inclusion or exclusion criteria. Study design had to be parallel group randomized controlled trial. Crossover, quasi-experimental, or cluster randomized trials were excluded.

We decided to exclude studies only focusing on acute bronchiolitis in infants, as the evidence regarding this indication is rather solid and covered already by several systematic reviews. Similarly, we decided to exclude all studies which focused on respiratory care of preterm infants and full-term newborns during transition to extrauterine life. However, we included studies where high-flow nasal cannula was used in postoperative care as post-extubation therapy (for example, cardiothoracic surgery for congenital cardiac defects). We excluded studies that did not present original results. Furthermore, we excluded non-English literature.

Main outcomes

Our main outcome for this scoping review was to identify the current indications for which the HFNC has been studied in randomized settings [17]. As we aimed especially to analyze the potential effectiveness of the intervention, we decided to focus on parallel group randomized controlled trials. These are typically the highest standard for evidence of effectiveness. Furthermore, we aimed to analyze the control interventions and the specific design of randomized studies (non-inferiority, superiority, etc.). Finally, we aimed to analyze the most used outcomes. We expect that main outcomes can be stratified in to three themes: clinical outcomes, laboratory parameter outcomes, and cost-effectiveness outcomes.

Critical appraisal

We assessed the risk of bias in the each of the included study by using Cochrane risk of bias 2.0 tool [18]. As the tool is designed to be outcome specific, we decided to conduct the assessment based on the intended primary outcome. Risk of bias analysis was performed by one author with prior expertise of this method (I.K.). Risk of bias figures were generated by using the Robvis shinyapp [19].

Data extraction

Data was extracted by one author and validated by a second author to reduce potential extraction errors. For this scoping review we extracted the following information: authors, journal, title, publication year, study period, country, study setting, intervention, control interventions, inclusion criteria, exclusion criteria, main outcomes, and secondary outcomes.

Permissions and ethics

Permissions for publication were not needed due to study designs. Similarly, our study did not need ethical committee evaluation.

Protocol registration

This review protocol was registered in Open Science Framework (https://osf.io/a3y46/).

Results

Search results and study characteristics

We screened 1276 abstracts and further assessed 43 full reports. Finally, 22 studies were included [20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41] (Fig. 1). Of these, 10 were conducted in Europe, 8 in Asia, 2 in Australia, 1 in South America, and 1 in Africa (Table 1). All studies were conducted in the 2010s or 2020s. Three of the included studies were single blinded and 19 were unblinded.

Fig. 1
figure 1

PRISMA flow chart of the study selection process

Full size image
Table 1 Characteristics of the included studies grouped by study setting
Full size table

Risk of bias

Overall risk of bias was low in 11 studies, had some concerns in 7 studies, and was high in 5 studies (Fig. 2). Most issues were due to bias in randomization process and in outcome measurement. The majority of studies were completely unblinded and caused some problems in the outcome assessment. Furthermore, some issues were seen in reporting the outcomes, as not all studies had prespecified protocol presented or referenced.

Fig. 2
figure 2

Risk of bias assessed in five domains, and overall by using Cochrane risk of bias 2.0 tool

Full size image

Indications and control interventions

The indications and patient groups had high variability. Eight studies were conducted in PICU patients, five studies focused on emergency departments, and eight studies were conducted in perioperative care patients. The patients could be categorized into three main groups: first, patients requiring primary respiratory support for acute respiratory failure; second, patients in the perioperative period needing respiratory support during or after the procedure; third, PICU patients with HFNC as post-extubation respiratory support for other than perioperative use. The control intervention was conventional oxygen therapy in 15 studies, CPAP in six studies, and laryngeal mask airway in one study (Table 1).

Study design

All included studies were randomized controlled trials, of which three were single blinded. The specific designs were superiority trial (13 studies), pilot or feasibility trial (5 studies), non-inferiority trial (3 studies), and equivalence trial (1 study) (Table 1).

Most frequently reported outcomes

Most frequently reported outcomes were clinical outcomes, such as asthma severity, reintubation rate, mortality, and length of stay (PICU and overall). Laboratory outcomes were used especially in perioperative studies where the main interest was gas exchange, typically assessed by arterial pCO2, pO2, and pO2 to FiO2 ratio. Few studies assessed imaging findings, such as presence of atelectasis by lung ultrasound or magnetic resonance imaging. None of the included studies provided cost-effectiveness analysis (Table 1). Adverse effects were infrequently and incompetently reported (presence of hyperoxia or cumulative exposure to supplementary oxygen, rate of accumulation of air into intestines with effect on incidence of nausea and vomiting or on the time needed to achieve full enteral feeds).

Summary of reported results

Six studies analyzed HFNC utilization in emergency departments and general pediatric wards. Three of the studies indicated possible benefits associated with HFNC while the remaining three studies did not identify any significant difference between the HFNC and comparator interventions (Table 2). Additionally, eight studies examined the use of HFNC during procedures or in operated patients. Among these, four studies reported benefits (reduced atelectasis, improved oxygenation), and four of the studies reported no evidence of a benefit of HFNC use. Notably, none of the studies reported increased rates of adverse events (Table 2). Furthermore, eight studies analyzed HFNC use in the context of PICUs. Out these, five studies reported positive outcomes and concluded that HFNC is a feasible or non-inferior option to CPAP or superior to COT. Meanwhile, two studies did not detect differences between treatment groups, and one study found HFNC to be less effective than CPAP as post-extubation therapy (Table 2).

Table 2 Main results and author conclusions of the included studies
Full size table

Discussion

In this systematic scoping review, we found that HFNC has been studied in a variety of pediatric patients and conditions. We identified three key patient groups: acute respiratory failure, perioperative care, and PICU post-extubation respiratory support. Key outcomes assessed were clinical and laboratory outcomes. None of the studies assessed cost-effectiveness.

The most studied patient groups and indications were patients needing primary respiratory support due to acute respiratory failure, followed by perioperative care and PICU post-extubation therapy. The indications were similar for which previous studies in adults have shown benefit or equal effectiveness of HFNC treatment compared to COT or CPAP therapies [42,43,44].

The most frequently used control intervention was COT. All studies comparing HFNC to COT aimed at showing the superiority of HFNC treatment. The second most used control intervention was CPAP, for which either non-inferiority or equivalence designs were used. The design choices were rational, as HFNC should provide benefit over COT and be at least non-inferior to CPAP to be a justified respiratory support mode.

Main outcomes were mostly clinical or laboratory parameters. However, the lack of adverse effect reporting and the complete missing of cost-effectiveness estimations were unfortunate, as in general novel therapies should be safe and preferably cost-effective. Previous systematic review in neonatal patients concluded that there is currently no evidence of HFNC cost-effectiveness against nCPAP in preterm patients [45]. In adult patients HFNC has shown cost-effectiveness in intubation or reintubation prevention in ICU patients, and for chronic obstructive pulmonary disease patients in chronic respiratory failure [46, 47]. A recent systematic review found that HFNC and CPAP were better than COT in preventing extubation failures in infants and young children [12]. In their review CPAP seemed to be the best performing post-extubation support, although the studies were conducted in relatively heterogenous patients.

Enhanced clarity and precision in patient population definitions within future studies would significantly contribute to the interpretability of results. For instance, the inclusiveness of a wide age range (1–14 years) within the same trial investigating acute asthma exacerbations could potentially confound findings. Physiologically, the nature of acute asthma considerably varies between a 1-year-old and a teenager [29]. Moreover, PICU studies have included both all-cause patients or cardiac surgery patients. Notably, trials focused solely on cardiac surgery patients have demonstrated outcomes that hold greater applicability in clinical settings due to the more well-defined patient population. Considering the broad spectrum of patient categories within the PICU, it is evident that HFNC is not the universal solution to all cases.

We detected issues in the risk of bias assessment in the original studies. Most of the issues came from the randomization process and outcome measurement. These issues should be remarked in future trials where the researchers should focus on proper allocation concealment and randomization process and describe those in depth in the final report. Furthermore, an attempt to blind at least outcome assessors in some parts of the studies should be made to improve the reliability. A positive sign was that we did not detect issues with missing outcome data.

This is the largest effort to gather systematic assessment of current literature on HFNC use outside of neonatal respiratory care and acute bronchiolitis infants. We performed a rigorous systematic assessment according to a pre-specified protocol and we did not have any major protocol deviations. Our scoping review provides a basis for future studies and reviews on the use of HFNC.

Our main limitation is the lack of non-English literature, as most likely we have missed some RCTs published in other languages. Furthermore, only one author performed the risk of bias assessment, which can be seen as a limitation. Furthermore, we did not proceed to meta-analysis due to substantial variation in the studies and indications and instead conducted a scoping review of current knowledge and evidence.

Conclusion

In conclusion we found that HFNC has been studied in a variety of settings and indications in children. We identified three key patient groups where HFNC was studied: acute respiratory failure, perioperative care, and post-extubation respiratory support in PICU patients. Key outcomes assessed were clinical outcomes, and none of the studies assessed cost-effectiveness. Further studies should aim to better study quality and assess cost-effectiveness alongside the clinical effectiveness and treatment-related harms or adverse events.