High Flow Nasal Cannula for Weaning Nasal Continuous Positive Airway Pressure in Preterm Infants: A Systematic Review and Meta-Analysis Open Access

Preterm infants often require respiratory support in neonatal intensive care units due to their immature lungs and respiratory system. Nasal continuous positive airway pressure (CPAP) is a widely used form of respiratory support for preterm infants, providing a constant positive pressure to the airways, thereby improving oxygenation and preventing alveolar collapse. However, there is ongoing debate and limited evidence regarding the optimal weaning strategies for CPAP in preterm infants [1]. Traditionally, weaning from CPAP involves a gradual reduction in pressure until reaching a minimum threshold, usually around 4 or 5 cm H₂O, before complete discontinuation. However, certain approaches involve a more gradual reduction in CPAP pressure, tapering off to 3 or 4 cm H₂O before discontinuation. Further, weaning approaches include cycling the CPAP off for shorter periods and gradually increasing the duration of off periods based on the infant's tolerance. However, this method may pose a risk of atelectrauma, a condition characterised by lung injury caused by repetitive opening and closing of the alveoli [2].

In recent years, the use of high-flow nasal cannula (HFNC) as an alternative weaning strategy for CPAP has gained increasing attention. HFNC delivers a heated and humidified blend of oxygen and air at a high flow rate, creating a positive pressure in the airways. It offers several advantages over traditional weaning methods, including better patient comfort [3], reduced nasal trauma [4], and improved access to feeding and care. HFNC can aid in maintaining oxygenation, optimising ventilation, and enhancing respiratory mechanics during the weaning process, much like CPAP, while also providing the additional benefits mentioned earlier.

Despite the increasing use of HFNC as a weaning strategy for CPAP, the available evidence remains limited and contradictory [5‒7]. Few randomised controlled trials (RCTs) have shown encouraging outcomes, indicating that HFNC is effective and comparable to CPAP [5, 6]. However, others have raised concerns about possible lung derecruitment, inconsistent pressure delivery, and prolonged respiratory support duration associated with HFNC [7]. Given these knowledge gaps and conflicting findings, conducting a systematic review becomes essential to consolidate the existing evidence and provide a comprehensive evaluation of the efficacy and safety of HFNC as a weaning strategy for CPAP in preterm infants. By synthesising the results from relevant RCTs, this systematic review aims to inform clinical decision-making and guide future research in this field.

This systematic review and meta-analysis were conducted and reported following the guidelines provided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [8]. The systematic review protocol was registered with PROSPERO, the international prospective register of systematic reviews, under the registration number CRD42022370322 (https://www.crd.york.ac.uk/prospero/.).

We conducted a comprehensive search across Cochrane Central, EMBASE, Medline, and Web of Science from inception through July 15, 2023. No language restrictions were applied. We also reviewed reference lists and citations of included studies and searched the clinical trial registry at https://beta.clinicaltrials.gov/ for potential studies (online suppl. material; for all online suppl. material, see https://doi.org/10.1159/000536464).

Only RCTs focussing on preterm infants with a gestational age of less than 37 weeks and assessing at least one of the identified outcomes of this systematic review were included. Nonrandomised studies, narrative reviews, systematic reviews, case series, and case reports were excluded. Additionally, studies that included both preterm and term populations without separate data available for preterm infants and studies utilising alternative respiratory support for weaning were excluded.

Studies including preterm infants <37 weeks gestational age or a mixed population but providing separate data for preterm infants were included.

The intervention involved using HFNC for weaning CPAP, with a minimum flow rate of 1 L per minute, whereas the comparator consisted of direct CPAP weaning, irrespective of the weaning plan in either of the groups.

The primary outcome assessed in this study was the duration of respiratory support, which is the duration of respiratory support from randomisation until successful weaning off respiratory support according to the authors’ definition. Secondary outcomes included weaning failure, nasal trauma (any and moderate to severe), chronic lung disease (CLD) (any and moderate to severe), severe retinopathy of prematurity (ROP), time to achieve full feeds, and length of hospital stay. The specific definitions and time points for these outcomes were determined by the authors.

The study authors (Y.B. and A.R.) independently screened the titles and abstracts and created a list of full-text articles. Further, they independently evaluated the full-text articles for eligibility. They also independently read the eligible articles and extracted the relevant data: the country, inclusion criteria, exclusion criteria, interventions, weaning protocols, primary outcomes data and secondary outcomes data. Lastly, they resolved the discrepancies through discussion and consensus.

Both authors (Y.B. and A.R.) independently assessed the risk of bias in each study using the Revised Cochrane risk-of-bias tool for randomised trials (RoB 2 version, Aug 22, 2019) outlined in the Cochrane Handbook for Systematic Reviews of Intervention for RCTs [9]. Publication bias analysis was assessed using a funnel plot if ten or more RCTs contributed to the meta-analysis.

Random-effects model meta-analyses were conducted using Review Manager 5.4. For dichotomous outcomes, relative risk (RR) was pooled, while mean difference (MD) was pooled for continuous outcomes, along with their respective 95% confidence intervals (CIs). Median and interquartile range data from the included RCTs were converted to mean and standard deviation in an appropriate manner [10]. The certainty of the evidence was assessed independently using the Cochrane Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach.

Forest plots were visually assessed apart from the p value for χ² (<0.10) and I² values (>50%) for determining significant heterogeneity. Data and analyses were checked in the presence of significant heterogeneity, and subgroup analyses (<28 weeks GA or >28 weeks GA) were considered to explore the cause for between studies heterogeneity if two or more studies are available.

Figure 1 illustrates the study selection procedure and screening. Out of the initial 843 records reviewed, 50 duplicates were identified and removed. Among the remaining 793 records, 751 were excluded based on title and abstract screening. After thoroughly examining the full texts of 42 records, 35 were further excluded for various reasons, including incorrect interventions (n = 19), inappropriate study design (n = 15), and irrelevant outcomes (n = 1). Consequently, a total of 7 RCTs were included in the analysis (Cresi F, 2023 [11], Clements J, 2022 [5], Gregoraci Fernández A, 2021 [6], Soonsawad S, 2016 [12], Tang J, 2015 [13], Badiee Z, 2015 [14], and Abdel-Hady H, 2011 [7]). Included studies were conducted in Afghanistan, Australia, Egypt, Italy, New Zealand, Spain, and Thailand. Table 1 provides a summary of all included RCTs, while Table 2 presents the meta-analysis findings and the GRADE assessment.

A total of 781 preterm infants were assessed in the seven included RCTs. Five of these studies enrolled preterm infants with a mean gestational age of less than 30 weeks. However, in two trials [7, 14], the preterm infants had a mean gestational age of more than 30 weeks. All trials used nasal CPAP as primary support or post extubation support. The studies in the CPAP arm showed some variability in the pressures used, ranging from 5 to 7 cm H₂O. Six studies directly weaned babies from CPAP to room air, while one trial [13] followed a gradual weaning approach, alternating periods of CPAP and time off CPAP. In the HFNC arm, the high flow ranged from as low as 2 to as high as 7 L per minute across the studies, and the weaning process varied based on the individual study protocols, as presented in Table 1. The risk of bias assessment is illustrated in Figure 2, and a detailed summary of the assessment can be found in the online supplemental Material.

Overall, six RCTs demonstrated appropriate randomisation methods and allocation concealment. However, one trial had an unclear randomisation generation method, though the allocation was adequate, ensuring no differences between the groups [7]. Three trials [7, 13, 14] were at high risk of performance bias due to unblinded interventions with insufficient information on adherence to the protocol; however, the other trials were rated as low risk, as they were conducted according to suitable protocols, and all participants received the interventions as per protocol with no deviations. Additionally, two trials [5, 11] had unclear detection bias risk due to insufficient information about whether the outcome could be influenced by knowledge of the intervention received, while the other trials were at a high risk of detection bias since knowledge of the intervention received could likely affect the outcome assessment. Attrition bias was not observed in any of the studies, and while four trials were considered low risk for reporting bias, two trials [7, 14] were registered after the study's completion and were therefore scored as unclear.

The meta-analysis incorporated five trials [6, 7, 11‒13] that provided data on the duration of respiratory support. The findings showed that using HFNC for weaning CPAP did not result in a statistically significant difference in the duration of respiratory support compared to weaning CPAP alone (MD (95% CI) in days 3.52 (−0.02, 7.05); 5 RCTs; participants = 488; I² = 29%). The evidence was downgraded due to limitations within the studies and the imprecision, leading to an overall grade of low certainty (Table 2; Fig. 3). Publication bias was not assessed due to the inclusion of only five RCTs in the meta-analysis. No subgroup analysis was performed due to lack of available data.

Apart from any nasal trauma, no significant differences were observed in the secondary outcomes. The meta-analysis showed HFNC for weaning CPAP resulted in reduced nasal trauma compared to weaning CPAP alone ((RR (95% CI) 0.61 (0.38, 0.99); 4 RCTs; participants = 335; I² = 0%) [5, 6, 12, 13] (Table 2; Fig. 4). The evidence certainty was downgraded due to study limitations and imprecision, resulting in low certainty of evidence. No significant differences were observed in other outcomes, such as moderate to severe nasal trauma ((RR (95% CI) 0.24 (0.04,1.40); 2 RCTs; participants = 185; I² = 0%), weaning failure ((RR (95% CI) 0.86 (0.59,1.26); 6 RCTs; participants = 700; I² = 30%), any CLD ((RR (95% CI) 0.85 (0.08, 8.74); 3 RCTs, participants = 395; I² = 78%), moderate to severe CLD ((RR (95% CI) 0.83 (0.53,1.31); 4 RCTs; participants = 335; I² = 17%), ROP ((RR (95% CI) 0.40 (0.10,1.55); 5 RCTs; participants = 582; I² = 68%), time to achieve full feeds ((MD (95% CI) 0.40 (−2.73, 3.53); 4 RCTs; participants = 481; I² = 6%), length of hospital stay ((MD (95% CI) −0.86 (−5.30, 3.58); 4 RCTs; participants = 421; I² = 0%) (Table 2 and online suppl. e-figures 1–7). The evidence certainty for these secondary outcomes was graded as low to very low certainty, as highlighted in Table 2.

The findings of this systematic review, based on data from seven RCTs encompassing 781 preterm infants, suggest that the utilisation of HFNC for weaning CPAP in preterm infants may not significantly affect the duration of respiratory support but may potentially reduce the occurrence of nasal trauma. It is important to note that the evidence is of low certainty due to limitations in RCTs and imprecision. Additionally, the meta-analysis revealed no statistically significant differences between groups for all other secondary outcomes, with the certainty of evidence ranging from low to very low.

Among the 7 included RCTs in the systematic review, primary outcome data could only be extracted from 5 studies for the meta-analysis. While the meta-analysis did not yield statistically significant differences, a closer examination of the data suggests a potential concern. The pooled estimate indicates that when using HFNC for weaning CPAP compared to weaning CPAP alone, the duration of respiratory support may increase by 3.5 days, with a confidence interval ranging from a slight reduction of 0.02 days to a potential increase of up to 7 days. These findings raise questions about the use of HFNC for weaning CPAP support in preterm infants, as noted in other studies [7, 15‒18], and emphasise the need for further investigation with larger RCTs. Including data from future RCTs may improve the confidence in these estimations.

Similarly, the estimates for all the other evaluated outcomes were imprecise, suggesting that the overall statistical power was inadequate to detect any potential differences between them if they existed. Our findings regarding lower nasal trauma with HFNC align with similar observations in other studies [4, 12], but no significant difference was observed in moderate to severe nasal trauma between the groups. It is crucial to highlight that the confidence in the estimate for any nasal trauma is low due to the imprecise nature of the results, with the smallest observed benefit being either negligible or not clinically meaningful.

Large RCTs and systematic reviews investigating the use of HFNC in various contexts, including primary support [4, 19, 20] and post-extubation [4, 21], consistently report unfavourable outcomes. Additionally, data have raised concerns about HFNCs ability to deliver adequate pressure in preterm infants reliably [7, 15‒18]. As a result, the use of HFNC in preterm infants, especially as primary or post-extubation support, is generally limited. However, its use for weaning CPAP support continues [22, 23], partly due to the absence of data demonstrating that HFNC may not be beneficial. Additionally, clinicians may support the use of HFNC for weaning CPAP as it can aid in transitioning neonates from intensive care to special care units. The simplicity of HFNC use and its comfort, particularly during skin-to-skin care, further endorse its adoption. Nonetheless, our meta-analysis, based on up-to-date data synthesis, provides valuable insights into the use of HFNC for weaning CPAP support, offering guidance to clinicians for informed clinical decision-making and future research directions in this field. Moreover, it is crucial to bear in mind that CPAP use in infants may promote lung growth, as highlighted by studies demonstrating a greater increase in functional residual capacity in babies subjected to extended CPAP [24], an effect that may not occur with HFNC.

Our study has several notable strengths. It is the first systematic review dedicated to examining the benefits and harms of using HFNC in weaning CPAP support for stable preterm infants. By conducting an extensive and comprehensive search of important databases, we have ensured the inclusion of all relevant studies exploring this specific question. Additionally, our use of the GRADE framework to appraise the evidence enhances the interpretability of our results for clinicians. These strengths emphasise the importance and relevance of our study in contributing to the existing knowledge on this subject and guiding clinical decision-making for the management of preterm infants during CPAP weaning with HFNC.

Nevertheless, our study is not without its limitations. The primary constraint lies in the relatively smaller number of participants included in the trials, which likely has affected the statistical power and precision of our results. Another significant challenge arises from the nature of the intervention, as blinding is impractical in this context, leading to potential risks of bias. These factors collectively weakened the certainty of evidence in our study. Furthermore, there is a lack of sufficient data on infants with <28 weeks GA, leaving a gap in understanding the applicability of HFNC weaning in this subgroup, and in the current meta-analysis subgroup data was not available to explore the differential treatment effects.

Moreover, the diversity in flow rates and weaning practices, as well as the variability of CPAP devices used and the population studied among the studies, may introduce heterogeneity and potentially impact the reliability of the comparison. Additionally, the heterogeneity in NICU care, including variations in quality, facilities, and staff competency, is noteworthy, especially given that the studies stem from seven different countries, suggesting potential differences in overall NICU care despite some similarities in the protocols. Nevertheless, we employed a random-effects model and did not find any significant statistical heterogeneity for all except two secondary outcomes. Lastly, the flow rates used in the studies might not adequately reflect current clinical practices, further limiting the generalisability of the findings. It is also crucial to highlight that outcomes may vary based on the timing of high-flow therapy application in the course of ongoing respiratory disease, considering factors such as the severity of lung disease and the duration of respiratory support before initiating the therapy.

In summary, this systematic review suggests that using HFNC for weaning CPAP in preterm infants may not significantly impact the duration of respiratory support but may potentially reduce nasal trauma. However, the evidence certainty is low due to limitations in the trials and smaller sample sizes. Clinicians should exercise caution when considering HFNC for weaning CPAP until additional supportive evidence on its safety becomes available. Future research should focus on larger cohorts, including infants with gestational ages less than 28 weeks, using current clinical practice flows and addressing weaning protocol variability to provide more reliable evidence for clinical decision-making in this vulnerable population.

We want to express our gratitude to Dr. Francesco Cresi and Dr. Elena Spada for providing the pertinent data for the meta-analysis.

An ethics statement is not applicable because this study is based exclusively on published literature.

The authors have no conflicts of interest to declare.

No specific funding was sought for this project. Dr. Razak receives a doctoral scholarship from Monash University and The Lions Cord Blood Foundation. The funding body played no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.

Yasser Balhareth: conception and design, acquisition, analysis, or interpretation of data, and drafting the manuscript. Abdul Razak: conception and design, acquisition, interpretation of data, drafting the manuscript, critical revision of the manuscript for important intellectual content, and supervision. Both authors contributed equally and are joint first co-authors.

All relevant data are submitted with the publication. Further enquiries can be directed to the corresponding author.