Abstract
High Flow Nasal Cannula therapy (HFNC) is a form of respiratory support for bronchiolitis. Recent evidence confirms HFNC reduces the risk of treatment escalation by nearly half (45%) compared to standard oxygen therapy (SOT), although most patients (75%) with mild-moderate respiratory distress manage well on SOT. The majority of children (60%) failing SOT respond well to HFNC making rescue use of HFNC a more cost-effective approach compared to its first-line use. HFNC is compared to CPAP in the setting of moderate to severe bronchiolitis. Patients on HFNC have a slightly elevated risk of treatment failure especially in severe bronchiolitis, but this does not translate to a significant difference in patient or healthcare centred outcomes. HFNC has improved tolerance, a lower complication rate and is more easily available in peripheral hospitals. It is therefore the preferred first line option followed by rescue CPAP. HFNC is clinically effective and safe to use in bronchiolitis of all severities.
Introduction
Bronchiolitis is the most common respiratory illness necessitating the use of respiratory support in infants in developed countries . Hospitalisation is required in up to 10 % cases , with up to 24 % of hospitalised patients needing intensive care . Smaller airway anatomy, developing alveoli, and weaker respiratory muscles make the infant prone to more severe disease . Prematurity, coexisting heart disease, chronic respiratory illness, neuromuscular disease, or immunodeficiencies increase the risk of needing respiratory support . Occlusion of airways by mucus and debris leads to increased airway resistance with consequent increased work of breathing accompanied by hypoxemia and hypercarbia,while profound central autonomic dysfunction causes apnoeas .
USA and UK guidelines recommend supportive therapy with no evidence for use of steroids or bronchodilators . Determining the most effective form of respiratory support is therefore the focus of current research.
When faced with an hypoxic child with respiratory distress, clinicians must choose from amongst standard oxygen therapy (SOT, up to 4 L/min), heated humidified high-flow nasal cannula (HFNC), or continuous positive airway pressure (CPAP). Factors influencing this decision are clinical outcomes, cost-effectiveness, and availability/tolerance of therapies. The aim of this article is to summarise the current role of HFNC in comparison with low-flow SOT and CPAP in managing bronchiolitis.
Current practice is to start cold dry SOT, typically with 2–4 LPM nasal prong, or up to 15 LPM by face mask as first line treatment for children with mild-moderate respiratory distress, targeting SpO2 > 90 % if aged over 6 weeks, or > 92 % if aged under 6 weeks or with significant comorbidities .
HFNC is a therapy capable of delivering adjustable and relatively constant high-flow (2–60 L/min) of gas with desired oxygen concentrations (21 %∼100 %), temperature (34 ∼ 39 ℃) and relative humidity (up to 100 %) through nasal prongs . It has a number of physiological effects: pharyngeal dead space washout, reduced work of breathing, precise and titratable FiO 2 delivery and improved mucociliary clearance .
CPAP stents open oedematous bronchioles preserving functional residual capacity preventing lung collapse. HFNC cannot provide a consistent PEEP above 5 cm H2O . UK NICE guidelines recommend CPAP as first line treatment in very sick infants .
Prior to introduction of HFNC, the only available option in children needing escalation beyond SOT, was CPAP/BiPAP followed by intubation. But peripheral hospitals lack the resources to deliver CPAP to children weighing > 5 kg in the UK and many other countries. This means babies aged approximately > 6 months failing SOT need transfer to a paediatric intensive care unit (PICU) for CPAP/ intubation. A proportion of these children are intubated for safe transfer rather than a need for invasive ventilation. HFNC provides an interim ‘rescue’ support, bridging the gap between SOT and intubation in this subgroup allowing these children to be managed without transfer to another hospital. It is not surprising therefore that intubation rates have dropped following introduction of HFNC . There is a gap in paediatric respiratory support escalation pathways, which is filled by the easy-to-use and well-tolerated HFNC.
Comparison of HFNC with SOT in mild-moderate bronchiolitis
Evidence suggests 67–87 % children under five years of age with respiratory distress and mild hypoxemia improve with SOT alone . It is therefore important to review the evidence comparing HFNC with SOT to prevent unnecessary escalation of treatment to high flow.
Treatment failure
‘Treatment failure’ refers to the necessity to escalate treatment. A recent 2024 Cochrane review comparing HFNC vs SOT (2012–2022, 8 studies, 2215 patients) reported that primary HFNC use reduced risk of treatment escalation by nearly half (45 %) for an individual patient (RR 0.55 (0.39 to 0.79) . Four RCTs did not have objective criteria to define treatment failure and all trials allowed escalation based on clinician preference. This, along with significant heterogeneity in the treatment failure criteria set by the remaining four trials, lowered the quality of evidence to moderate. The addition of hypertonic saline and use of a different oxygen delivery system (Oxymask) did not change the risk of treatment failure . In contrast, a more recent multicentre RCT including 268 patients < 6 months with moderate bronchiolitis reported no difference in failure rates between 3L/kg/min HFNC (14 %) vs. SOT (20 %) (OR 0.66, 95 % CI 0.35–1.26; p = 0.21). This is despite HFNC showing a significant improvement in respiratory rate (RR) and work of breathing in the first 1–6 hrs of treatment.
The largest included RCT, Franklin et al. (n = 1472 children) shows 11 % more children (95 % confidence interval, −15 to − 7; P < 0.001) in the SOT group (23 %) needed escalation. The study concluded 9 children treated with HFNC prevented one instance of escalation (95 % CI, 7 to 14). The higher treatment failure rate was not associated with younger age < 3 months, compared to 3–6 or > 6 months. Children receiving SOT in hospitals without an on-site PICU had a higher risk of treatment failure (28 %) (RD − 21 percentage points vs. HFNC) compared to those with an on-site PICU (20 %) (RD −6 percentage points vs. HFNC). The unblinded nature of these studies meant that clinician bias led to greater difference in treatment failure depending on ease of access to PICU care. True difference in treatment failure rate was lower at −9% (7 % in HFNC vs. 16 % in SOT; 95 % CI, −12 to − 5; P < 0.001) once objective criteria for escalation are applied similar to the –6% reported by Durand et al. 34 % infants not meeting prespecified objective failure criteria still ‘failed treatment’.
Evidence confirms faster improvement in physiological parameters with use of HFNC may explain the lower treatment failure risk. HFNC reduces length of wheezing (two RCTs, total 114 patients, MD − 2.29, 95 % CI − 2.85 to − 1.72, p < 0.01) . HFNC significantly reduces ‘clinical severity scores’, respiratory rate (RR) and heart rate (HR) compared to baseline in 1–6 h of starting treatment, with the effect persistent at 24 h . A PICU study of 60 patients, with 93 % described as severe bronchiolitis, similarly shows significant improvement in saturations in the first 12 h of treatment in the HFNC group but no significant difference in CO 2 clearance. Yang et al. in their small RCT (54 patients) reported a significant reduction in PaCO 2 and increase in PaO2 using HFNC compared to SOT.
Various studies have tried to identify risk factors for treatment failure on HFNC. Prematurity, RSV aetiology and previous PICU admission were not seen to significantly affect treatment failure in either group. . A large retrospective study of 2657 children aged < 24 months on HFNC for bronchiolitis identified age < 3 months and greater respiratory distress score as risk factors predicting intubation. . Factors associated with lower risk of intubation were a reduction in tachycardia after initiation of HFNC and presentation after day 5 of illness. Another retrospective study highlighted significant tachycardia, dehydration, and a venous pH < 7.30 as risk factors predicting intubation. . A high pCO 2 value has also been correlated with a higher risk of HFNC therapy failure .
‘Treatment failure’ is irrelevant if it does not translate into a significant difference in outcome measures that matter to families and healthcare systems. Simply put, a higher risk of treatment failure implies the need to have HFNC available as a mode of rescue treatment but does not make a case for early HFNC.
A meta -analysis in 2020 (four RCTs, 1753 children in PICU, ED and general wards) concluded that 60 % children who fail SOT are successfully managed on HFNC and no difference is seen in the need for PICU transfer between the treatment groups .
PICU admission and rates of intubation
Lin et al. ( meta -analysis of 1526 patients, dominated by Franklin et al .) suggested no significant difference in incidence of intubation between HFNC and SOT groups (RR 1.98, 95 % CI 0.6 to 6.56, p = 0.26) and no significant reduction on transfer to ICU in the HFNC group (RR 1.30, 95 % CI 0.98 to 1.72, p = 0.06). These findings were echoed in subsequent meta -analysis (rates of intubation, 4 studies, OR 0.94, 95 % CI 0.42 to 2.13, p = 0.17; admission to PICU, 7 studies, 1223 patients, OR 1.10, 95 % CI 0.84 to 1.42, p = 0.49). Increasing the flow rate to 3L/kg/min also does not change the PICU admission rate (15 % vs, 19 % in SOT, p = 0.45) .
The reduction in intubation rate associated with HFNC use from retrospective studies is therefore unlikely to be related to an actual therapeutic advantage and more likely related to easy availability of an interim rescue respiratory support system, which was lacking earlier.
Length of hospital stay (LoS)
HFNC may influence LoS because of a lower likelihood of treatment failure, clinician perception of more invasive respiratory support, and lack of weaning protocols.
Recent pooled data meta -analyses suggest HFNC reduces LoS by 15.6 h compared to SOT (7 studies, 1951 patients, MD −0.65 days lower (95 % CI −1.23 to −0.06)) . It is however worth noting exclusion of two quasi-randomised trials moves the result closer to the null with a mean difference of −0.32 i.e., 7 h (95 % CI −0.82 to 0.2).
One review suggests infants < 6 months (mean difference − 3, 95 % CI − 4.33 to − 1.67, p < 0.001) and those with LoS ≥ 4 days (MD − 2.32, 95 % CI − 3.12 to − 1.52, p < 0.001) might benefit more from earlier HFNC, while higher flow rates do not reduce LoS.
Length of oxygen supplementation
HFNC allows accurate titration of oxygen . This is both clinically and economically advantageous. While earlier reviews including large RCTs failed to show a clear advantage of HFNC in reducing length of oxygen supplementation, the addition of newer studies since 2019 has meant the 2024 Cochrane meta -analysis (7 studies, 2132 patients) now shows HFNC reduces the duration of oxygen supplementation by approximately 14 h (MD of −0.59 (95 % CI −1 to −0.18). Unfortunately, considerable heterogeneity with absence of clear weaning protocols in most RCTs lends this result low certainty evidence. Also, it is worth noting the largest RCT in the analysis (though not contributing the greatest weight) shows no difference in the two groups. Durand et al. showed HFNC reduces length of oxygen use, but the difference is insignificant in line with Franklin’s results .
Cost effectiveness
The consumables cost of the HFNC circuit is up to 16 times higher than SOT . Kepreotes et.al . demonstrated using HFNC as rescue therapy is cheaper (AUS$ 111 990) than using it as primary therapy ($135 513), even after accounting for the higher treatment failure in the former group .
A cost-effectiveness analysis using decision tree modelling estimates the treatment cost of each episode of bronchiolitis is lower at €1,312–2,644 ($1,786–3,600) in the HFNC group compared to €1,598–3,764 ($2,175–5,125) in the SOT group . The lower costs in the former are driven by lower rates of PICU admission. This is based on data from a retrospective case-control study rather than a randomised controlled trial. It does not factor in the cost of higher nursing: patient ratios needed to manage a HFNC patient and the resource burden this adds during seasonal peaks.
HFNC, independent of its comparison with SOT, has over time reduced the need for PICU admissions and intubations by providing an effective rescue treatment. Its cost-effectiveness therefore lies in its judicious use in firstly reducing ICU bed occupancy during seasonal epidemics and secondly, reducing inter-hospital transfers for CPAP in > 5 kg infants .
Availability, tolerance and complications
SOT requires a tubing attached to an oxygen flowmeter and is therefore more widely available compared to HFNC which requires a medical device to deliver it. Paediatric teams may only invest in a limited number of machines which become a precious resource at the height of the bronchiolitis season.
HFNC is well tolerated as reported by multiple RCTs . Subjectively, parents’ report better comfort and ability to feed with HFNC but no difference in ability to sleep . Objectively, HR and RR at 1 h and 24 h is lower with HFNC . Weak evidence suggests infants are fed for slightly longer in the HFNC group and there is no significant difference in nasal mucosal or skin trauma .
The relative risk of complications (pneumothorax, pressure sores, epistaxis and tubing/circuit connections) is higher with HFNC compared to SOT (RR 1.20 (95 % CI 0.38 to 3.74; 4 studies, 1789 participants; low-certainty evidence) . However, studies report wide confidence intervals with high heterogeneity. Franklin reported only one pneumothorax in both groups, no emergency intubations or cardiac arrest . TRAMONTANE 2 comparing 2L/kg/min HFNC to 3 L/kg/min reported a higher risk of discomfort (43 % vs. 16 %, p = 0.002) and longer PICU stay (6.4 vs. 5.3 days, p = 0.048) with higher flow rate but no increase in the risk of air leak . Durand et.al . in contrast reported a higher incidence i.e., three self-resolving air leak syndromes using HFNC at 3L/kg/min .
Use of HFNC versus CPAP in moderate to severe bronchiolitis
Prior to the introduction of HFNC, nasal CPAP/BIPAP were the key escalation therapies in patients failing SOT. In recent years, several large, high-quality studies, and subsequent meta -analyses, have compared HFNC versus CPAP, in moderate to severe bronchiolitis. Key outcome parameters of interest include efficacy, safety, cost-effectiveness, availability and treatment compliance.
Treatment failure
Pooled data meta -analysis (5 RCT 2012–22, 541 children) shows the risk of treatment failure is significantly lower with CPAP than HFNC [OR: 0.51, 95 % CI (0.36, 0.75), I 2 = 0 %] (low quality evidence) . Both treatments show equal improvement in RR, saturations, work of breathing, pO2 and pCO2 over 2––48 h .
The TRAMONTANE trial, an RCT of 142 infants with moderate to severe bronchiolitis, compared HFNC 2L/kg/min with CPAP 7 cm H 2 O in children aged < 6 months . HFNC had a higher risk of treatment failure (51 %) compared to CPAP (31 %), a risk difference of −19 % (95 % CI −35 to −3%)). They reported a rise in modified Woods Clinical Asthma Score (mWCAS) in 29.6 % of the high-flow group compared to 14.1 % in the CPAP group (P = 0.04). Discomfort was the leading cause of treatment failure with CPAP, while respiratory distress was more frequent with HFNC . Overly sensitive failure criteria (1-point increase in mWCAS or discomfort EDIN score, RR rise > 10/min with RR > 60/min) resulted in a higher overall failure rate (40 %) than initially predicted (15 %) from previous studies. This amplified the difference .
In contrast, a subsequent Brazilian pilot RCT showed no significant difference in treatment failure between HFNC and CPAP for moderate-severe bronchiolitis (n = 63, <9months, RR 1.07, 10 (35.7 %) CPAP group vs. 13 (37.1 %) HFNC group (p = 0.88)). Treatment failure in this study was defined as clinician need for escalation to BiPAP or intubation; crossover between HFNC and CPAP was not allowed. The equity comes from the lower failure rate with HFNC in this trial. Possible explanations for this include a greater proportion of moderate rather than severe bronchiolitis, smaller numbers, or absence of the option to swap therapies leading to higher clinician threshold for declaring treatment failure to avoid intubation. This explains the considerably longer median time to treatment failure (15.2 h in the HFNC group and 18.5 h in the CPAP group), compared to 6 h for both groups in the TRAMONTANE study. The lower HFNC failure rate is mirrored in the more recent TRAMONTANE 2 trial (38.7 %, 2016–2017). Increasing confidence in use of HFNC over time and changes in practice, for example, accepting lower oxygen saturations prior to escalation, may account for this trend.
Switching between CPAP and HFNC after primary treatment failure results in 70–80 % success with no significant difference in the duration of overall respiratory support . In fact, in the TRAMONTANE trial, the success rate for HFNC after failing CPAP is higher (18/22; 82 %) compared to rescue CPAP (26/36; 72 %) . This means patient specific factors influence treatment success independent of advantages of either therapy, disease related factors or clinician bias. Future studies should focus on patient-specific factors which help individualise therapy, rather than what is better as blanket first line therapy for all patients.
A higher FiO 2 need is the only predictor of HFNC treatment failure (p = 0.02) in TRAMONTANE . Higher weight predicts failure in the CPAP group. In both arms, apnoea is a risk factor for treatment failure . Prematurity is not a significant risk factor .
Patients ‘failing treatment’ have timely and protocolised escalation to alternate modes of respiratory support. This means treatment failure does not correlate with more intubations or PICU admissions, longer PICU or hospital LoS and greater mortality .
Rate of intubation
The most recent meta -analysis (5 RCT, 541 patients) shows intubation risk is not significantly different between CPAP and HFNC [OR: 1.18, 95 % CI (0.74, 1.89), I 2 = 0 %].
The intubation rate in both arms of TRAMONTANE is very small (3/71 (4 %) CPAP and 5/71 (7 %) HFNC, p = 0.72) because cross-over between treatment arms was allowed . In contrast, studies not allowing crossover show a higher intubation risk in both groups (3/28 (10.7 %) CPAP vs. 7/35 (20 %) in the HFNC group) .
Length of hospital and PICU stay
Pooled data meta -analysis (2 RCTs , 318 children) suggests no significant difference between CPAP and HFNC in hospital [MD =: −0.22, 95 % CI (−0.91, 0.48), I 2 = 27 %] or PICU LoS (MD − 0.15, 95 % CI − 1.27 to 0.98, p = 0.80) .
Vahlkvist et al. described longer treatment using HFNC (95 (27–193) hours) vs. CPAP (70 (6–240) hours), but the study was not sufficiently powered to demonstrate a significant difference .
In contrast, HFNC did not significantly prolong time on respiratory support in the FIRST-ABC step-up trial in the subgroups with bronchiolitis and severe disease . But in the overall cohort including LRTI patients, length of PICU stay was significantly shorter in the HFNC group (5 days vs 7.4 days, MD − 3 days; 95 % CI, −5.1 to − 1 days). The mean duration of hospital stay was also lower in the HFNC group (13.8 days vs 19.5 days; MD − 7.6 days (95 % CI, −13.2 to − 1.9 days).
Cost-effectiveness
Cost-effectiveness information is lacking, specifically in a bronchiolitis cohort comparing HFNC to CPAP. A cost-effectiveness trial in special care nurseries in Australia looking at neonatal respiratory distress syndrome concluded no difference in cost effectiveness if both therapies were easily available and patients could escalate from HFNC to CPAP in the event of treatment failure .
Availability, tolerance and complications
Nasal CPAP is currently easily available for infants < 5 kg (on average < 6–8 months of age) but specialised equipment (for example, ventilators, NIPPY™/Resmed™) to deliver CPAP in older infants weighing > 5 kg, is not available in hospitals with no on-site PICU. General paediatric staff are not trained to provide CPAP using specialised ventilators. By contrast, HFNC is widely available now in all paediatric areas and staff are well-trained in its use.
Patient tolerance is an important factor determining success of non-invasive ventilation. In the TRAMONTANE trial, patient comfort was assessed using the EDIN score, and a higher proportion of patients in the CPAP group experienced treatment failure due to a rise in their EDIN score than in the HFNC group (18.3 % vs 8.5 %) . Similarly, Vahlkvist et al. showed the neonatal infant pain score was higher in the CPAP group . In the FIRST-ABC step-up trial, 50 % of patients on HFNC continued HFNC at 24 h compared to only 30 % on CPAP, indicating better tolerance of HFNC Overall, fewer patients switched from HFNC to CPAP (20 % vs. 30.9 % from CPAP to HFNC). Patients predominantly switched from CPAP to HFNC due to patient discomfort. In addition, the use of sedation was significantly lower in the HNFC group (27.7 % vs 37 % for CPAP) which is an indirect measure of better tolerance.
In general, for both HFNC and CPAP, complications are low in the setting of bronchiolitis, and mostly relate to pressure damage or nasal trauma which is more common with CPAP (2 studies, OR 0.20, 95 % CI 0.06 to 0.63, p = 0.006) . Combined risk of complications (skin lesions, poor tolerance, abdominal distension) is significantly higher with CPAP [OR: 3.39, 95 % CI (1.48, 7.77), I 2 = 7 %] (low quality) .
In the TRAMONTANE trial, skin lesions were rare in both groups, CPAP (8.5 %) and HFNC (2.8 %), and were staged at NUPAP pressure injury stage 1, predominantly nasal redness . In the FIRST-ABC step up trial, the proportions with nasal trauma are 6/295 (2.0 %) in the HFNC group and 18/278 (6.5 %) in the CPAP group .
An RCT looking at feed tolerance in neonates with respiratory distress shows no difference in the two therapies . Choice of initial support does not significantly affect mortality .
Conclusions
HFNC is more effective than SOT in reducing the work of breathing and improving gas exchange in bronchiolitis, therefore providing higher odds of treatment success. Additionally, the two most recent reviews suggest, albeit based on low quality evidence, a significant reduction in hospital length of stay and length of oxygen utilisation ( Table 1 ). These patient-centred and cost-effective outcomes raise the question − should HFNC be used as primary therapy rather than rescue therapy?
