What is it?

The use of high flow nasal cannula (HFNC) has increased as a popular modality of treatment for patients hospitalized with bronchiolitis. Since the introduction and rapid adoption of HFNC, data has shown that both in and outside of the ICU, HFNC is being used as a therapeutic intervention for respiratory symptoms in the majority of children hospitalized for bronchiolitis. .

In clinical practice, HFNC is a tool that has been used to treat both hypoxia and increased work of breathing associated with respiratory distress. The HFNC circut is a closed system machine consisting of an oxygen blender, a heated humidifier, a water reservoir and a pressure valve that can be used to deliver heated and humidified oxygen at flow rates higher than what is typically delivered via a standard nasal cannula. The mix within the circuit is a blend of air and oxygen that can produce variable oxygen concentrations, and can be delivered at adustable flow rates starting at 2 L per minute. The ability to independently adjust the flow rate and provided FiO ₂ (fraction of inspired oxygen) is a benefit of HFNC over standard supplemental oxygen therapy. Additionally, the high flow gases that are inhaled are both heated and humified, making it more comfortable for the patient. The ability to deliver high flow rates may also contribute to dead space wash out and reduction in upper airway resistance. These benefits of use, however, can be nullified by patient specific factors including patient size, habitus, positioning, nasal prong sizing, rate of flow and open vs closed mouth breathing. .

Physiology of HFNC

Two key physiologic features that have been studied in HFNC are dead space wash out and delivery of positive pressure. Nasopharyngeal and oropharyngeal dead space washout occurs by flushing in oxygen continuously while flushing out carbon dioxide continuously. Theoretically, this dead space washout is occurring in HFNC administration in combination with positive pressure delivery. Positive end expiratory pressure (PEEP) is created when the opposing forces of the high flow air moving into the larynx pushes against the expiratory flow of the patient breathing out, which creates resistance and as a result generates positive pressure. However, the ability to deliver positive pressure via HFNC can be affected by several factors including nasal prong size, open vs closed mouth of the patient, patient size, size of nares and nasal secretions. These variable patient factors make it such that HFNC cannot provide a consistent or measurable PEEP.

Studies that have investigated PEEP generated by HFNC have either been conducted by measuring the generated tracheal or pharyngeal pressure in airway models. Most of these studies found that ideal scenarios for generating PEEP included a closed mouth patient, which in clinical practice is difficult to replicate. In simulated upper airway models of term infants and toddlers attached to a mechanical test lung, the flow required to generate PEEP of 6 cm H2O in a term neonate was 5–7 L/min and in a toddler was 14–20 L/min in models with closed mouths. In these models, the closed mouth system showed improved CO ₂ clearance with utilization of HFNC as low as 2 L/min in term neonates and toddlers. However, a fifty percent decrease in PEEP was seen in these same patient models when simulated with an open mouth. In studies that measured PEEP within the trachea of paediatric airway models, there was an overall increase in tracheal PEEP with increase in flow rate. In closed mouth preterm infant airway models, as little as 7–8 L/min generated a tracheal PEEP of 6 cm H ₂ O. However, in a closed mouth tracheal model of a young child (5yrs, 20 kg) the flow needed to achieve a tracheal PEEP of 5 cm H ₂ O was 28–32 L/min. These closed mouth models required significantly more flow to reach this PEEP when compared to the closed mouth infant model. The observed variability between closed and open mouth models as well as a flow rate that is required for larger infants or toddlers, which is higher than is typically used in common practice for bronchiolitis, makes it unlikely that consistent positive pressure is occurring for patients hospitalized for bronchiolitis who are on HFNC therapy.

Similar limitations in PEEP generation have been demonstrated in live patients. When pharyngeal pressures were measured in hospitalized infants ≤ 6 months of age with RSV, pharyngeal pressures increased from 0.2 cm H ₂ 0 on a flow rate rate of 1 L/min to 4 cm H ₂ 0 on a flow of 6 L/min or higher. However, consistent positive pressure was again affected by air leak both around the nasal prong and through the mouth. The air leak from the patient’s mouth was attempted to be minimized by the use of a pacifier, which was only intermittently successful and not a reliable mitigation approach. .

Early data in support of HFNC

Outside of the controlled laboratory setting, the clinical application and impact of HFNC utilization on patient centered outcomes was being evaluated at hospitals across the world.

There is likely a subset of patients with bronchiolitis who benefit from HFNC, and initial observational, small, single center studies looked promising. Early in the evolution of HFNC utilization, HFNC was identified as a possible tool to improve patients’ vital signs, reduce work of breathing, decrease arterial CO ₂ and decrease paediatric intensive care unit (PICU) admission rates. These studies, although illuminating, had several limitations to them including that most were non-randomized, single institution studies with small sample sizes. Mayfield found that early initiation of HFNC instead of LFNC (low flow nasal canula) resulted in decreases in heart rate, respiratory rate and PICU admission. Importantly, this was shown to be done without adverse events such as pneumothorax, bradycardia, bradypnea, emergency intubation or cardiopulmonary resuscitation. .

Other studies supported early initiation of HFNC leading to decrease in heart rate, respiratory rate and higher oxygen saturations. Milani reported similar outcomes with HFNC producing faster improvements in patient’s respiratory rate, heart rate, respiratory effort and ability to feed, as well as decreasing the length of hospitalization. HFNC was also posited to keep kids more comfortable, with one study showing that HFNC both improved respiratory distress scores (a combination of respiratory rate, dyspnea, retraction and wheeze) and improved COMFORT scores as defined by alertness, calmness, physical movement, mean arterial pressure, heart rate, muscle tone and facial tension. .

Meta-analyses have been done to establish the safety and efficacy of high flow oxygen for children with bronchiolitis within acute hospital settings. These studies show that HFNC is superior to low flow oxygen in terms of treatment failure and reduced length of stay. These studies also show a lack of adverse events which support its safety profile and use on the general paediatric wards. It is however important to note that not all studies had the same definition of treatment failure. .

The safety and efficacy of HFNC had been established for children with bronchiolitis within acute hospital settings. These initial in vivo studies in infants suggested that HFNC was superior to low flow oxygen without associated adverse events. Not only was there improvement in objective findings such as vital signs, several small studies subsequently identified it as an intervention that could significantly reduce intubation rates. These single-center studies identified significant reductions in intubations by comparing intubation rates prior to the introduction of HFNC at their institution to intubation rates after the introduction of HFNC as a treatment for respiratory failure in bronchiolitis. Three studies showed a statistically significant reduction in intubation between bronchiolitis seasons, with rates of intubation decreasing by 50 % in one study to as high as 80 % in another, when comparing pre HFNC seasons to post HFNC seasons. .

After these small, observational studies showed favorable outcomes in use of HFNC, small randomized controlled trials were undertaken. In one study, patients with bronchiolitis admitted to the paediatric ward were treated with HFNC at a flow rate of 2L/kg/min and were then compared to patients treated with LFNC. Those on HFNC were found to have a shorter duration of oxygen therapy before returning to baseline heart rate and respiratory rate. Additionally, the children treated with HFNC in this study were noted to have improvements in perceived respiratory distress more quickly than those on LFNC. Comparisons were made not only to LFNC but other forms of oxygen delivery. When compared to diffuser oxygen mask, HFNC was found to have a greater decrease in respiratory rates, heart rates, time to weaning off oxygen and length of ICU stay when compared to use of a diffuser mask. .

However, it is important to note that these improvements in respiratory distress and vitals may not be appreciated at all flow rates. These benefits may not be seen until flows of 1.5–2.0L/kg/min. Studies like those above drove a general trend to use HFNC for work of breathing in infants, and not necessarily for hypoxemia. Most studies, however, compare respiratory distress scores between LFNC and HFNC in hypoxemic infants. Use of HFNC in non-hypoxemic infants is not currently supported by the evidence and should not be considered. .

Large randomized controlled trials told a different story

Following these smaller single center studies, several larger randomized controlled trials (RCTs) were undertaken. Three RCTs, including nearly 2,000 patients, compared early use of HFNC (1L/kg/min to 3L/kg/min) versus standard oxygen therapy (max 2L), and examined outcomes such as length of stay, ICU utilization and intubation rates.

Franklin et al (2018) included 1638 patients with mild to moderate bronchiolitis who were randomized to HFNC at a rate of 2L/kg or supplemental oxygen via LFNC. Primary outcomes showed that there was a larger degree of treatment failure (23 %) among the standard therapy group when compared to the HFNC group (12 %), however treatment failure in this study was defined as escalation to HFNC. Secondary outcomes showed that there was no difference between groups when it came to duration of hospital stay, duration of stay in the ICU, duration of oxygen therapy or intubation rates. .

A separate RCT included 202 patients who were randomized to standard oxygen therapy or HFNC at 1L/kg. In this study, fewer children in the HFNC group experienced treatment failure when compared to standard therapy, where treatment failure was defined as critically abnormal vital signs, severe respiratory distress scores or as deemed by the medical provider. Similarly, they did not find any significant difference in length of hospital stay or adverse events. .

Durand (2020) included 268 infants randomized to HFNC at 3L/kg or standard therapy, and did not show a difference between groups when it came to treatment failure, reduced ICU admission rates or intubation rates. However, the HFNC group showed improved respiratory status scores at 1 h and improved respiratory rates at hours 6 and 12. .

All three studies, which included 2108 patients, showed no major differences in duration of hospital stay, ICU admission rates or duration of stay in the ICU, duration of oxygen therapy, intubation rates, heart rate, respiratory rate or comfort scores. Additionally, two of the three studies allowed for crossover from LFNC to HFNC if the patient had treatment failure on LFNC. In both studies, patients who underwent crossover also did not have any differences when it came to secondary outcomes when compared to patients who were started on HFNC initially. Taken together, this data suggests that early use of HFNC compared to later use after failure of standard oxygen therapy (“rescue use”) does not alter disease course or impact important patient outcomes. .

An updated Cochrane Review published in early 2024 included 16 RCTs, and aimed to assess the HFNC therapy compared to standard oxygen therapy in the management of infants with bronchiolitis. Despite inclusion of 16 trials and over 2800 patients, the authors were unable to make strong conclusions or recommendations regarding HFNC use. The study authors concluded that there may be a reduction in hospital LOS when patients are initially treated with HFNC, with a reduction in LOS of 0.65 days. It should be noted that the avaerge LOS for the studies included in this review was 4.15 days, which is longer than the average length of stay for US hospitals. Authors of the review also conclude that “these reductions are modest and may not have significant clinical impact when applied in the real world.” We would posit that after 20 plus years of research, multiple randomized controlled trails and systematic reviews— data supporting early, routine use of HFNC in bronchiolitis remains lacking. .

Protocols

Despite the lack of convicing data on routine use of HFNC in patients hospitalized with bronchiolitis, its use continued to rise. Multiple studies have now shown that approximately half of all patients admitted with bronchiolitis are treated with HFNC therapy. Associated with this increase in utilization was the increase in the protocolized use of HFNC in the non-ICU setting. Studies looking at nationwide use of HFNC protocols found that approximately half of hospitals surveyed used HFNC on the wards, and of those 76 % of them used locally developed protocols. Protocols varied in terms of initiation flow rates, which were determined by several factors including weight, age or local protocols, and maximum flow rates ranged from 6-15 L/min, which was determined based on age, weight or neither, with some insitutions having a standard maximum flow rate outside of the ICU. .

With the evolving landscape of HFNC outcomes data, efforts were made to standardize HFNC use outside of the ICU setting. This use and protocolization had the potential to decrease ICU utilization and overall cost of the hospitalization. It also had the potential to decrease hospital length of stay as HFNC was no longer solely provider driven but rather managed by a multidisciplinary team including providers, nurses and respiratory therapists. Implementation of protocols was also found to decrease average hospital LOS, average ICU LOS and average HFNC duration without an increase in ICU readmission or escalation of respiratory support. These protocols helped the local teams decrease variability surrounding when to start HFNC, how much flow and oxygen to start, frequency of reassessment, and de-escalation/weaning parameters. This standardization helps to promote high value care—but there is no agreed upon standard for any of these variables outside of each institution. This reveals the need for evidence based recommendations on HFNC use in bronchiolitis.

Flow rates

Flow rate determination has shifted over time, and varies institution to instiution. Prior to 2010, the majority of hospitals with HFNC protocols reported having aged based crtierias for determining appropriate initial flow rates, with the transition to more weight based flow rate protocols over the last 5 years. This change to weight-based protocols over time was supported by evidence documenting increased ICU utilization in institutions with age based flow parameters (36 % compared to 21 % in weight based protocols) and the length of stay was also significantly longer at institutions that followed age based protocols (2.9 days compared to 1.9 days). However, even weight based flows differed from institution to institution and many were left wondering, how much flow is ideal?

Understanding how to determine the ideal flow rate for each patient is essential to guiding appropriate use of HFNC. The initial in vitro based studies had been performed on closed mouth, artificial lung model systems- and it is not clear how these results should translate into clinical practice. Physiologic studies suggest that a weight-based flow rate would be more likely to achieve similar effects of positive pressure by providing adequate pharyngeal pressure, and to subsequently improve breathing patterns and reduce the increased workload on the respiratory muscles. With limited success in identifying meaningful clinical outcomes with the use of HFNC, it was hypothesized that perhaps we are using the wrong “dose” for the treatment. Several studies set out to determine the optimal weight-based flow rate to prevent treatment failure, as defined by persistent abnormal heart rate, respiratory rate or oxygen levels or escalation of respiratory support, and possibly deliver positive pressure in patients instead of in lung models. Positive inspiratory pressure can only be achieved if the high flow rate is greater than the patient’s own inspiratory flow which on average was seen to be 1.7L/kg in patients 12 months old and younger. This was congruent with earlier in vitro studies measuring either pharyngeal or esophageal pressure. Flow rates of 1.7L/kg/min in human subjects have been shown to demonstrate significantly higher esophageal pressures than with low-flow oxygen therapy. When comparing 1L/kg/min, 1.5L/kg/min and 2L/kg/min the optimal flow rate was found to be either 1.5L/kg/min or 2L/kg/min. Those initiated on 1L/kg/min were found to have a longer length of stay and three times more likely to experience treatment failure. Therefore, to achieve positive inspiratory pressure it is theorized that flows greater than 1.7L/kg are needed to achieve 4 cm H ₂ O of pharyngeal pressure. Thus, there was support for a shift in age based protocols to weight based protocols given these flow rates had been shown to be necessary to generate any form of positive pressure or to improve work of breathing. A subsequent pattern of increased median maximum flow rates was seen in hospital protocols, with early maximum flow rates around 0.85L/kg/min, when it was provider driven without a standardized protocol, to 1.63L/kg/min when driven by weight-based protocols. .

Papers continued to show a dose dependent effect on pharyngeal pressures but there was uncertainty over whether the dose-dependent pattern would eventually plateau or cause adverse events. In other words, how much HFNC is too much? Perhaps increasing flows to 3L/kg/min would further support patients with bronchiolitis and improve outcomes. Milesi et al. found that HFNC acts through a threshold effect rather than a dose effect. When increased to 3L/kg/min infants were found to have longer durations of stay in the PICU mainly due to treatment failure secondary to discomfort and the eventual need to reduce the flow rate. Additionally, measures of work of breathing (an equation using esophageal manometry and respiratory rate) to assess response to different weight-based rates of high flow showed a plateau effect at 1.5L/kg/min. Perhaps the correct dose is where a potential of positive inspiratory pressure is met but the flow is not affecting the patient’s comfort levels. Unfortunately, objectively definied clinical improvement on 1.5-2L/kg/min has not been a consistently reproducible finding with some papers finding no difference in treatment failure, intubation rates or ICU admissions when randomizing children to either 1L/kg/min or 2L/kg/min. .

Weaning

Just as protocolization became standard for HFNC initiation, so too did hospitals implement standardized weaning protocols in the hopes of mitigating prolonged HFNC duration. One study found that of the reported hospitals with protocols, fifty-seven percent of them reported guidance on weaning of HFNC. Of those institutions, 65% of them weaned from HFNC to standard nasal cannula and 22% of sites weaned from HFNC directly to room air. There is however still no agreement upon recommendations on when to wean HFNC. Should this be based on work of breathing, respiratory rate, FiO ₂ needs, day of illness or other clinical factors? Once ready to wean, is it required to do so slowly and if so what should be weaned first, the FiO ₂ or the flow? Some pathways weaned slowly with RTs being instructed to decrease flow rate by 1L every 2 h for patients who were clinically improving and requiring less than 30 % FiO ₂ . Once at 2L for 2 h then HFNC could be discontinued. Other institutions protocols performed rapid disctoninuation. Knowing most benefits of high flow come at ∼ 2L/kg/min flow rate it was felt not to be necessary to wean slowly, especially once the child is experiencing a subtherapeutic flow rate (<2L/kg/min). This rapid discontinuation can also lead to an overall reduction in mean LOS without any significant change in adverse outcomes, such as transfers to the ICU. There are now multiple studies that show rapid discontinuation of HFNC is safe and effective. .