Pages Navigation Menu

Meet 'em, greet 'em, treat 'em and street 'em

SGEM#228: Winds of Change – High Flow Nasal Oxygen for Acute Bronchiolitis?

SGEM#228: Winds of Change – High Flow Nasal Oxygen for Acute Bronchiolitis?

Podcast Link: SGEM228

Date: September 5th, 2018

Reference:  Franklin et al. A Randomized Trial of High-Flow Oxygen Therapy in Infants with Bronchiolitis. NEJM March 2018.

Guest Skeptic: Dr. Ben Lawton is a paediatric emergency physician in Brisbane Australia. He divides his time between a tertiary children’s hospital and a community hospital that is busy enough to have its own paediatric emergency department. He is part of the Don’t Forget the Bubbles team.

Case: Elsie is five months old and presents on day two of a bronchiolitic illness. She has taken just under half of her usual feeds so far today and has a respiratory rate of 58 breaths per minute and oxygen saturation of 90% on room air with moderate work of breathing. She is not clinically dehydrated and has a temp of 38.2C with clear rhinorrhea, red ears, a red throat and equal air entry with widespread crackles and wheeze. She was born at term, is immunised and has no significant medical history. Her parents Dave and Tony have driven 20 minutes from home to bring her to the regional hospital where you work. Your hospital has an inpatient paediatric ward but is a 90-minute drive from the nearest children’s hospital with PICU facilities.

Background: We have covered bronchiolitis before on the SGEM#167 with expert Dr. Amy Plint. That episode looked at how bronchiolitis was managed in community hospitals. The bottom line was that there seemed to be a knowledge gap when it comes to managing bronchiolitis in the community setting (previous evidence as suggested a knowledge gap also exists in the academic pediatric hospitals).

Although the vast majority of infants with bronchiolitis can be managed with supportive care at home, due to its high incidence, it is the number one reason for infants to be hospitalized (Njoo et al 2001, Langley et al 2003, Craig et al 2007 and Shay et al 1999).

Since bronchiolitis is a clinical diagnosis, there is no test, including viral testing and radiography, which rules it in or out (Schuh et al 2007). Sadly, despite multiple guidelines (NICEAAPCPS), there has also been no “magic bullet” in terms of treatment.

Hypertonic saline has been tried for acute bronchiolitis. A systematic review of this treatment modality was covered on SGEM#157. The bottom line at that time was that the data did not support the routine use of hypertonic saline for mild to moderate acute bronchiolitis.

The American Academy of Pediatrics guideline says that oxygen therapy in infants with saturation of 90% or greater may not be needed (Ralston et al Pediatrics 2014)

  • “Clinicians may choose not to administer supplemental oxygen if the oxyhemoglobin saturation exceeds 90% in infants and children with a diagnosis of bronchiolitis (Evidence Quality: D; Recommendation Strength: Weak Recommendation [based on low level evidence and reasoning from first principles]).”

Clinical Question: Will Elsie benefit from the use of heated humidified high-flow oxygen via nasal prongs in preference to low flow oxygen via standard nasal cannula?


Reference: Franklin et al. A Randomized Trial of High-Flow Oxygen Therapy in Infants with Bronchiolitis. NEJM March 2018.

  • Population: Infants less than 12 months of age with bronchiolitis and needing supplemental oxygen based on their institutional practice.
    • Bronchiolitis was defined using the American Academy of Pediatric criteria as symptoms of respiratory distress associated with symptoms of a viral respiratory tract infection.
    • Exclusions: Critically ill infants who had an immediate need for respiratory support and ICU admission; infants with cyanotic heart disease, basal skull fracture, upper airway obstruction, or craniofacial malformation; and infants who were receiving oxygen therapy at home.
  • Intervention: Heated humidified high-flow oxygen via nasal prongs (HFOT) at 2L/Kg/min with FiO2 titrated to maintain oxygen saturation of 92%-98% or 94%-98% depending on the institutional practice
  • Comparison: Oxygen via standard nasal low flow nasal prongs (LFOT) titrated to maintain oxygen saturation of 92%-98% or 94-98% depending on institutional practice with max of 2L/min
  • Outcome:
    • Primary Outcome: Treatment failure resulting in escalation of care during hospital admission. Escalation of care was defined as needing increased respiratory support or transfer to an ICU. Treatment failure defined by meeting three of four clinical criteria. Clinicians were permitted to escalate therapy if they were worried for other clinical reasons not captured in the four clinical criteria:
      1. Heart rate unchanged or increased since admission
      2. Respiratory rate unchanged or increased since admission
      3. FiO2 requirement exceeding 0.4 on high-flow or 2L/min on low-flow to maintain oxygen saturation in target range
      4. Hospital Early-Warning Tool triggered
    • Secondary Outcome: The proportion of infants transferred to an ICU, duration of hospital stays, duration of ICU stays, duration of oxygen therapy, intubation rates, and adverse events (any event that was fatal, life threatening, permanently disabling or resulted in a prolonged hospital stay).

Authors’ Conclusions: Among infants with bronchiolitis treated outside the ICU, those who received high-flow oxygen therapy had significantly lower rates of escalation of care due to treatment failure than those in the group that received standard oxygen therapy.”

 Quality Checklist for Randomized Clinical Trials:

  1. The study population included or focused on those in the emergency department. Yes
  2. The patients were adequately randomized. Yes
  3. The randomization process was concealed. Yes
  4. The patients were analyzed in the groups to which they were randomized. Yes
  5. The study patients were recruited consecutively (i.e. no selection bias). Yes
  6. The patients in both groups were similar with respect to prognostic factors. Yes
  7. All participants (patients, clinicians, outcome assessors) were unaware of group allocation. No
  8. All groups were treated equally except for the intervention. Yes
  9. Follow-up was complete (i.e. at least 80% for both groups). Yes
  10. All patient-important outcomes were considered. Unsure
  11. The treatment effect was large enough and precise enough to be clinically significant. Unsure

Key Results: They enrolled 1,472 infants with acute bronchiolitis with 733 assigned to receive standard oxygen therapy with nasal prongs (LFOT) and 739 assigned to receive high-flow oxygen therapy (FHOT). The mean age was six months and just over 1/3 were female.


Treatment failure was 12% in high-flow oxygen therapy group vs. 23% in the low-flow oxygen therapy comparison group.


  • Primary Outcome: Treatment failure resulting in escalation of care
    • HFOT 87/739 (12%) vs. LFOT 167/733 (23%)
    • Risk difference -11% (95% CI -15% to -7%, P<0.001)
    • NNT of 9 to prevent 1 episode of escalation of care

  • Secondary Outcomes: 
    • There were no significant differences in duration of hospital stay, duration of ICU stays or duration of oxygen therapy.
    • HFOT was used as a rescue therapy in all 167 infants in the standard care group who required escalation of care, 61% of these infants responded to HFOT as a rescue therapy.
    • Twelve infants required intubation (eight from HFOT group and four from the LFOT group) so HFOT did not prevent intubation within the study cohort.
    • There was no difference in adverse event rates with one pneumothorax in each group.

Thank you to our FOAMed friend in Australia, Jesse Spurr (Injectable Orange). He happened to be hanging out with the lead author on this paper and got us connected. We reached out to Dr. Donna Franklin and she was kind enough to include written responses to our nerdy questions.

1) Objective vs. Subjective Clinical Criteria: There were five clinical criteria to trigger an escalation of care. You needed three or more criteria to trigger the escalation. Three of the criteria were very objective (heart rate, respiratory rate and oxygen requirements). One was a fuzzy criterion, because different hospitals use different pediatric early warning (PEW) tools. These PEW tools could be triggered by a single or multi trigger meaning a rise in a single vital sign (e.g HR alone) would trigger escalation of care but not others thereby ticking two of the four boxes. The fifth clinical criterion was subjective, if the clinicians’ spidey senses were tingling. I understand why they included clinical gestalt, but it would have made the study easier to interpret if they had just used objective criteria or been pragmatic and said if the clinician is worried that is enough to escalate care.

Agree that this was always going to be difficult in the subjective vs objective. Each of these early warning tools was comparable in scoring therefore when a child reached a certain HR, RR, SpO2 etc composite score and triggered a medical review, this occurred at the same level for all participating institutions.

What was interesting was that those patients who triggered the subjective 5th criteria we went back and checked all physiological criteria on these patients and they actually met the 3 of 4 criteria at that time of escalation. Thereby telling us that the Criteria 5 being the clinician’s ‘gut feel’ is most of the time ‘spot on’.

Agree it would have been best to use objective criteria however this was not always available if poor documentation occurred and we would find that the medical notes provided more information and had to use what was available.

2) High-Flow Oxygen Therapy: There has been some debate about what constitutes “high-flow” with some trials using 1L/kg or other flow rates. PARIS provides a dataset around a way of defining and using high-flow which is as defensible as any standard in this area. i.e. it is now a strong argument that we should accept “standard high-flow”as meaning flow of 2L/Kg/min with titrated FiO2 from 0.21 up, we know the adverse event rates do not increase up to an FiO2 of 0.4 so can argue that children being managed within these parameters should be cared for on standard wards with standard nursing ratios unless there are other contextual factors which make this problematic.

Agree with this – we are now performing a dose finding trial on the sicker cohort in PICU to see what the ‘sweet spot’ is for these children with flows delivered. However, this would be the more unwell Intensive Care patient rather than the PARIS patient who was mild to moderate in severity when randomised.

3) Composite Outcome: The primary outcome of escalation of care was a composite outcome. It included either requiring more respiratory support or transfer to an ICU. For the standard oxygen therapy group, it meant being moved to HFOT. When you looked at the individual outcomes there was no difference in transfer to an ICU. Therefore, the difference in composite outcome was driven by moving LFOT patients to HFOT.

This one was difficult as hospitals were obviously already used to using high flow in the wards.  We recognised that the study would have been better designed for both groups to immediately transfer to ICU if a child failed either treatment arm however this was not practical as all centres know how to care for high flow in the ward and it was not sensible to go directly to ICU from standard oxygen therapy. However, we recognised this was a flaw in the design but unfortunately not able to alter.  For both common sense reasons and health care costs associated with ICU and transfers etc.

4) Escalation of Care: Escalation of care was twice as likely to occur (7% vs 14%) in hospitals without an onsite PICU vs hospitals with a PICU. Not using a PICU bed has significant cost savings and is nicer for the family not to be in PICU but arguably not as disruptive for the family than being transferred to another hospital – though a study on family experience of care would be good and I think is coming. Transfers are both expensive and disruptive to families so on a system/economic level HFOT may be disproportionately useful in hospitals without PICU.

Yes, use of high flow in hospitals without onsite ICU is useful.  They regional centres outperformed the tertiary centres and kept these children in their centre.  The children who required escalating and transferred actually did transfer and they met 3 of the 4 criteria and were the more unwell cohort.

A HF paper has been submitted this week to the American Journal of Respiratory and Critical Care Medicine.

We will also be pulling together a paper looking at the regional vs tertiary centres in the near future.

5) Secondary Outcomes: None of the secondary outcomes were statistically different (transferred to an ICU, duration of hospital stays, duration of ICU stays, duration of oxygen therapy and intubation rates). These seem very patient oriented compared to escalation of care.

6) Intubation Rate: The lack of difference in intubation rates was odd on the face of it given that anecdotally we are intubating noticeably fewer kids with bronchiolitis than we did before the widespread use of HFOT. A difference may have been lost in the fact that all of the patients in the standard group who required escalation were put on HFOT so it is possible that without HFOT as a rescue therapy the intubation rate would have been much higher but there is no way to know this.

The ANZPIC database shows we still have up to 10% intubation rate for bronchiolitis, whereas PARIS showed a 0.8% intubation rater (12/1472) infants.  Our unit we work in is around 3% but we have used high flow for >12years in PICU.  The paper was alluding to the much lower than what was current in other centres around Australia and NZ specifically where it is higher.  Schlapbach et al showed in his ERJ paper ‘Burden of disease and change in practice in critically ill infants with bronchiolitis’ a ‘risk to be intubated’ graph whereby there are some centres intubating up to 35% of these infants which is quite shocking to see in 2018.

7) Lack of Blinding: The two different oxygen supplementation setups were different and obvious. This means the study was unblinded. It could have introduced some bias. Especially given the subjective criteria that was part of the escalation of care. It is unclear which direction this potential bias due to lack of blinding would have influenced the results.

Agree that it could introduce some bias. All our centres were familiar with high flow however not all clinicians were believers. A difficult component of the design but could not change as you know what you are randomised to and it is therefore seen and managed as standard oxygen or high flow.

8) External Validity: Within the parameters defined by the trial (flow of 2L/kg/min with FiO2 <0.4) there was no increase in adverse events with HFOT with a nursing ratio of 1:4. Nursing ratios of 1:2 (until recently required in many places) would prevent use of HFOT in many contexts as it would block too many beds leading to pressure for the child to be transferred anyway but these nursing ratios seem unnecessary based on PARIS. Important to note that PARIS came with paediatric nurse educator support provided by the trial team at no cost to the hospital. Many non-tertiary facilities do not have access to specialised paediatric nurse educator support and it would be important to observe for any increase in adverse events due to differences in nursing education capacity as HFOT is implemented outside of the trial.

As with any study protocol in place the performance of care and management was somewhat improved during the course of the study as we were onsite at tertiary centres every day and visiting every1-2 weeks after initial set up at all regional centres for the course of the study.  What was interesting was that prior to the study taking place all infants <12mths with bronchiolitis mild to moderate (PARIS patients description) would immediately go to resus bay.  Within a few months of PARIS commencing and education in place the staff were more comfortable with these infants and they were cared for in the acute area and then randomised if meeting the inclusion criteria.  We are performing a follow up study next year on translation of knowledge and using the Nurse Educators who work in these departments already as the people who will educate. We will screen these patients admitted and see how they are cared for and managed etc.  Will be interesting to see – this will occur in six regional centres.

9) Oxygen Saturation: This study had different oxygen saturation targets (92%-98% or 94%-98% depending on the institutional practice). Periods of hypoxia has been documented in health infants with saturation as low as 84% to 86% (Hunt et al J Peds 1999 and American Thoracic Society Am J Respir Crit Care Med 1999). Do these targets matter or should we allow lower oxygen saturations before providing supplemental oxygen?

The AAP gives a weak recommendation based on low-level evidence that supplemental oxygen may not be administered if the level is >90%. The Bronchiolitis of Infancy Discharge Study (BIDS) demonstrated equivalence in their primary outcome (time to resolution of cough) between a minimum target oxygen saturation of at least 90% vs. >94%. Those in the modified group (minimum target of at least 90%) were fit for discharge sooner, off oxygen one day sooner and were actually discharged 10 hours earlier than the standard group.

Schuh et al published a RTC in JAMA 2014 on hospitalization of infants with bronchiolitis. The took patients with mild to moderate bronchiolitis and true oxygen saturations of >87%. They randomized them to have a true saturation reading or altered to read 3% higher than the true reading. The primary outcome was hospitalization within three days. Those with the artificially elevated pulse oximetry were less likely to be admitted to hospital within three days. This finding suggests the other factors besides oxygen saturation should be used in determining the management of infants with bronchiolitis.

There was also the Principi et al (JAMA 2016) that looked at desaturation of infants with bronchiolitis discharged home. It showed a majority of infants had hypoxic episodes (oxygen saturation of <90% for at least one minute). Of the infants who had desaturation events, the vast majority had oxygen levels <80% and more than a third <70% for at least one minute. However, there was no difference in unscheduled return medical visits and delayed hospitalizations between infants that had desaturations and those who did not.

Do these oxygen values really matter? What oxygen level should we use? Certainly, oxygen saturation should not be the only factor in making management decisions for patients with acute bronchiolitis.

We were governed by what was current standard practice in each facility with the thresholds.  Agree that this is a difficult number to state which is correct.  Then you also need to factor in altitude in various parts of the world too.

10) Failure? The primary outcome as mentioned was escalation of care or transfer to an ICU (failure of treatment). There was no difference in transfers to the ICU in this study.

It would be totally reasonable based on the results of this trial to adopt a policy of putting infants on low flow oxygen initially with escalation to HFOT as the first thing to do once they have triggered those pre-determined clinical criteria, how you attach the low flow prongs to the infant is worth thinking about to avoid the need to pull tape off the child’s faces in order to change the prongs should they need an escalation to HFOT.

To drill down on those numbers just a little more in order to make our point. Of the 167 patients who failed standard care with LFOT, 167 (100%) were put on HFOT. Of those, 102/167 (61%) improved representing 14% of the overall patients in the standard group. This means 65/167 (39%) did not improve and were transferred to the ICU. Therefore, 65/733 (9%) of patients that started on NP, were escalated to HFOT and ultimately ended up in the ICU. Remember that 12% (87/739) of the patients randomized to the HFOT group were admitted to the ICU. Now consider that the NNT of 9. The interpretation could be that 9 infants would need to be put on HFOT initially to prevent 1 from being put on HFOT later with no difference in ICU admissions. Does this actually represent a failure? How much equipment, time and staff would be needed for this intervention? Would it not be more practical just to put everyone on LFOT and only transition those to HFOT if needed?

The difficulty and acknowledgement in the design occurs here again, allowing escalation from standard oxygen to high flow in the ward and thus the patient could be managed in the ward and family and patient not transferred to ICU.  The cleaner design would have been if failing either arm then both go to ICU.  From a HF point of view this will soon be available for comment when published.

It can be definitely viewed as your ‘bottom line’ comment and indeed it has been interpreted this way in some of our institutions who participated.  They continue to use HF as a rescue option only in the wards.  This is some but not all hospitals who participated who now do this.

Comment on Authors’ Conclusion Compared to SGEM Conclusion: The authors conclusion that HFOT reduces need for escalation of care is reasonable and the parameters they have chosen as denoting a need for escalation in the trial reflect parameters around which clinical management and resource use would commonly be affected in clinical contexts where early warning tools are used (i.e would have triggered a Medical Emergency Team [MET] call). Within the system that I work this would reduce resource use, it is not clear that the escalation that this trial avoids actually benefits these patients i.e we may have just drawn an artificial line and then found a therapy that helps us stay the right side of that line forgetting that the line had no real meaning in the first place.  The Authors have neither shown nor claimed that HFOT makes bronchiolitis better any quicker.


SGEM Bottom Line: High-flow oxygen therapy is not required for every child in hospital with bronchiolitis. It will continue to have a role in supporting those with more severe disease, but the potential benefits and harms will need to be considered within the context of where it is being used.


Case ResolutionElsie was commenced on low flow oxygen via nasal prongs at 1L/min, the flow was increased to 1.5L/min on day four and weaned over day five and she was discharged on day six. She started daycare three weeks later and currently has a runny nose, Dave and Tony are keeping a careful eye on her work of breathing.

Dr. Ben Lawton

Clinical Application: We have not found a magic bullet for bronchiolitis. HFOT oxygen for treating bronchiolitis in infants is associated with no increase in adverse events and is probably useful in enabling moderately unwell children with bronchiolitis be cared for outside of PICU in most developed healthcare systems where PEW scores are used. It does not make any difference to rate of recovery. There may well benefits in terms of cost and family experience in some contexts, but these have not been adequately explored yet.

What Do I Tell My Patient? I would tell Dave and Tony that Elsie needs some more help with her breathing. We will start with some oxygen through some nasal prongs, if she gets worse we can try her on high flow.

Keener Kontest: Last weeks’ winner was Dr. Steven Stelts an emergency specialist from Auckland, New Zealand. He knew it was Dr. Perry Cox from Scrubs who once said to his junior colleagues “Are you a real doctor, or a doctor like Dr. Pepper is a doctor?”

Listen to the SGEM podcast on iTunes to hear this the new keener question. If you know the answer send an email to TheSGEM@gmail.com with “keener” in the subject line. The first correct answer will receive a cool skeptical prize.

Other FOAMed:


Remember to be skeptical of anything you learn, even if you heard it on the Skeptics’ Guide to Emergency Medicine.