(Last Updated On: January 8, 2019)

A Randomized Trial of High-Flow Oxygen Therapy in Infants with Bronchiolitis

Franklin, Donna, et al. The New England Journal of Medicine, March 2018. [paper]

Why I Chose this Study

During my recent PICU and Pediatric ED rotation, a large amount of my patients with bronchiolitis were started on HFNC liberally. Other hospitals that we rotate in often just start standard nasal cannula in non-critically ill hypoxemic bronchiolitic infants. I was interested to see if there was any difference in outcomes between these different treatment options.


Bronchiolitis is the most common reason for non-elective hospital admissions in infants. The respiratory support that has historically been provided has been via standard nasal cannula. The pathophysiology of bronchiolitis causes small airway inflammation. This inflammation can increase work of breathing and cause hypercarbia, which can respond to positive pressure. This positive pressure has typically been administered in the ICU setting. High-flow nasal cannula (HFNC) is heated humidified air with oxygen via nasal cannula that can provide a small amount of positive pressure that does not always require ICU admission.

Research Question

Does early treatment with HFNC in infants with bronchiolitis and hypoxemia versus standard oxygen therapy result in less treatment failure and decreased escalation of care?

Study Design

This was multicenter, randomized, controlled trial performed at 17 different tertiary and regional hospitals in New Zealand and Australia in the EDs and general pediatric inpatient units performed during 10/2013-08/2016. The study analyzed 1472 patients.

Inclusion Criteria

  • Infants younger than 12 months
  • Diagnosis of bronchiolitis
  • Need for supplemental oxygen therapy to keep SaO2 above predetermined levels (92-94%)

Bronchiolitis was defined in the article as symptoms of respiratory distress with an associated viral respiratory tract infection.

Exclusion Criteria

  • Critically ill infants admitted directly to ICU
  • Bronchiolitis that did not require oxygen therapy
  • Use of oxygen at home
  • Cyanotic heart disease
  • Basal skull fracture
  • Upper airway obstruction
  • Craniofacial malformations


Primary outcome: treatment failure which resulted in escalation of care during the admission. Escalation of care is defined as an increase in respiratory support or a transfer to an ICU. Treatment failure was based on having 3 of 4 positive criteria from the list below and/or clinicians thought an escalation of care was necessary.

  • Heart rate remained unchanged or increased from admission.
  • Respiratory rate remained unchanged or increased from admission.
  • Oxygen requirement in HFNC group exceeded an FiO2 of at least 0.4 to keep SaO2 above 92-94% depending on the hospital as some had different cut-offs for hypoxia.
  • Hospital internal early-warning tool triggered a review of case. Each hospital used an early warning to trigger escalation of care (scoring systems used as these hospitals that looks at HR, RR, SaO2, BP, temp, respiratory distress, capillary refill time, altered mental status).

Secondary Outcome: Duration of hospital/ICU stay, duration of oxygen therapy, intubation rates, adverse events, and transfers to the ICU were the secondary outcomes.


In HFNC group 12% of patients had escalation of care and in the standard therapy group 23% of patients who had their care escalated. The p-value for the primary outcome was <0.001 with a NNT of 9.

Several of the regional hospitals did not have an on-site ICU and the treatment effect of the intervention differed between hospitals with an ICU and hospitals without on-site ICU (P<0.001)

  • Escalation of care in hospitals without on-site ICU: 7% (HFNC group) vs 28% (standard group)
  • Escalation of care in hospitals with on-site ICU: 14% (HFNC group) vs 20% (standard group

There was no statistical difference between lengths of hospital/ICU stay, duration of oxygen therapy, or rates of ICU admissions. Also, no difference in outcomes between RSV negative and RSV positive infants.

A desperately unhappy mother cradling her sick child, Théophile Alexandre Steinlen 1902


There are several limitations to this study that likely affected the outcomes. First, the physicians could escalate the therapy regardless if the patient met the clinical criteria based on gestalt, and this happened in 34% of patients who did not meet escalation criteria. This probably led to bias as treatment failure was higher in hospitals in the study without on-site ICUs (28% vs 20% in standard group) as they may not have been as comfortable managing these patients.

I felt the study could have been improved by having all hospitals with same Sa02 cut-offs and also using the same hospital internal early-warning tool as there were two different ones used among the 17 hospitals.

Also, when looking at the demographics of who was randomized with each group, there was a statistical difference between the respiratory rates of each group with the HFNC group having on average a higher respiratory rate. This could mean the HFNC was generally sicker.

This study did have some good strengths. There were a large number of patients, private companies were not involved in the design or analysis of the study (however they did donate the supplies), and the study was performed at 17 different hospitals.

I was more interested in their secondary outcomes rather than their primary outcome as these seemed more clinically relevant. No difference was seen in the secondary outcomes including no difference in the length of stay in the hospital and ICU, duration of oxygen therapy, intubation rates, ICU transfers, or even adverse events. The article really tries to persuade one that HFNC does decrease escalation of care while glossing over the fact that most of the clinically important measurements showed no difference.

That being said, the data did show that early initiation of HFNC in these patients did decrease the need for increasing respiratory support going forward. While the data is not as robust as I would like and there was no difference in secondary outcomes, there still is a difference in escalation of care. Because of this difference, for my hypoxemic bronchiolitis infants I will consider starting HFNC on them early over standard NC as there was no harm shown and possible benefit.

Justin Kosirog is an EM/IM Resident, class of 2020

EM/IM Sessions are reviewed in journal club style by the current attendings and residents, as well as alumni of the UIC IM/EM program prior to publication. This post was specifically reviewed by Sukhi Bains, MD, Assistant Professor of IM/EM. Elspeth Pearce, MD Editor.