Is High-Flow Nasal Cannula Noninferior to Nasal CPAP for the Initial Management of Preterm Infants?

April 04, 2017


Roberts CT, Owen LS, Manley BJ, Frᴓisland DH, Donath SM, Dalziel KM, Pritchard MA, Cartwright DW, Collins CL, Malhotra A, Davis PG, for the HIPSTER Trial Investigators. Nasal High-Flow Therapy for Primary Respiratory Support in Preterm Infants. N Engl J Med 2016; 375:1142-51. PMID 27653564.


Megan J. Kirkley, MD, MPH
Fellow in Neonatal-Perinatal Medicine
University of Colorado Department of Pediatrics

Sunah S. Hwang, MD, MPH/MSPH
Assistant Professor in Neonatal-Perinatal Medicine
University of Colorado Department of Pediatrics




Among infants 28 weeks 0 days to 36 weeks 6 days gestational age with early RDS, is nasal high-flow therapy as primary respiratory support non-inferior to nasal CPAP in terms of treatment failure within 72 hours?


  • Design: Multicenter, randomized, non-inferiority trial
  • Allocation: Concealed
  • Blinding: Unblinded
  • Follow-up period: Death or NICU discharge
  • Setting: 9 NICUs in Australia and Norway
  • Patients:
    • Included:
      • GA 28 weeks 0 days to 36 weeks 6 days
      • Less than 24 hours old
      • Not previously received endotracheal intubation or surfactant
      • Clinician had decided to continue noninvasive respiratory support.
    • Excluded:
      • Urgent need for intubation and ventilation as determined by clinician
      • Had already met criteria for treatment failure (see below)
      • Known major congenital abnormality or pneumothorax
      • Had already received ≥ 4 hours of CPAP support
  • Intervention: Nasal High-Flow cannula, 6-8 LPM
    • Standard Arm: CPAP 6-8 cmH2O via nasal prongs or mask
  • Outcomes: The primary outcome was treatment failure within 72h, defined as meeting one or more of the following criteria:
    • FiO2 ≥4
    • pH ≤7.2 and PaCO2 > 60 in an arterial or capillary sample obtained ≥ 1 hour after starting treatment
    • ≥2 episodes of apnea requiring PPV within 24 hours or ≥6 episodes requiring any intervention within 6 hours
    • Urgent need for intubation and mechanical ventilation as determined by the treating clinician.
    • Secondary outcomes were diverse and included:
      • The individual reason(s) for treatment failure
      • Respiratory complications and outcomes: need for mechanical ventilation at any time during admission, incidence of BPD, incidence of pneumothorax or air leak, nasal trauma, total number of days and type of respiratory support, duration of need for supplemental oxygen, treatment with postnatal corticosteroids for lung disease.
      • Complications and outcomes of prematurity: (explicitly listed in the appendices) sepsis, meningitis, PDA, NEC stage 2 or above, intestinal perforation, grade III or IV IVH, cystic PVL, post-hemorrhagic hydrocephalus, ROP requiring laser intervention
      • Other measures of neonatal health (including duration of caffeine therapy, pain scores, time to full enteral feeds, mode of feeding at discharge, weight gain)
        • Other metrics including length of stay, incidence of transfer, parental stress, and cost of care.
  • Analysis and Sample Size: Based on a predicted treatment failure rate of 17% in the CPAP group, the authors prespecified a noninferiority margin for high-flow treatment of 10 percentage points (or within 27%.) For the study to have 90% power, a sample size of 750 was required. 278 infants were enrolled in the high-flow group, 286 infants in the CPAP group. The study was stopped by DSMB after data review for the first 515 infants demonstrated a highly significant difference favoring the CPAP group in the rate of treatment failure
  • Patient follow-up: Of 3349 screened infants, 1093 were eligible and 583 underwent randomization. Of the 510 mother-infant pairs who were not randomized, approximately 50% were never approached or available for consent, 25% declined participation, and the remaining 25% were not randomized due to language reasons, involvement in another study, or other factors. 19 of the 583 randomized infants were either randomized in error, had prospective consent withdrawn, or had retrospective consent declined by the parents. 100% of the 564 infants who were assigned treatment were followed to death or NICU discharge.


For the primary outcome, 13.3% of infants randomized to nCPAP had treatment failure by 72 hours, compared to 25.5% of infants randomized to HFNC. This risk difference of 12.3% (5.8-18.7) exceeded the prespecified non-inferiority margin of 10 percentage points, thus did not demonstrate noninferiority.


(n = 278)


(n = 286)

Risk Difference (95% CI) p-value
Primary Outcome Treatment Failure within 72 hours 71 (25.5%) 38 (13.3%) 12.3% (5.8 to 18.7) < 0.001
Secondary Outcomes Intubation within 72h 43 (15.5%) 33 (11.5) 3.9 (-1.7 to 9.6) 0.17
Duration of respiratory support* 4 days (2, 7) 3 days (2, 6) 1.0 (0.3 to 1.7) 0.005
Need for supplemental oxygen at 36 weeks 17 (12.1%) 17 (11.4%) 0.7 (-6.7 to 8.2) 0.85
Death before discharge 1 (0.4%) 1 (0.4%) 0.0 (-1.0 to 1.0) 0.98
Nasal trauma 23 (8.3%) 53 (18.5%) -10.3 (-15.8 to -4.7) <0.001
Pneumothorax or other air leak during assigned treatment 0 6 (2.1%) -2.1 (-3.8 to -0.4) 0.02
Pneumothorax or other air leak at any point 10 (3.6%) 8 (2.8%) 0.8 (-2.1 to 3.7) 0.59
Cost of hospital stay $29,785 $32,036 0.40

*Denoted as median (IQR) and difference in median (95% CI)

Results shown are from the primary intention-to-treat analysis, but were not different in the per-protocol analysis. In the gestational age subgroup analysis (≥32 and <32 weeks) the rates of treatment failure remained higher in the HFNC group regardless of gestational age, and the rates of intubation were not different between groups regardless of gestational age.


The authors conclude that high-flow cannula used as primary respiratory support in preterm infants >28 weeks GA resulted in a significantly higher rate of treatment failure. The study was stopped early, after 75% of the target sample had been recruited, due to this significant between-group difference.

Visit Acta to access a pdf copy of this EBNEO commentary!


The study question builds on the group’s earlier work, which demonstrated HFNC was noninferior to nCPAP as post-extubation support (1). Noninferiority trials are designed to show whether a new treatment is “not unacceptably less efficacious” than the standard of care (2). A new treatment that is nearly as effective may be acceptable to providers and patients if it is less invasive, less costly, or more convenient. (3) Of note, the authors defined 20% margin for noninferiority in their earlier study (1), a 10% difference in the rate of treatment failure between HFNC and nCPAP here may be too conservative.

In contrast to studies of HFNC as post-extubation support, (1, 4) Roberts et al found HFNC as primary support to be not only inferior to nCPAP, but the difference was obvious prior to full recruitment, and was significant in both the intention-to-treat and per-protocol analyses. The authors posit the higher rate of treatment failure in the HFNC group “may reflect its reduced effectiveness in infants with surfactant-deficient lungs.” Were this the case, the HFNC group should have had higher need for subsequent surfactant therapy than the CPAP group, but the rates of surfactant administration were equivalent. Treatment failure (in either group) may have occurred after the window during which surfactant is considered to be most efficacious (

Defining “treatment failure” by two objective and three somewhat subjective criteria complicates the interpretation of the results. Apnea in particular is subjective in terms of nursing documentation, and the inability to blind caregivers to mode of respiratory support may have impacted the perceived frequency and severity of apneic episodes. In practice, failure of noninvasive respiratory support typically means need for intubation and mechanical ventilation, but intubation rates during the treatment period were not significantly different between the HFNC and nCPAP groups. Approximately 30 infants on HFNC had treatment failure but stabilized on nCPAP; 40 infants failed HFNC and ultimately required intubation and surfactant. While this confirms something about nCPAP therapy was superior to HFNC, it seems unlikely to indicate such a substantial treatment effect that continued enrollment in the study would have been unsafe or unethical. Overall rates of intubation and pneumothorax were the same, length of stay was unchanged, and the rate of nasal breakdown Ihigher in the nCPAP group. It would be helpful to know if there was a clinically important increase in the frequency of pneumothorax and nasal breakdown specifically in the crossover group from HFNC to nCPAP. Because an increased cumulative oxygen exposure in the HFNC group is plausible, reporting the average FiO2 requirement for both groups also would be helpful.

Based on these findings, the standard use of HFNC as primary respiratory support in preterm infants who would otherwise receive CPAP cannot be recommended. However, the important complications of pneumothorax and intubation were ultimately not different between groups, nor did length of stay or cost of care differ significantly. This suggests HFNC may be an option for some infants without fear of significant adverse events.


1. Manley BJ, Owen LS, Doyle LW, Andersen CC, Cartwright DW, Pritchard MA, et al. High-flow nasal cannulae in very preterm infants after extubation. N Engl J Med. 2013;369(15):1425-33.
2. Hahn S. Understanding noninferiority trials. Korean J Pediatr. 2012;55(11):403-7.
3. Kaji AH, Lewis RJ. Noninferiority Trials: Is a New Treatment Almost as Effective as Another? JAMA. 2015;313(23):2371-2.
4. Soonsawad S, Swatesutipun B, Limrungsikul A, Nuntnarumit P. Heated Humidified High-Flow Nasal Cannula for Prevention of Extubation Failure in Preterm Infants. Indian J Pediatr. 2017. Epub ahead of print.