Randomized Controlled Trial of High-Flow Nasal Cannula in Preterm Infants After Extubation

April 07, 2021


Uchiyama A, Okazaki K, Kondo M, Oka S, Motojima Y, Namba F, Nagano N, Yoshikawa K, Kayama K, Kobayashi A, Soeno Y, Numata O, Suenaga H, Imai K, Maruyama H, Fujinaga H, Furuya H, Ito Y, Non-Invasive Procedure for Premature Neonates (NIPPN) Study Group. Randomized Controlled Trial of High-Flow Nasal Cannula in Preterm Infants After Extubation. Pediatrics 2020; 146(6):e20201101. PMID: 33214331.


A/Prof Brett J. Manley

Dr Kate A. Hodgson

Prof Peter G. Davis

All: The Royal Women’s Hospital, Parkville, Victoria, Australia




In Preterm infants born < 34 weeks’ gestation who required non-invasive ventilation after extubation from mechanical ventilation (P) High-flow nasal cannula (HFNC) (I) compared nasal continuous positive airway pressure (NCPAP) and/or nasal intermittent positive pressure ventilation (NIPPV) (C) Treatment failure (O) within 7 days (T)


  • Design: Randomised controlled trial
  • Allocation:Infants were centrally randomly assigned by a computer-generated randomization sequence with 10 block sizes to either an HFNC group or NCPAP/ NIPPV group. Allocation was stratified by study site. In the NCPAP/NIPPV group the choice of either support was at the clinician’s discretion.
  • Blinding: Not blinded
  • Follow-up period:
    • Primary outcome: 7 days after extubation from mechanical ventilation
    • Study: Death before discharge from hospital, or hospital discharge
  • Setting: Six neonatal intensive care units in Japan
  • Patients: Preterm infants born <34 weeks’ gestation who required non-invasive ventilation after extubation from mechanical ventilation. The criteria for non-invasive ventilation requirement were not stated.
  • Interventions:
    • HFNC: Optiflow Junior (Fisher &Paykel Healthcare, Co Ltd, Irvine, CA) device. Starting flow rate was >2 L/min and adjusted according to the patient’s respiratory condition up to 8 L/min. The fraction of inspired oxygen (FIO2) was adjusted to maintain oxygen saturation (SpO2) levels between 92% and 95% (>92% if in room air). Weaning occurred when the infant’s respiratory condition was stable at a flow rate of 2 L/min and FiO2 <0.3 for 24 hours.
    • NCPAP/NIPPV: Several devices used: Infant Flow SiPAP (CareFusion, San Diego, CA); medinSindi (Medin Medical Innovations GmbH, Olching, Germany); medinCNO (Medin Medical Innovations GmbH); Babylog 8000 Plus (Dräger Medical AG & Co, Lübeck, Germany); Babylog VN500 (Dräger Medical AG & Co); and Bear Cub 750 (VIASYS Healthcare, Conshohocken, PA). NCPAP pressure of 4 to 5 cm H2O. The FIO2 with NCPAP therapy was determined in the same way as HFNC therapy. NIPPV therapy was performed to maintain a positive end expiratory pressure (PEEP) of 4 to 5 cm H2O and peak inspiratory pressure of 3 to 4 cm H2O above PEEP.
  • Outcomes:
    • Primary outcome: Treatment failure within 7 days of extubation as defined by any of (1) supplemental oxygen requirement of 40%, (2) sustained blood gas measurements of pH <7.20 and pCO2 >60 mm Hg, (3) episodes of apnea requiring 1 bag mask ventilation within 24 hours or episodes of apnea requiring 1 stimulation within 1 hour over a 6 hour period, or (4) urgent reintubation as determined by the attending physician.
    • Secondary outcomes: reintubation within 7 days (168 hours), nasal skin or mucosal injury, chronic lung disease at 36 weeks’ corrected gestational age, death before hospital discharge, results of blood gas analysis such as pH and pCO2, and base excess, modified COMFORT scale between 60 and 90 minutes after extubation, variables predicting HFNC treatment failure.
  • Analysis and Sample Size: Trial designed to determine if HFNC is noninferior to NCPAP/NIPPV in preventing treatment failure after extubation. The investigators assumed a 15% intubation rate in the NCPAP/NIPPV group, an expected treatment failure rate of 24% in the HFNC group, and chose a 20% non-inferiority margin. The calculated sample size was a total of 310 infants (alpha 0.1; power 0.8). The sample size was then inflated for anticipated dropouts from the study and a target sample size of 340 infants was chosen.
  • Patient follow-up:
    • HFNC group: 177 infants randomised, 176 (99.4%) included in the primary analysis.
    • CPAP/NIPPV group: 201 randomised, 196 (97.5%) included in the primary analysis.


A total of 378 infants were randomised, 177 to HFNC and 201 to NCPAP/NIPPV. One infant was excluded post-randomisation from the HFNC group and five infants from the NCPAP/NIPPV group. Ultimately 176 infants and 196 infants were included in the HFNC and NCPAP/NIPPV groups respectively. Demographics between groups were similar, with a mean gestational age at birth of 28 weeks’ and birth weight of approximately 1.1 kg.

Treatment failure within 7 days after extubation occurred in 54/176 infants (31%) in the HFNC group and 31/196 (16%) in the NCPAP/NIPPV group: risk difference (95% CI) 14.9% (6.2% to 23.2%). Among 54 infants with treatment failure in the HFNC group, 16 and 28 infants were successfully treated with NCPAP and NIPPV, respectively, without reintubation. Among 31 infants with treatment failure in the NCPAP/NIPPV group, 14 infants were successfully treated with NIPPV. Treatment failure within 72 hours after extubation occurred in 45/176 (26%) infants in the HFNC group and 30/196 (15%) infants in the NCPAP/ NIPPV group: risk difference (95% CI) 10.2% (1.9% to 18.3%). Among those infants who had treatment failure within 7 days, treatment failure occurred within 72 hours after extubation in 83% (45 of 54) of the HFNC infants, and 97% (30 of 31) of the NCPAP/NIPPV infants. In extremely preterm infants born <28 weeks’ gestation (n=168 infants), HFNC resulted in a higher rate of treatment failure than NCPAP: risk difference (95% CI) 21.0% (7.1 to 34.1%, p=0.003) (Supplemental Table 5, online).

The rates of treatment failure, reintubation within 7 days after extubation, death before discharge, CLD at 36 weeks’, and nasal trauma were similar between groups. The pH, pCO2 between 60 and 90 minutes after the start of treatment, and base excess values did not show significant differences between the two groups. Modified COMFORT scale values between 60 and 90 minutes after the start of treatment were similar between the groups.

A multivariate logistic regression analysis was performed to examine clinical predictors of HFNC treatment failure: histologic chorioamnionitis (adjusted odds ratio [aOR], 2.92; 95% CI, 1.17 to 7.31; P=0.02), treated patent ductus arteriosus (aOR, 3.61; 95% CI, 1.62 to 8.07; P=0.002), and corrected gestational age at the start of treatment (aOR, 0.76; 95% CI, 0.61 to 0.94; P=0.008) were independently associated with HFNC treatment failure.


HFNC revealed a significantly higher rate of treatment failure than NCPAP or NIPPV after extubation in preterm infants. The independent factors associated with treatment failure with HFNC use were histologic chorioamnionitis, treated patent ductus arteriosus, and a younger corrected gestational age at the start of treatment.


HFNC therapy is increasingly used as ‘non-invasive’ respiratory support for preterm infants. A 2016 Cochrane Review found that HFNC had similar efficacy to NCPAP for preterm infants following extubation (1), but that further studies were needed, particularly in extremely preterm infants born <28 weeks’ gestation, who are at highest risk of extubation failure.

The trial by Uchiyama et al. is the largest randomised trial of HFNC as post-extubation support in preterm infants, and one of only a few to enrol extremely preterm infants (168 infants enrolled; 76 HFNC, 92 NCPAP/NIPPV). This multicentre non-inferiority trial found NCPAP/NIPPV was superior to HFNC (and HFNC did not satisfy non-inferiority criteria) for the primary outcome of treatment failure within 7 days. Reintubation rates were similar; more than 80% of infants who met HFNC treatment failure criteria avoided reintubation due to crossover to NCPAP/NIPPV.

Like previous similar trials (2-4) the intervention was not blinded, although objective treatment failure criteria were used. In the NCPAP/NIPPV group, 149 infants received NCPAP and 47 received NIPPV at extubation. In both groups, the escalation of respiratory support was at clinician discretion. As the control group could receive either NCPAP or NIPPV, the definition of “treatment failure” is blurred: some infants received NIPPV before “treatment failure” and others didn’t. It is not reported whether NIPPV inflations were synchronised with the infant’s breathing or not, which may be important (5).

A non-inferiority trial design is reasonable given the existing evidence, although the sample size calculation assumed a HFNC treatment failure rate 9% higher than that of NCPAP/NIPPV, and the chosen non-inferiority margin was rather large. There is an imbalance in the allocation of infants to HFNC (47%) and NCPAP/NIPPV (53%). A total of 378 infants were ultimately recruited to the study (38 more than required, even with inflation for anticipated dropouts), however this represents only 33% of eligible infants which has implications for generalisability of the results. The use of respiratory stimulants prior to extubation is not reported. The trial was prospectively registered, however there are inconsistencies between the registration and the published trial in the stated sample size, comparator and primary outcome.

The starting HFNC gas flow was >2 L/min, lower than some previous studies (3, 4), and it is unclear whether infants received the maximal permitted gas flow of 8 L/min prior to treatment failure being determined. As increasing gas flow is associated with increased delivered pressure (6), this may have contributed to decreased efficacy in the HFNC group.

Extubation failure in preterm infants is an important concern for neonatal clinicians, and the optimal mode of post-extubation respiratory support remains uncertain. The results of the trial by Uchiyama et al. may tip the balance in favour of NCPAP/NIPPV as initial post-extubation support however, consistent with previous evidence, reintubation rates are similar when HFNC is used with the ‘backup’ of NCPAP/NIPPV. For the most immature infants at highest risk of extubation failure it seems prudent to avoid using HFNC as initial post-extubation support.


  1. Wilkinson D, Andersen C, O’Donnell CP, De Paoli AG, Manley BJ. High flow nasal cannula for respiratory support in preterm infants. Cochrane Database Syst Rev. 2016;2:CD006405. doi: 10.1002/14651858.CD006405.pub3. PubMed PMID: 26899543.
  2. Kadivar MM, Mosayebi ZM, Razi NM, Nariman SM, Sangsari RM. High Flow Nasal Cannulae versus Nasal CPAP in Neonates with Respiratory Distress Syndrome Managed with INSURE Method: A Randomized Clinical Trial. Iran J Med Sci. 2016;41(6):494-500. Epub 2016/11/18. PubMed PMID: 27853329; PubMed Central PMCID: PMCPMC5106564.
  3. Manley BJ, Owen LS, Doyle LW, Andersen CC, Cartwright DW, Pritchard MA, et al. High-flow nasal cannulae in very preterm infants after extubation. NEJM. 2013;369(15):1425-33. doi: 10.1056/NEJMoa1300071. PubMed PMID: 24106935.
  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;84(4):262-6. Epub 2017/01/06. doi: 10.1007/s12098-016-2280-2. PubMed PMID: 28054235.
  5. Lemyre B, Davis PG, De Paoli AG, Kirpalani H. Nasal intermittent positive pressure ventilation (NIPPV) versus nasal continuous positive airway pressure (NCPAP) for preterm neonates after extubation. Cochrane Database Syst Rev. 2014(9):Cd003212. Epub 2014/09/05. doi: 10.1002/14651858.CD003212.pub2. PubMed PMID: 25188554.
  6. Wilkinson DJ, Andersen CC, Smith K, Holberton J. Pharyngeal pressure with high-flow nasal cannulae in premature infants. J Perinatol. 2008;28(1):42-7. Epub 2007/11/09. doi: 10.1038/sj.jp.7211879. PubMed PMID: 17989697.

Leave a comment