ADC-FNN Online First, published on December 23, 2016 as 10.

1136/archdischild-2016-311388
Original article

High flow nasal cannula versus NCPAP, duration

to full oral feeds in preterm infants: a randomised
controlled trial
Sinead J Glackin,1 Anne O’Sullivan,1 Sherly George,1,2 Jana Semberova,1,3
Jan Miletin1,2,3,4
1
Department of Neonatology, ABSTRACT
Coombe Women and Infants Objective To compare the time taken by preterm What is already known on this topic?
University Hospital, Dublin,
Ireland infants with evolving chronic lung disease to achieve full
2
UCD School of Medicine, oral feeding when supported with humidified high flow
▸ High flow nasal cannula therapy and nasal
University College Dublin, nasal cannula (HFNC) or nasal continuous positive
Dublin, Ireland continuous positive airway pressure (NCPAP)
airway pressure (NCPAP).
3
Institute for the Care of are safe and effective respiratory support
Design Single centre randomised controlled trial.
Mother and Child, Prague, options for preterm infants with respiratory
Czech Republic Setting Level III neonatal intensive care unit at the
distress syndrome.
4
3rd Faculty of Medicine, Coombe Women and Infants University Hospital, Dublin,
▸ High flow nasal cannula is less bulky than
Charles University, Prague, Ireland.
Czech Republic NCPAP.
Patients Very low birthweight (birth weight <1500 g)
▸ Oral feeding is difficult for preterm infants on
Correspondence to infants born before 30 weeks’ gestation who were
respiratory support.
Associate Clinical Professor NCPAP-dependent at 32 weeks corrected gestational age
Jan Miletin, Department of were eligible to participate.
Neonatology, Coombe Women Interventions Enrolled infants were randomised in a
and Infants University Hospital,
1:1 ratio to receive HFNC or NCPAP. Participants were
Cork Street, Dublin 8, Ireland; What this study adds?
[email protected] monitored daily until full oral feeding was established
and the baby was off respiratory support.
Received 7 June 2016 Main outcome measures Our primary outcome was ▸ There is no advantage of high flow nasal
Revised 24 September 2016 the number of days taken to establish full oral feeds cannula therapy compared with NCPAP in
Accepted 29 November 2016
(defined as oral intake ≥120 mL/kg/day) from the time establishing oral feeds in preterm infants.
of randomisation. We estimated that enrolling 44 ▸ It seems safe to feed orally stable preterm
subjects (22 in each group) would allow us demonstrate infants on NCPAP.
a 7-day difference in our primary outcome with 80%
power and α of 5%.
Results Forty-four infants were randomised (22 to
HFNC vs 22 to NCPAP). The mean time to achieve full while there was no difference in the failure rates of
oral feeding was not different between the groups HFNC and NCPAP, nearly half of those who failed
(HFNC 36.5 (±18.2) days vs NCPAP 34.1 (±11.2) days, HFNC were successfully managed on NCPAP
p=0.61). without the need for subsequent reintubation.2
Conclusions Preterm infants treated with HFNC did Although concerns have previously been
not achieve full oral feeding more quickly than infants expressed about the safety of HFNC due to the dif-
treated with NCPAP. ficulty in accurately measuring and controlling the
Trial registration number ISRCTN66716753. pressure generated, no differences in the rates of
pulmonary air leaks or death in preterm infants
were found.3 In the meta-analysis of five studies
involving 857 infants, HFNC was associated with a
INTRODUCTION significantly reduced risk of nasal trauma over
Heated humidified high flow nasal cannula NCPAP.1
(HFNC) is a widely used form of respiratory Several RCTs comparing HFNC and NCPAP
support in preterm infants with respiratory distress included time to attain full enteral feeding within
syndrome (RDS) as an alternative to other forms of the secondary outcomes and did not find any sig-
non-invasive ventilation (NIV). In a recently pub- nificant differences between the groups.4–6 Yoon
lished meta-analysis that included eight randomised et al7 reported in their retrospective study
controlled trials (RCTs) and 1112 preterm infants, decreased time to reach full enteral feeds and
HFNC was found to be as well tolerated as nasal regain birth weight in the HFNC group. A poten-
To cite: Glackin SJ, continuous positive airway pressure (NCPAP), the tial advantage of HFNC over NCPAP is that it may
O’Sullivan A, George S, most frequently used form of NIV in preterm facilitate oral feeding and kangaroo care. Oral
et al. Arch Dis Child Fetal
Neonatal Ed Published
infants.1 No significant differences between HFNC feeding is usually trialled in infants older than
Online First: [ please include and NCPAP were found in terms of efficacy as 32 weeks corrected gestational age (CGA), when
Day Month Year] primary respiratory or postextubation support for they have developed the coordination for sucking,
doi:10.1136/archdischild- preterm infants with RDS. However, one of the swallowing and breathing. Many clinicians wait
2016-311388 studies included in the meta-analysis showed that until after infants are off NCPAP support before
Glackin SJ, et al. Arch Dis Child Fetal Neonatal Ed 2016;0:F1–F4. doi:10.1136/archdischild-2016-311388 F1
Copyright Article author (or their employer) 2016. Produced by BMJ Publishing Group Ltd (& RCPCH) under licence.
 Original article

offering oral feeds.8 9 Time to achieve full oral feeds was investigations and management. If they were switched back to
reported as a secondary outcome in the RCT by Yoder et al.10 NCPAP, they would remain on it for 48 hours before being
They did not find any significant differences between HFNC changed back to HFNC. If this occurred on more than two
and NCPAP in infants born between 28 and 42 weeks of gesta- occasions, these infants would remain on NCPAP until the end
tion. Recently, Shetty et al8 reported quicker attainment of full of the trial. Infants were weaned off respiratory support at the
oral feeds using HFNC in preterm infants requiring non- discretion of the attending clinicians and there were no specific
invasive respiratory support at 34 weeks of gestation in their criteria mandated in the trial. Respiratory support was reinstated
case series. However, this clinical question has not been tested if required, using these criteria: (a) more than one self-correcting
in an RCT to date. We hypothesised that preterm infants with apnoeic episode per hour (defined as a bradycardia <100/min
evolving chronic lung disease (CLD) treated with HFNC would with oxygen saturations (SpO2) <88% lasting >20 s); (b) one
achieve full oral feeding earlier than preterm infants treated apnoeic episode requiring either moderate stimulation or bag
with NCPAP. and mask ventilation; (c) need for oxygen to maintain SpO2
>88%; (d) a score of 6–10 on the Silverman-Anderson
PATIENTS AND METHODS Respiratory Scale, indicating moderate-to-severe respiratory
We conducted an RCT at the Coombe Women and Infants distress.11
University Hospital, Dublin, Ireland. The study was approved by Oral feeds were offered in both groups at least once every
the hospital research ethics committee (reference number 18-2012) 72 hours and additional feeds were offered when infants
and registered with the International Standard Randomized demonstrated feeding cues (eg, sucking efficiently on a soother,
Controlled Trial Number register (ISRCTN66716753) before the waking at feeding times and settling post feeds). Information on
first patient was enrolled. every feed offered, including the type and adverse events (eg,
Very low birthweight (VLBW) infants born before 30 weeks’ desaturation events—falls in oxygen saturation >10% from
gestation who still received NCPAP respiratory support with baseline for >10 s; bradycardia—HR <100 bpm for >10 s),
<30% oxygen at 32 weeks CGA and who were on full enteral were recorded on proforma data sheets.
(≥120 mL/kg/day without supplemental intravenous nutrition), Our primary aim was to establish if there was a difference
orogastric tube feeding were eligible for enrolment. Infants who between infants on HFNC and infants on NCPAP in the dur-
were supported with a pressure <5 cm H2O with no supple- ation it would take them to become fully established on full oral
mental oxygen requirement (FiO2=0.21) were trialled off feeds, either breast or bottle. Infants were determined to be
NCPAP. If these infants were restarted on NCPAP within fully established on oral feeds when they had taken ≥120 mL/
24 hours, they were eligible for randomisation. Infants with sig- kg/day by mouth for 24 hours. Our secondary objectives were:
nificant congenital, respiratory, cardiac or airway abnormalities the duration to the first attempted oral feed, the duration of
or infants who were still on patient-assist NCPAP (added pres- respiratory support, CLD defined as respiratory support at
sure support breaths) were not eligible for enrolment. 36 weeks of gestational age, duration of hospital stay, episodes
Investigators, clinicians and caregivers were not masked to the of apnoea. We recorded basic demographic data (including ges-
infant group assignment. tational age at birth, birth weight, Apgar scores, antenatal ster-
Written informed consent was obtained from the infant’s oids, mode of delivery). We have also recorded incidence of
parent or guardian prior to enrolment. Infants were randomised RDS, need for endotracheal ventilation, incidence of patent
at 32 weeks CGA, to continue on NCPAP at their current set- ductus arteriosus (defined as presence of patent ductus arterio-
tings (control group) or to HFNC commencing at 7 L/min with sus beyond the first 72 hours of life), incidence of necrotising
oxygen as required to keep their oxygen saturation (SpO2) enterocolitis (≥IIA according to modified Bell’s criteria)12 and
between 88% and 95% (intervention group). HFNC was admi- incidence of early onset sepsis (using definitions published in
nistered using the Fabian Therapy Evolution (Acutronic Medical the National Institute for Health and Care Excellence
Systems AG, Switzerland). Prongs were selected so as to sit guidelines).13
inside and occlude less than one-third of the nostrils as per the Our sample size was based on our primary outcome and our
manufacturer’s recommendations. NCPAP was administered end point was the number of days taken to establish full oral
using the Fabian Therapy Evolution or the Infant Flow SiPAP feeds from the time of randomisation (32 weeks CGA). From a
system (CareFusion, USA). NCPAP was given using nasal prongs retrospective chart review on our institution, we estimated that
and/or masks of an appropriate size as per the manufacturer’s it would take infants on NCPAP at 32 weeks CGA 35 days
recommendations. The treatment allocation schedule was gener- (±8 days) to achieve full oral feeding (infants in this retrospect-
ated using a statistical software package (StatsDirect V.2.6.1, ive review were managed solely on NCPAP). We calculated that
StatsDirect, UK) and was concealed from the treating clinicians. in order to demonstrate a reduction of 7 days in our primary
Group assignment was written on cards and placed in sequen- outcome in infants receiving HFNC with 80% power and α
tially numbered, sealed opaque envelopes by an independent 5%, we needed to enrol 44 subjects (22 in each group). We ana-
administrative assistant. Immediately following enrolment, the lysed outcome data using the ‘intention-to-treat’ principle with
study investigator opened the next envelope in the sequence in StatsDirect, V.2.6.1 (StatsDirect) using Student’s t-test, Fisher’s
the presence of an attending physician and revealed the group exact test and Mann-Whitney U test as appropriate.
assignment. Enrolled infants were monitored until full oral
feeding was established. The monitoring included four hourly RESULTS
observations of heart rate, respiratory rate, work of breathing, There were 149 VLBW infants born at <30 weeks of gestation,
oxygen requirement, frequency and duration of apnoeic epi- who had survived to 32 weeks of gestation during the recruit-
sodes defined as breath holding for >20 s and daily physical ment period ( January 2013 to December 2014). Of those, 59
examinations. If any infant on HFNC deteriorated clinically, the infants were eligible for randomisation and 44 infants were ran-
clinician on call was to assess the infant and determine whether domised to the trial (figure 1). There were no statistical differ-
the infant was stable on HFNC, or if they needed to be ences between the NCPAP group and HFNC group in relation
switched back to NCPAP along with any other required clinical to the patient characteristics at enrolment (table 1).
F2 Glackin SJ, et al. Arch Dis Child Fetal Neonatal Ed 2016;0:F1–F4. doi:10.1136/archdischild-2016-311388
 Original article

The number of days taken to achieve full oral feeding was not
different between the groups (HFNC group 36.5 days±18.2 vs
NCPAP group 34.1 days±11.2, p=0.61).
There were no statistically significant differences between the
groups in terms of our predefined secondary outcomes (table 2).
Six infants (27%) in the NCPAP group were off respiratory
support when offered first oral feeds compared with one infant
(4.5%) in the HFNC group ( p=0.09). There were no other
adverse outcomes or events in any of the infants in either group
(including nasal trauma) in this study and no infants required to
be changed from HFNC to NCPAP due to clinical
deterioration.

DISCUSSION
Our hypothesis that there would be a difference of 7 days to
reach full oral feeds between the group randomised to NCPAP
and the group randomised to HFNC was rejected. While the
difference of 7 days may have seemed excessive, we were
looking for a clinically relevant result. A recent case series
looked at the establishment of full oral feeds as their primary
outcome in two groups of preterm infants with bronchopul-
monary dysplasia (defined as oxygen dependency beyond
28 days), managed either with NCPAP alone or with NCPAP for
2 weeks postextubation followed by HFNC.8 In their initial ana-
lysis, they found no difference between the two groups in the
duration to reach full oral feeds ( p>0.5). However, when they
compared infants still on respiratory support at 34 weeks, they
found that those on HFNC reached full oral feeds significantly
earlier than those on NCPAP (39.4 vs 41 weeks CGA;
p=0.002). However, this subgroup analysis result was likely
confounded by the fact that they waited to feed the NCPAP
infants until they were off respiratory support due to concern
Figure 1 Consort flow diagram of study enrolment. HFNC, high flow about aspiration, which may have led to some infants missing
nasal cannula; NCPAP, nasal continuous positive airway pressure. their opportunity to learn how to feed and developing oral aver-
sion. Their NCPAP group were also of earlier gestation and
lower birth weight, and therefore represented a different popu-
lation to the HFNC group.
Our results are supported by an RCT which included duration
Table 1 Patient characteristics at enrolment to reach full oral feeds in their secondary outcomes and found
HFNC NCPAP no difference in these results between the HFNC group and the
(n=22) (n=22) p Value

Gestational age at birth (weeks)* 26.9±1.5 27.3±1.5 0.41

Birth weight (g)* 868±160 891±202 0.68
Table 2 Primary and secondary outcomes
Apgar score 1st minute† 6 (5–7) 5 (4–6) 0.35 HFNC NCPAP
Apgar score 5th minute† 8 (7–8) 8 (6–8) 0.41 (n=22) (n=22) p Value
Complete course antenatal steroids‡ 77 86 0.7
Days to full oral feeds from enrolment 36.5±18.2 34.1±11.2 0.61
Caesarean delivery‡ 77 82 >0.99 (days)*
Respiratory distress syndrome‡ 100 100 NS First oral feed (days from enrolment)* 9.3±6.5 10.9±4.8 0.37
Endotracheal ventilation‡ 68 77 0.74 Days of respiratory support from 21.9±12.7 18.8±9.4 0.4
Patent ductus arteriosus‡ 91 67 0.07 enrolment (days)*
Necrotising enterocolitis (≥IIA)‡ 10 5 >0.99 Length of hospital stay from enrolment 80.4±17.3 78.2±14.1 0.68
Early onset sepsis‡ 9.5 5 >0.99 (days)*
Day of life at enrolment* 36±10 33±10 0.44 Chronic lung disease (at 36/40)† 64 68 >0.99
Weight at enrolment (g)* 1327±253 1272±181 0.42 Episodes of apnoea (episodes/day on 0 (0–0.036) 0.014 (0–0.151) 0.21
respiratory support)‡
*Mean±SD compared with t-test.
†Median (IQR) compared with Mann-Whitney U test. *Mean±SD compared with t-test.
‡n (%) compared with Fisher’s exact test. †n (%) compared with Fisher’s exact test.
HFNC, high flow nasal cannula; NCPAP, nasal continuous positive airway pressure; ‡Median (IQR) compared with Mann-Whitney U test.
NS, not significant. HFNC, high flow nasal cannula; NCPAP, nasal continuous positive airway pressure.

Glackin SJ, et al. Arch Dis Child Fetal Neonatal Ed 2016;0:F1–F4. doi:10.1136/archdischild-2016-311388 F3
 Original article

NCPAP group.10 In a retrospective study, Yoon et al7 found a although, the sample size was based on our primary outcome,
reduction in the duration to reach full enteral feeds in their we achieved full recruitment and lost no participants to
HFNC group. However, a major limitation of their study was follow-up.
that there were only 55% of infants in the HFNC group actually
on HFNC. CONCLUSION
One of the benefits of HFNC is the lack of nasal trauma to We found no difference in the duration to reach full oral feeds
infant’s friable mucosa. There were no episodes of nasal trauma between stable preterm infants managed on HFNC and those
in either group in our study. This was likely due to the popula- managed on NCPAP. While there is widespread concern about
tion studied who were >32 weeks CGA and whose skin was less oral feeding NCPAP infants, we did not show any difference in
friable and therefore at less risk of nasal trauma than younger episodes of apnoea, desaturations or bradycardias and we had
infants. no episodes of aspiration in either group. Future studies should
The results of our secondary outcomes are consistent with investigate if there is a difference in breast-fed infants between
other published studies, which have shown no difference in the HFNC and NCPAP managed infants.
safety or efficacy of HFNC over NCPAP in stable preterm
infants. We also showed that there is no difference between the Contributors SJG performed literature search, was involved in the study design,
groups in the duration of respiratory support requirement. In data collection, data analysis, data interpretation and has written first draft of the
manuscript. AO was involved in the study design, staff training prior to the study,
some previous studies, infants on NCPAP were not offered oral data collection and reviewed final version of the manuscript. SG was involved in the
feeds due to concern about aspiration. We did not have any data collection, data analysis and reviewed final version of the manuscript. JS was
cases of aspiration or acute respiratory deterioration following involved in the study design, data collection, data interpretation and reviewed final
oral feeds in either group. This is likely because of the stable version of the manuscript. JM supervised the conduct of the study, was involved in
the study design, data analysis and data interpretation and reviewed final version of
population of infants studied. the manuscript.
We did not get formal feedback from parents or nursing staff
Competing interests None declared.
in this study. Anecdotally, the nursing staff found it easier to
manage infants on HFNC in comparison to those on NCPAP Ethics approval Research Ethics Committee, Coombe Women and Infants
University Hospital, reference number 18-2012.
and the parents preferred HFNC as they had a clearer view of
their infant’s face, which facilitated bonding and kangaroo care. Provenance and peer review Not commissioned; externally peer reviewed.
In our neonatal unit, prior to the commencement of the RCT,
occasionally infants on NCPAP were cautiously commenced on REFERENCES
oral feeds if they were displaying feeding cues. Being part of the 1 Kotecha SJ, Adappa R, Gupta N, et al. Safety and efficacy of high-flow nasal
cannula therapy in preterm infants: a meta-analysis. Pediatrics 2015;136:542–53.
RCT made the nursing staff more confident than they had been
2 Manley BJ, Owen LS, Doyle LW, et al. High-flow nasal cannulae in very preterm
prior to the trial to orally feed infants on NCPAP. infants after extubation. N Engl J Med 2013;369:1425–33.
While we planned to include infants who were both bottle 3 Kubicka ZJ, Limauro J, Darnall RA. Heated, humidified high-flow nasal cannula
and breast fed, we had disappointingly few infants who were therapy: yet another way to deliver continuous positive airway pressure? Pediatrics
trialled on breast feeding and no infants who breast fed success- 2008;121:82–8.
4 Collins CL, Holberton JR, Barfield C, et al. A randomized controlled trial to compare
fully. It is possible that if more infants were breast fed, it may heated humidified high-flow nasal cannulae with nasal continuous positive airway
have influenced the results in favour of the HFNC group. The pressure postextubation in premature infants. J Pediatr 2013;162:949–54.e1.
low numbers of breast-fed infants likely reflects our general 5 Kugelman A, Riskin A, Said W, et al. A randomized pilot study comparing heated
population rather than either respiratory support mechanism. humidified high-flow nasal cannulae with NIPPV for RDS. Pediatr Pulmonol
2015;50:576–83.
The strengths of our study include the fact that it was an
6 Campbell DM, Shah PS, Shah V, et al. Nasal continuous positive airway pressure
RCT and it was safely conducted in an appropriate population from high flow cannula versus infant flow for preterm infants. J Perinatol
with proper observations and no infants were lost to follow-up. 2006;26:546–9.
Our selected population were a group of stable preterm infants 7 Yoon S-h, Kwon Y-h, Park H-k, et al. High flow nasal cannula versus nasal CPAP in
who were still NCPAP-dependent at 32 weeks CGA, as the preterm infants. J Korean Soc Neonatol 2011;18:293–300.
8 Shetty S, Hunt K, Douthwaite A, et al. High-flow nasal cannula oxygen and nasal
primary aim of the study was to determine the feeding outcome continuous positive airway pressure and full oral feeding in infants with
of these infants. bronchopulmonary dysplasia. Arch Dis Child Fetal Neonatal Ed 2016;101:F408–11.
The principal limitation of our study is that caregivers and 9 Hanin M, Nuthakki S, Malkar MB, et al. Safety and efficacy of oral feeding in
outcome assessors were not masked to the infants’ group assign- infants with BPD on nasal CPAP. Dysphagia 2015;30:121–7.
10 Yoder BA, Stoddard RA, Li M, et al. Heated, humidified high-flow nasal cannula
ment. This introduced the potential for bias in that nursing staff
versus nasal CPAP for respiratory support in neonates. Pediatrics 2013;131:e1482–90.
could offer extra feeds to infants who they felt were showing 11 Silverman WA, Andersen DH. A controlled clinical trial of effects of water mist on
feeding cues according to their preferred intervention. Most obstructive respiratory signs, death rate and necropsy findings among premature
nursing staff preferred handling the infants on HFNC, so they infants. Pediatrics 1956;17:1–10.
could influence the results of the study and offer the HFNC 12 Bell MJ, Ternberg JL, Feigin RD, et al. Neonatal necrotizing enterocolitis.
Therapeutic decisions based upon clinical staging. Ann Surg 1978;187:1–7.
group more oral feeds. However, we found no difference in the 13 Caffrey Osvald E, Prentice P. NICE clinical guideline: antibiotics for the prevention
number of oral feeds offered to infants in either group. Another and treatment of early-onset neonatal infection. Arch Dis Child Educ Pract Ed
limitation of our study was the small number of patients, 2014;99:98–100.

F4 Glackin SJ, et al. Arch Dis Child Fetal Neonatal Ed 2016;0:F1–F4. doi:10.1136/archdischild-2016-311388