OBJECTIVE: To determine whether nasal continuous positive airway pressure (NCPAP) given with nasal prongs compared with nasal mask reduces the rate of intubation and mechanical ventilation in preterm infants within 72 hours of starting therapy.
METHODS: Infants <31 weeks’ gestation treated with NCPAP were randomly assigned to receive it via either prongs or mask. Randomization was stratified by gestational age (<28 weeks, 28–30 weeks) and according to whether NCPAP was started as a primary treatment for respiratory distress or postextubation. Infants were intubated and ventilated if they fulfilled 2 or more of 5 failure criteria (worsening signs of respiratory distress; recurrent apnea treated with mask positive pressure ventilation; fraction of inspired oxygen >0.4 to keep oxygen saturation >88% sustained for 30 minutes; pH <7.2 on 2 blood gases ≥30 minutes apart; Pco2 >9 kPa [68 mm Hg] on 2 blood gases ≥30 minutes apart) within 72 hours of starting therapy. The groups were treated the same in all other respects. We recorded relevant secondary outcomes and analyzed data by using the intention-to-treat principle.
RESULTS: We enrolled 120 infants. Thirty-two of 62 (52%) infants randomly assigned to prongs were intubated within 72 hours, compared with 16/58 (28%) of those randomly assigned to mask (P = .007). There were no statistically significant differences between the groups in any secondary outcomes.
CONCLUSIONS: In premature infants, NCPAP was more effective at preventing intubation and ventilation within 72 hours of starting therapy when given via nasal masks compared with nasal prongs.
- CPAP —
- continuous positive airway pressure
- Fio2 —
- fraction of inspired oxygen
- IQR —
- interquartile range
- NCPAP —
- nasal continuous positive airway pressure
- NIPPV —
- nasal intermittent positive pressure ventilation
- NMH —
- National Maternity Hospital
What’s Known on This Subject:
Nasal continuous positive airway pressure (NCPAP) is commonly given to premature infants with nasal prongs and nasal masks. Prongs and masks appear to injure the nose of preterm infants with equal frequency.
What This Study Adds:
Nasal masks are more effective than nasal prongs for preventing intubation and mechanical ventilation in premature infants within 72 hours of starting NCPAP.
Nasal continuous positive airway pressure (NCPAP) is used for the primary treatment of respiratory distress syndrome in preterm infants.1–3 Giving NCPAP to preterm infants after extubation reduces the need for reintubation and mechanical ventilation.4 The interfaces most commonly used to deliver NCPAP are prongs and masks. Short binasal prongs (short tubes that fit in both nostrils) are more effective than a single nasal prong (longer tube in 1 nostril)5,6 or nasopharyngeal prongs (long tubes in both nostrils).7 Masks that fit over the nose were developed many years ago8,9 and are commonly used today (Fig 1).10 Nasal trauma has been reported with the use of both nasal masks and prongs11,12 and occurs equally often with each interface.13 The effectiveness of NCPAP given with these interfaces has not previously been compared.
Like the majority of Irish10 and UK14 nurseries, we use the Infant Flow Driver, Infant Flow Advance, and Infant Flow SiPAP (Viasys Healthcare, Yorba Linda, CA) devices to give NCPAP to preterm infants in our NICU. These devices can administer NCPAP with both short binasal prongs and nasal masks (Viasys Healthcare) that are available in 3 sizes (small, medium, and large). Before the study, infants who started NCPAP in our NICU were placed on nasal prongs or nasal mask at the discretion of the treating doctors and nurses, and prongs were used more often.
We hypothesized that giving NCPAP with nasal prongs compared with nasal mask would reduce the rate of intubation and mechanical ventilation in preterm infants within 72 hours of starting NCPAP therapy in the NICU.
We conducted this randomized controlled trial at the National Maternity Hospital (NMH), Dublin, a stand-alone university maternity hospital with ∼10 000 deliveries annually. The hospital has a tertiary level NICU to which ∼150 very low birth weight infants are admitted annually. Infants were eligible for inclusion if they were born at <31 weeks’ gestation by best obstetric estimate (dated by early obstetric ultrasound or last menstrual period) and were starting NCPAP for the first time in the NICU. We included both infants who were given NCPAP as primary treatment of respiratory distress and infants who were given NCPAP postextubation. Infants with major congenital anomalies were excluded. The study protocol was approved by the Ethics Committee at the NMH.
Written informed consent was obtained before enrollment in the study from a parent or guardian by 1 of the members of the research team. Infants were randomly assigned to receive NCPAP with either nasal prongs or nasal mask in a 1:1 ratio. The randomization was stratified by gestational age (<28 weeks and 28–30 weeks) and according to whether the infant started NCPAP as a primary treatment of respiratory distress or as an aid to extubation. The treatment allocation schedule was generated in permuted blocks of 4 by using a random number table. The treatment allocation was contained in sequentially numbered sealed opaque envelopes. The next envelope in the sequence was opened after the decision to start NCPAP had been made and the continuous positive airway pressure (CPAP) driver had been set up in the NICU. Infants receiving NCPAP postextubation were placed on CPAP directly after removal of the endotracheal tube. Infants of multiple pregnancies were randomly assigned individually. Infants were nursed supine if umbilical catheters were in situ and were otherwise nursed prone. Active mouth closure with chin straps or other methods were not attempted. All infants started on a NCPAP pressure of 5 to 7 cm H2O that could be increased up to a maximum of 9 cm H2O. NCPAP pressures were not continuously recorded electronically but were recorded manually once an hour from the machine. Nonsynchronized nasal intermittent positive pressure ventilation (NIPPV) could be given to infants who had increasing clinical respiratory distress, increasing oxygen requirements and/or apnea. The decision to use NIPPV and all changes to NCPAP pressure and the pressures and rate used during NIPPV were at the discretion of treating clinicians.
All infants enrolled in the study received a loading dose of 10 mg/kg caffeine base. Infants who received NCPAP as primary treatment of respiratory distress were given caffeine shortly after starting NCPAP. Infants who received NCPAP postextubation were given caffeine before extubation. Caffeine base was given regularly at a dose of 2.5 mg/kg 24 hours after the loading dose and at daily intervals thereafter. The regular dose of caffeine could be increased to a maximum of 5 mg/kg daily. Infants on NCPAP were to remain on their allocated interface for the whole duration of treatment with NCPAP (ie, if an infant randomly assigned to receive NCPAP via nasal prongs was intubated at any time, he or she received NCPAP by nasal prongs at all subsequent extubation attempts).
The primary outcome of our study was intubation and ventilation within 72 hours of starting NCPAP. Infants were intubated and ventilated if they met 2 or more of 5 failure criteria:
worsening clinical signs of respiratory distress (increasing tachypnea; expiratory grunting; intercostal, subcostal, and/or sternal recession);
apnea treated with positive pressure ventilation (PPV) by mask on 2 or more occasions in 1 hour;
fraction of inspired oxygen (Fio2) >0.4 to maintain pulse oxygen saturations ≥88% for >30 minutes;
pH <7.2 on 2 arterial or capillary blood gases taken >30 minutes apart; and
Pco2 >9 kPa (68 mm Hg) on 2 arterial or capillary blood gases taken >30 minutes apart.
We recorded complications of prematurity as secondary outcomes including death, necrotizing enterocolitis, intraventricular hemorrhage, respiratory support on day 28, oxygen therapy at 36 weeks’ corrected gestational age, and retinopathy of prematurity. All secondary outcomes were determined before discharge home from hospital unless stated otherwise.
In the 3 years before this study, ∼40% of infants with a birth weight of ≤1500 g or <29 weeks’ completed gestational age who received NCPAP as primary treatment of respiratory distress were ventilated at our hospital while the rate of reintubation among those extubated to NCPAP was lower. The interface on which infants started was not recorded but was likely to be prongs in the majority. We estimated that the rate of intubation in the mask group would be 50% and that we would need to enroll 120 infants to show a reduction in treatment failure from 50% to 25% with a 2-tailed type I error rate of 0.05 and a power of 80%. An external data monitor analyzed the data from the first 60 infants enrolled and recommended completing enrollment. Data were analyzed by using the intention-to-treat principle by using PASW version 18 software (SPSS Inc, Chicago IL). We report the primary outcome and dichotomous secondary outcome data as proportions and compared the groups by using nonparametric tests (Pearson χ2). We report secondary continuous outcome data as mean (SD) when the data were normally distributed and median (interquartile range [IQR]) when the data were skewed, and we compared the groups by using parametric (t test) and nonparametric tests (Mann-Whitney U test) as appropriate. We considered P values <.05 statistically significant.
One hundred forty-five infants born at <31 weeks’ gestation were admitted to the NICU of the NMH between August 2009 and November 2010. Twenty-five infants were not enrolled into the study (Fig 2); 120 infants were randomly assigned, 62 to prongs and 58 to masks. The groups were well matched for gestational age, birth weight, gender, mode of delivery, age at randomization, and duration of ventilation before randomization (Table 1). More than 90% of infants in each group were exposed to antenatal steroids, and approximately half of enrolled infants had been intubated and received surfactant before randomization. Among infants who were randomly assigned post-extubation, the median Fio2 at randomization was not different between the groups (Table 2).
Thirty-two of the 62 (52%) infants randomly assigned to nasal prongs were intubated and ventilated within 72 hours of starting NCPAP compared with 16/58 (28%) infants randomly assigned to nasal mask (P = .007, number needed to treat = 4). One infant randomly assigned to prongs was intubated without reaching the prespecified failure criteria. Per-protocol analysis of the data revealed a statistically significant difference in the primary outcome between the groups (P = .012). The most common reasons that infants reached failure criteria were clinical signs of respiratory distress and Fio2 > 0.4 (Table 3). The median NCPAP pressure among infants who reached the primary outcome was not different between the groups (prongs 6 [6–7] cm H2O versus mask 6 [5–7] cm H2O, P = .818). Infants who reached the primary outcome did so at a median (IQR) of 315 (101–960) minutes post randomization, the latest intubation occurring at 2670 minutes (45 hours) postrandomization.
Although more infants randomly assigned to prongs were treated with NIPPV within 72 hours, the difference between the groups was not statistically significant (Table 4). There were no differences between the groups in other secondary outcomes that we measured (Table 4). In particular, there were no differences in the rates of pneumothorax, severe cranial ultrasound abnormalities, oxygen therapy at 36 weeks’ corrected gestational age, or death before hospital discharge.
Only 3% of infants in both groups had nasal trauma sufficient to prompt clinicians or nursing staff to change the interface. Two infants randomly assigned to prongs were temporarily changed to mask due to nasal trauma on a total of 3 occasions, 1 on day 22 of NCPAP and the other on day 49 and again on day 65 of NCPAP. Two infants randomly assigned to mask group temporarily changed to prongs due to nasal trauma, on day 24 and 47 of NCPAP. One additional infant randomly assigned to mask changed to prongs during NCPAP treatment. This infant developed severe lung disease after Staphylococcus aureus pneumonia. After 13 days of NCPAP (day 32 after birth), he had increased oxygen requirements, and the attending clinician was concerned that leak from the interface might be a problem. There was no reduction in the Fio2 with a change to prongs, and the infant was mechanically ventilated 24 hours later. Infants from both groups who switched interface were treated with the alternative interface for median (IQR) of 2 (2–2) days, a median (IQR) of 3 (2–5) percent of the total duration they spent on NCPAP. No infant had nasal injury that prompted referral to a plastic surgeon.
More infants randomly assigned to prongs who were born <28 weeks (Table 5) and who started NCPAP postextubation (Table 2) were intubated within 72 hours of starting therapy; however, this study was not sufficiently powered to detect differences in subgroups, and these results should be interpreted with caution.
We studied infants treated with NCPAP both as a primary treatment of respiratory distress and as an aid to extubation because these are the clinical situations in which we routinely give NCPAP to preterm infants in our nursery. Though these populations are somewhat distinct, we aim to quickly extubate preterm infants intubated for treatment of respiratory distress syndrome at our hospital. Thus, when designing the study, we thought it reasonable to choose treatment failure within 72 hours of randomization as our primary outcome as failure within that time would likely be due to respiratory causes, whereas failure after this time would more likely be due to nonrespiratory causes. We believed that if we detected difference in the primary outcome between the groups, it would be due to more effective NCPAP in that group. Of the 21 infants who did not reach the primary outcome but were later intubated (8 randomly assigned to prongs, 13 to mask), 10 were intubated for surgery, 8 in association with infection, and 3 for apnea. We suspected that failure rates would differ according to whether infants had previously been intubated for respiratory distress and so stratified the randomization on that basis. Overall, 30/57 (53%) of the infants receiving CPAP as primary treatment were intubated and ventilated within 72 hours of starting treatment, whereas 18/63 (29%) of the infants treated postextubation were intubated within 72 hours. These rates are comparable with those seen in other trials of early NCPAP in preterm infants.1,2 We also stratified the randomization according to gestational age because we suspected that failure rates would be higher in more immature infants; 21/51 (41%) infants <28 weeks were intubated <72 hours, compared with 27/69 (39%) infants 28 to 30 weeks. Our method of randomization resulted in groups that were well balanced for demographic data. Infants <28 weeks allocated to prongs were randomly assigned earlier than infants to masks, though this difference was not statistically different (Table 5). We do not know whether this difference had an effect on the difference we observed between the groups.
The major weakness of our study was the unblinded nature of the intervention and the consequent potential for the bias of either or both caregivers and/or outcome assessors to influence our results. Although NCPAP pressures could be increased up to 9 cm H2O and NIPPV started at the discretion of the treating clinician, there were no predefined criteria as to when pressures should be increased or NIPPV commenced. The median NCPAP pressure among infants who reached the primary outcome was not different between the groups. Although NIPPV was given to more infants randomly assigned to prongs within 72 hours of randomization, the difference was not statistically significant. The use of NIPPV did not appear to delay intubation just beyond 72 hours in any enrolled infant; the earliest intubation among the 21 infants who did not reach the primary outcome but were subsequently intubated was on day 5 postrandomization. All enrolled infants received a loading dose of caffeine and were given it at 24-hour intervals thereafter at a dose of 2.5 mg/kg, which could be increased. We did not record the regular dose of caffeine given to enrolled infants. However, because most infants who reached the primary outcome did so early (75% within 16 hours), and only 9 infants (5 randomly assigned to prongs, 4 randomly assigned to mask) failed >24 hours after randomization, it is most unlikely that there was a significant difference in the dose of caffeine either group received. To minimize bias in assessment of the primary outcome, we used prespecified failure criteria. These criteria were agreed and are used in clinical practice by the neonatologists at our NICU and are similar to those used in recent randomized trials of early NCPAP in preterm infants.1–3 These criteria were observed in all but 1 infant, and per-protocol analysis of the data revealed a statistically significant difference in the primary outcome between the groups. It is noteworthy that before this study most staff at our nursery preferred to use prongs and that we hypothesized that prongs were more effective.
Although we did not directly measure it, we speculate that nasal masks were more effective than nasal prongs by delivering higher NCPAP pressure to the nasopharynx. By the nature of their design, binasal prongs reduce the internal diameter of the nose. They may, therefore, increase resistance and reduce the pressure delivered to the airway. An in-vitro study revealed reductions in the pressure across binasal prongs used for NCPAP in preterm infants with the largest pressure drops seen with the smallest prongs at the highest rates of flow of gas.15 This may partly explain the marked difference in failure rates we observed among more immature infants who were likely to have received NCPAP through the smallest binasal prongs.
Our findings are applicable in units where the CPAP devices that we studied are used. Nasal masks compatible with these devices have been available for many years and had been used at our hospital for >5 years before this study. They are not more difficult to use than binasal prongs and are used in the majority of Irish10 and many UK units.14 The connection to the CPAP circuit is the same for the mask as that for the binasal prongs, and no specific training is required to use the mask. We did not study other CPAP-generating devices (eg, bubble CPAP, ventilator-derived CPAP), and so we do not know if masks are more effective than prongs when used with these devices. Similarly, because we did not attempt active closure of the mouth in our study by using chin straps, pacifiers, or other methods, we do not know if our findings are applicable in these circumstances. However, given the large treatment effect we observed, these questions merit examination in randomized studies.
Our study agreed with a previous study that revealed that nasal trauma occurred in a small proportion of infants, with equal frequency with each interface and after several weeks of therapy.13 We also agree with other studies that revealed that trauma related to nasal prongs tends to be maximum around the medial aspect of the nasal septum and the columella, whereas trauma related to nasal masks is more often seen at the junction of the nasal septum and philtrum and at the glabella.11,12 As masks and prongs cause nasal trauma in differing distribution, the interface used is alternated in many units. If such an approach is to be used, we recommend that infants are started on nasal mask and that the interface only be alternated after 72 hours.
In premature infants, NCPAP was more effective at preventing intubation and ventilation within 72 hours of starting therapy when given via nasal masks compared with nasal prongs.
Thanks to Professor Peter Davis for his interim analysis of our data.
- Accepted July 10, 2012.
- Address correspondence to Colm O’Donnell, MB, MRCPI, MRCPCH, FRACP, PhD, Department of Neonatology, The National Maternity Hospital, Holles St, Dublin 2, Ireland. E-mail:
Dr Kieran made substantial contributions to the conception and design of the study. She acquired the data and made a substantial contribution to the analysis and interpretation of the data. She wrote the first draft of the article. Drs Twomey, Molloy, and Murphy made substantial contributions to the design of the study. They critically revised the article for important intellectual content. Dr O’Donnell conceived and designed the study. He analyzed and interpreted the data. He re-drafted the article and revised it for important intellectual content. All authors approve this version of the article.
This work was presented at the Pediatric Academic Societies/Asian Society for Pediatric Research Meeting; April 30, 2011 Denver, CO.
No company (or agent thereof) whose products feature in this study or any competing company (or agent thereof) had any role in study design or conduct; data collection or interpretation; the decision to present or publish the data; or writing this article.
This article is being submitted only to Pediatrics and we will not submit it elsewhere while it is under consideration at Pediatrics. Although the results have previously been published in abstract form (Arch Dis Child. 2011;96 [suppl 1]:A3 and PAS 2011), they have not previously been published in full.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: The investigators were supported with seed funding from the National Children’s Research Centre.
- Finer NN,
- Carlo WA,
- Walsh MC,
- et al.,
- SUPPORT Study Group of the Eunice Kennedy Shriver NICHD Neonatal Research Network
- Sandri F,
- Plavka R,
- Ancora G,
- et al.,
- CURPAP Study Group
- Davis PG,
- Davies MW,
- Faber B
- Roukema H,
- O’ Brien K,
- Nesbitt K,
- Zaw W
- Kattwinkel J,
- Fleming D,
- Cha CC,
- Fanaroff AA,
- Klaus MH
- Kieran EA,
- Walsh H,
- O’Donnell CPF
- Robertson NJ,
- McCarthy LS,
- Hamilton PA,
- Moss ALH
- Fischer C,
- Bertelle V,
- Hohlfeld J,
- Forcada-Guex M,
- Stadelmann-Diaw C,
- Tolsa JF
- Yong SC,
- Chen SJ,
- Boo NY
- Owen LS,
- Morley CJ,
- Davis PG
- De Paoli AG,
- Morley CJ,
- Davis PG,
- Lau R,
- Hingeley E
- Copyright © 2012 by the American Academy of Pediatrics