Sudden vs Pressure Wean From Nasal Continuous Positive Airway Pressure in Infants Born Before 32 Weeks of Gestation: A Randomized Clinical Trial

Key Points

Question  Are there differences between sudden wean and pressure wean from nasal continuous positive airway pressure (nCPAP) in very preterm infants with respect to early growth and respiratory support needs?

Findings  In this randomized clinical trial, preterm infants were randomized to sudden vs pressure nCPAP wean and there were no differences found in growth and respiratory support. More infants born before 28 weeks with pressure wean were weaned successfully during the first attempt.

Meaning  Overall, in very preterm infants, sudden wean and pressure nCPAP wean are equally good choices when early growth and respiratory support are considered; in infants born before 28 weeks, pressure wean may be preferred.

Importance  Nasal continuous positive airway pressure (nCPAP) is a well-established treatment of respiratory distress syndrome in preterm infants. Suboptimal weaning from nCPAP may be associated with lung injury, pulmonary morbidity, and infant weight gain. To our knowledge, the best weaning strategy from nCPAP is unknown.

Objective  To compare the effect of sudden wean and pressure wean from nCPAP in very preterm infants.

Design, Setting, and Participants  A randomized, clinical, open-label, multicenter trial was conducted at 6 neonatal intensive care units in Denmark from September 2012 to December 2016 and included infants born before 32 weeks of gestation.

Interventions  Sudden wean with discontinuation of nCPAP without a prior reduction in pressure. Pressure wean with gradual pressure reduction prior to the discontinuation of nCPAP.

Main Outcome and Measures  The primary outcome was weight gain velocity from randomization to postmenstrual age 40 weeks. Secondary outcomes included other measures of growth, nCPAP and the duration of oxygen supplementation, postmenstrual age at successful wean and at discharge, successful wean at the first attempt, the number of attempts to wean, and the length of the hospital stay. Prespecified subgroup analyses by gestational age were performed.

Results  Of the 372 randomized infants, 185 (49.7%) were randomized to sudden wean and 187 infants (50.3%) to pressure wean. A total of 177 infants in both groups completed the trial (median gestational age for sudden and pressure wean, 30 weeks [interquartile range, 29-31]; male: sudden wean, 89 [50%]; pressure wean, 96 [54%]). There was no difference in mean [SD] weight gain velocity from randomization to 40 weeks postmenstrual age between the 2 groups (22 [6] g/kg/day). No difference was found in any of the secondary outcomes. More infants born before 28 weeks of gestation were successfully weaned from nCPAP during the first attempt in the pressure wean group compared with the sudden wean group (risk difference, 31%; 95% CI, 13%-50%), but there was no difference in the duration of nCPAP and oxygen supplementation.

Conclusions and Relevance  Overall, we found no difference in weight gain velocity or any of the secondary outcomes between very preterm infants who were randomized to sudden wean or pressure wean from nCPAP. However, among infants born before 28 weeks’ gestation, infants from the pressure wean group were more often successfully weaned during the first attempt without a longer total duration of nCPAP treatment.

Trial Registration  ClinicalTrials.gov Identifier: NCT01721629

Nasal continuous positive airway pressure (nCPAP) is a widely used treatment of respiratory distress syndrome in preterm infants.^1-9 Nasal continuous positive airway pressure prevents alveolar collapse, reduces the work of breathing, prevents apneas of prematurity, and prevents lung injury.^4,5,10 However, nCPAP is also associated with gastric distension,¹¹ nasal septum necrosis,¹² and pneumothorax.⁶ The benefits and risks of nCPAP are well described, but the optimal strategy for nCPAP weaning remains to be determined.¹³

Optimal nCPAP weaning is of great clinical importance because too early weaning from nCPAP and repeated weaning attempts may eventually cause lung injury. Repetitive attempts to wean the preterm infant from nCPAP may cause cycles of atelectasis that are followed by the need for alveolar recruitment, which can damage the alveoli (atelectrauma), increase pulmonary morbidity, and cause chronic lung damage.^10,14 Pulmonary morbidity causes prolonged oxygen therapy, a prolonged length of stay in the neonatal intensive care unit (NICU),^15,16 and has been shown to be independently associated with impaired growth (ie, decreased weight gain velocity from birth to discharge from the NICU in preterm infants).^17,18 Lung injuries on the basis of inappropriate nCPAP weaning with atelectrauma could be one of the most important and potentially preventable elements of lung injury in very preterm infants.

Three main weaning methods have been described: graded time off wean, sudden wean, and pressure wean.^13,19,20 Graded time off wean has been shown to prolong nCPAP and oxygen duration and increase the risk of bronchopulmonary dysplasia (BPD) and the length of hospital stay compared with either sudden wean or pressure wean.^21-25 However, it is unknown whether sudden wean or pressure wean is the optimal weaning strategy.

The primary aim of this study was to compare the effect of sudden wean and pressure wean from nCPAP on weight gain velocity in very preterm infants. Several secondary outcomes related to measures of growth, weaning attempts, nCPAP, and oxygen therapy were also studied.

This randomized clinical trial (RCT) was conducted in 6 NICUs in the Central and North Regions of Denmark between September 2012 and December 2016. The Central Denmark Region committees on research ethics approved the protocol and written informed consent was obtained from the parents of every study infant. The trial was registered at Clinical Trials (ClinicalTrials.gov) as NCT01721629, and the trial protocol is in Supplement 1. Further details are in Supplement 2.

All infants born before 32 weeks of gestation who required nCPAP for more than 24 hours were assessed for eligibility. To be randomized. the infant had to fulfill all the following inclusion criteria: a current postmenstrual age (PMA) of 29 weeks and 0 days or older, nCPAP of 24 hours or more, nCPAP pressure of less than 8 cm H₂O, an oxygen requirement of less than 30% and not increasing, a respiratory rate of less than 70 breaths per minute, less than 3 episodes of bradycardias (<70 beats/minute) or desaturations (<70%) in the preceding 24 hours, and tolerate time off nCPAP during nursing procedures (up to 15 minutes).

Infants were ineligible if they had undergone gastrointestinal surgery, had known or suspected congenital neuromuscular disease, or had chromosomal anomalies or congenital malformations of the heart, lung, or gastrointestinal tract. Furthermore, infants of parents who were unable to speak or read Danish were considered ineligible.

A computer-generated web-based randomization program with a random allocation sequence and log-on procedure that was specific to each center was developed by an independent statistician. This program prompted a systematic assessment of the eligibility criteria, provided a unique randomization number, and assigned the infant or the twin pair to the nCPAP weaning method in a 1:1 ratio. The attending physicians had responsibility for enrolling participants.

The randomization was stratified by study center and gestational age (<28 or 28-32 weeks) and balanced within randomly sized blocks of 2 to 4 to 6 infants in each center. Twins were randomized to the same weaning strategy. Masking to treatment allocation was impossible.

According to the Danish National Guideline, very preterm infants were stabilized on nCPAP immediately after birth if possible. If surfactant was needed the intubation surfactant extubation procedure was used.^26,27 Infants that failed this procedure and needed mechanical ventilation were always extubated to nCPAP when ready. Weaning from nCPAP was initiated as per the randomization allocation, when the infant met the inclusion criteria, which also served as readiness criteria for weaning from nCPAP.

In the sudden wean group, nCPAP was discontinued without prior reduction in nCPAP pressure. If the infant failed the discontinuation of the nCPAP, according to the failure criteria, nCPAP was recommenced (eFigure in Supplement 3). A minimum of 1 of the following failure criteria had to be met: persistent tachypnea (respiratory rate ≥ 70 breaths per minute), increased work of breathing (intercostal retractions), an increased oxygen requirement of more than 10% to maintain oxygen saturations or more than 90%, more than 3 episodes of bradycardias (<70 beats per minute) or desaturations (<70%) in the preceding 24 hours, major apnea or bradycardia that required positive pressure ventilation, and a transcutaneous carbon dioxide increase of more than 2.0 kPa from the starting point (if measured). A new attempt of sudden wean was at the earliest made 24 hours from the latest attempt and only when the readiness criteria were met again (eFigure in Supplement 3).

In the pressure wean group, the nCPAP pressure was decreased by 1 cm H₂O every 24 hours until a pressure of 4 cm H₂O was achieved. Then the nCPAP was discontinued. If any of the failure criteria (previously mentioned) were met during the pressure reduction, the nCPAP pressure was increased. If any of the failure criteria were met after discontinuation, nCPAP was recommenced. In both scenarios, a new attempt of pressure wean was at the earliest made 24 hours from the latest attempt and only when the readiness criteria were met again (eFigure in Supplement 3). In the sudden wean and pressure wean groups, a successful wean was defined as no need for recommencing nCPAP according to failure criteria for 3 days after nCPAP discontinuation (eFigure in Supplement 3).

The primary outcome was total infant weight gain from randomization to a PMA of 40 weeks divided by the weight at randomization and the number of days from randomization to 40 weeks PMA (ie, weight gain velocity [in grams/kilograms/day]). The weight was measured by digital baby scales.

The secondary prespecified outcomes were weight (in grams) at 40 weeks PMA, weight gain velocity (in grams/kilograms/day) measured from the date of randomization to the date of a successful wean, weight at successful wean (in grams), total days in nCPAP after randomization to a PMA of 40 weeks, PMA at successful wean, a successful wean during the first attempt defined as the ability to stay off nCPAP for 3 days without meeting any failure criteria after the first discontinuation of nCPAP, days to a successful wean measured from the date of randomization to the date of a successful wean, the number of attempts to discontinue nCPAP, supplemental oxygen after randomization to a PMA of 40 weeks (yes/no), days with supplemental oxygen after randomization to a PMA of 40 weeks, BPD defined as the need for supplemental oxygen at a PMA of 36 weeks, a need for supplemental oxygen at discharge from the NICU, length of stay in the NICU, and PMA at discharge.

We estimated the sample size on the basis of prior knowledge on weight gain velocity in preterm infants,²⁸ taking into account clustering by twinning.^29,30 We aimed to be able to show a difference in weight gain velocity of 2.5 g/kg/day between the 2 groups. For this difference to be detected with a power of 80% and a 2-sided significance level of P = .05, a total of at least 92 infants in each group had to be recruited. However, anticipating an average cluster size of 2 and an intracluster correlation coefficient of 0.35, the adjusted sample size was 125 infants in each group. To allow for a loss to follow-up and a lack of compliance of 15%, the size of the study population was estimated to be 150 infants in each group.

All analyses were carried out as strict intention-to-treat analyses. Per protocol analyses were performed to examine the robustness of the primary estimates. Analyses were adjusted for the center and clusters of siblings of twin pairs. Clinical characteristics of infants in the 2 nCPAP weaning groups were presented as means with standard deviations for continuous, normally distributed variables, as medians and interquartile ranges (IQRs) for continuous non-normally distributed variables, and as numbers and percentages for categorical variables.

The associations between the interventions and the continuous outcomes were analyzed using a random-effects mixed model regression to account for clustering within the twin pairs who were included in the study. The analyses were also adjusted for the study center. The adjusted mean differences were presented with 95% confidence intervals. The association between the interventions and the dichotomous outcomes were analyzed using a generalized linear model for the binomial family accounting for clustering and adjusted for the study center (the binreg command). The results of these analyses were presented as risk differences with 95% confidence intervals. Furthermore, we performed prespecified subgroup analyses by gestational age. All analyses were conducted using Stata, version 13 (StataCorp).

Of the 611 infants born who were before 32 weeks of gestation and admitted to 1 of the 6 participating NICUs between September 2012 and December 2016, 428 infants (70.0%) were eligible and 372 infants (60.9%) were randomized. A total of 354 infants had adequate data for the analysis of the primary outcome (Figure). The clinical characteristics were similar in the 2 groups (Table 1).

There was no statistical significant difference in mean (SD) weight gain velocity from the date of randomization to a PMA of 40 weeks between the sudden wean and pressure wean group (22 [6] g/kg/day in both groups) (Table 2). There was no statistical significant difference between the sudden wean and pressure wean group in any of the secondary outcomes (Table 2). A per protocol analysis of the 171 infants in the sudden wean group and the 173 infants in the pressure wean group who followed the protocol showed similar results (eTable in Supplement 3).

A total of 58 infants (16%) were born with a gestational age younger than 28 weeks and 296 infants (84%) were born with a gestational age between 28 and 31 weeks and 6 days (Table 3).

In this subgroup of extremely preterm infants, there was no statistical significant difference between the sudden wean and the pressure wean group in mean (SD) weight gain velocity from the date of randomization to a PMA of 40 weeks (19 [6] g/kg/day vs 19 [8] g/kg/day).

Furthermore, there was no difference between sudden wean and pressure wean in any of the secondary outcomes except for successful wean during the first attempt, in which significantly more infants in the pressure wean group were successfully weaned during the first attempt (41% vs 11%; risk difference, 31%; 95% CI, 13%-50%) compared with the sudden wean group (Table 3).

In this subgroup of very preterm infants there was also no statistical significant difference between the sudden wean and pressure wean group in mean (SD) weight gain velocity from the date of randomization to a PMA of 40 weeks (23 [6] g/kg/day vs 23 [5] g/kg/day). There was no difference in any of the secondary outcomes (Table 4).

In the pressure wean group 1 infant received a diagnosis of nasal septum necrosis. No other adverse events were observed.

We performed this randomized clinical trial to help resolve the uncertainty of how to optimally wean infants born before 32 weeks of gestation from nCPAP. To our knowledge, this is the largest RCT to date on nCPAP weaning, and we found no difference between sudden wean and pressure wean with respect to weight gain velocity, nCPAP and oxygen duration, length of stay in the NICU, or any of the other secondary outcomes that were all related to growth and respiratory support. In the analysis that was stratified by gestational age, more infants who were born before 28 weeks of gestation were successfully weaned from nCPAP during the first attempt in the pressure wean group compared with the sudden wean group without any effect on the secondary outcomes, including the total duration of nCPAP or supplementary oxygen treatment.

The results of this study agree with findings of what to our knowledge is the only other RCT that compares sudden wean and pressure wean in very preterm infants. This study, by Amatya et al,³¹ had a sample size of 68 infants born between 26 to 32 weeks of gestation and used successful wean during the first attempt as a primary outcome. They also found no difference between sudden wean and pressure wean in markers of pulmonary morbidity; weight gain, nCPAP and oxygen duration, BPD, and the length of hospital stay. Pressure wean was associated with a higher rate of successful wean during the first attempt compared with sudden wean. This differs from our results, in which the higher rate of successful weans during the first attempt only applied to infants who were born before 28 weeks of gestation. The study by Amatya et al³¹ was not powered to perform a stratified analysis.

A successful wean during the first attempt is an outcome that is highly dependent on the readiness and failure criteria used, in particular nCPAP pressure. The decision on when to wean preterm infants from nCPAP has limited evidence.^19,21-23,31 We used modified inclusion and failure criteria from the RCT by Todd et al.²¹ Because 4 cm H₂O was not used in our NICUs before the study, none of the infants were thought to be treated with this pressure at the time of inclusion. However, 4 infants were. They were all born after 28 weeks gestation. The results remained essentially unchanged after excluding those infants in secondary analyses.

The finding of a higher rate of successful weans during the first attempt in infants born before 28 weeks of gestation in the pressure wean group could indicate that our criteria may not be entirely appropriate for sudden wean in this subgroup of infants, because more infants in the sudden wean group had to recommence nCPAP. In the pressure wean group, the nCPAP was only discontinued if the infant had demonstrated that the gradual reduction in nCPAP pressure was tolerated; otherwise the nCPAP pressure was increased, indicating that the infants tolerated nCPAP discontinuation. Thus, if, for example, a lower nCPAP pressure was chosen as a readiness criterion for discontinuation, the number of infants weaned during the first attempt would probably have been higher in the sudden wean group.

We used weight gain velocity as a surrogate measure of the associated long-term health outcomes, neurodevelopment and pulmonary morbidity.^18,28,32-35 Inappropriate weaning from nCPAP may cause atelectasis, increased work of breathing, and episodes of desaturation and bradycardia, and animal studies have shown intermittent hypoxic episodes to suppress growth hormone and decrease postnatal weight gain.^36-38 A decreased weight gain velocity was therefore considered a biologically meaningful measure of inappropriate nCPAP weaning.

Previous randomized clinical trials of nCPAP weaning used either a successful wean during the first attempt or duration of nCPAP treatment as primary outcomes.^21-25,31 However, these outcomes may be influenced by the study design. Thus, the risk of self-fulfilling results would exist if, for example, nCPAP pressures prompting the first wean after randomization were too high for any infant to cease nCPAP during the first attempt of sudden wean, while infants may still cease during the first attempt by pressure wean. In fact, we had no infants who were randomized to sudden wean that ceased nCPAP at first attempt if enrolled at the upper limit of nCPAP pressure for participation. Thus, a successful wean during the first attempt is an outcome that depends on the readiness and failure criteria used. Days in nCPAP may also be influenced by the study design. The fact that nCPAP pressure had to be gradually reduced by 1 cm H₂O per day could contribute to an inherently longer duration of therapy in infants who were randomized to a gradual wean in studies with a low nCPAP pressure at inclusion. Therefore, we opted to present both a successful wean during the first attempt and nCPAP duration as secondary outcomes.

We were unable to mask the health care clinicians to the intervention. However, the possible effect on outcome was decreased by the use of strict criteria for weaning, and the weaning orders that were given by the tending physician at morning rounds rather than the bedside nurse. Furthermore, the randomization sequence was only known by the independent developer of the web-based randomization program, and the health care clinicians had no means of predicting the particular nCPAP weaning strategy for a particular infant, thus securing allocation concealment.

The loss to follow-up was modest and there were similar dropout rates in the groups. Furthermore, no apparent pattern in clinical characteristics distinguished infants with no primary outcomes at a PMA of 40 weeks of gestation from those who had. The results that were related to secondary outcomes were similar regardless of whether infants with missing primary outcome information were included in the analyses. Therefore, it seems unlikely that bias on this account could affect the validity of the results.

The presence of multiple gestations in the study population may violate the assumptions of the most commonly used statistical methods.³⁰ To account for this potential correlation between twins and potential false-positive findings, we used cluster analyses.³⁰ However, taking clustering into account in the analyses revealed essentially no change in the results. We used inclusion criteria that allowed for the participation of most of the infants who were born before 32 weeks of gestation, including infants who were born from multiple gestations and for whom only a few exclusion criteria applied, which may increase the generalizability of our results to most very preterm infants treated with nCPAP.

Based on the findings in this RCT, sudden wean and pressure wean from nCPAP could be considered equally good weaning strategies from nCPAP with respect to weight gain velocity and several other measures of growth and respiratory support. Pressure wean may be preferred in infants who are born before 28 weeks of gestation, as they had a higher rate of successful weans during the first attempt.

Accepted for Publication: June 5, 2018.

Corresponding Author: Christina Friis Jensen, MD, Child and Adolescent Medicine, Aarhus University Hospital, Palle Juul-Jensens Boulevard 99, 8200 Aarhus N, Denmark (christina.friis.jensen@clin.au.dk).

Published Online: July 23, 2018. doi:10.1001/jamapediatrics.2018.2074

Author Contributions: Dr Jensen (principal investigator) had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Jensen, Ebbesen, Cipliene, Johansen, Nielsen, Henriksen.

Acquisition, analysis, or interpretation of data: Jensen, Sellmer, Ebbesen, Johansen, Hansen, Nikitina, Petersen, Henriksen.

Drafting of the manuscript: Jensen, Sellmer, Ebbesen, Cipliene.

Critical revision of the manuscript for important intellectual content: Jensen, Sellmer, Ebbesen, Johansen, Hansen, Nielsen, Nikitina, Petersen, Henriksen.

Statistical analysis: Jensen, Ebbesen, Henriksen.

Obtained funding: Jensen, Ebbesen, Cipliene, Nikitina, Henriksen.

Administrative, technical, or material support: Jensen, Sellmer, Ebbesen, Hansen, Nielsen.

Supervision: Sellmer, Ebbesen, Johansen, Hansen, Nielsen, Henriksen.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was funded by Health, Aarhus University, and the Family Hede Nielsen’s Foundation.

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors.

Meeting Presentation: Preliminary results of this study were presented at the joint European Neonatal Societies; November 1, 2017; Venice, Italy.

Additional Contributions: We thank the participating families as well as the nurses and physicians in the participating neonatal units; they provided invaluable contributions in relation to study implementation and data registrations. These individuals were not compensated for their contributions.