Abstract
BACKGROUND:
There is no consensus on how to wean infants from Nasal Continuous Positive Airway Pressure (NCPAP). We hypothesized that ceasing NCPAP abruptly would decrease the duration required, compared with a gradual wean.
METHODS:
This retrospective chart review included preterm infants requiring NCPAP for over 48 hours. Cohort1 weaned NCPAP by cycling on and off, while cohort 2 ceased NCPAP abruptly. The primary outcome was total days on NCPAP. Secondary outcomes included rate of bronchopulmonary dysplasia, weight gain, duration of hospital stay, and compliance with the use of stability criteria.
RESULTS:
81 infants met inclusion criteria in cohort one, and 89 in cohort two. Median days on NCPAP were 17.0 and 11.0 days, respectively, not significant. There was no significant difference in secondary outcomes.
CONCLUSIONS:
There was no significant association between the two NCPAP weaning protocols and the outcomes studied.
Introduction
Preterm infants frequently receive respiratory support for Respiratory Distress Syndrome, apnea of prematurity and Broncho Pulmonary Dysplasia (BPD) [1]. Nasal Continuous Positive Airway Pressure (NCPAP) has been used in preterm infants as a mode of respiratory support since the 1970s and is now widely used in Neonatal Intensive Care Units (NICU) [2–4]. Early use of NCPAP is associated with decreased incidence of pulmonary [5, 6] and non-pulmonary morbidities associated with invasive ventilation [7]. However, there are risks associated with NCPAP, including nasal trauma and pneumothorax [8]. On the other hand, early weaning off NCPAP can lead to atelectasis, apnea and bradycardia, leading to failure to wean and prolonged use of NCPAP, and at times, the need for intubation with subsequent mechanical ventilation and prolonged oxygen use [9].
While NCPAP has been used extensively across NICUs, and the benefits and risks have been studied, surveys of neonatologists highlight a lack of consensus regarding the optimal method and timing of weaning NCPAP [10]. Current weaning methods include abrupt removal of NCPAP, gradual increase in time off NCPAP (cycling), gradual reduction of NCPAP pressure with or without oxygen supplementation, transition to high-flow nasal cannula, or a combination of any of these methods [11]. In one study, removing infants < 30 weeks gestational age (GA) from NCPAP using set stability and failure criteria improved patient outcomes like NCPAP duration, bronchopulmonary dysplasia, and hospital stay [12]. This study showed that infants who were taken off NCPAP abruptly, after meeting criteria, with a view to stay off NCPAP had better outcomes when compared to infants who were cycled on and off NCPAP either to room air or to oxygen.
After a review of the literature, including the studies mentioned above, a new protocol was introduced in our unit, where infants were taken off NCPAP abruptly, with a view to stay off, once they fulfilled set stability criteria (Appendix 1).We hypothesized that ceasing NCPAP abruptly at standard criteria would decrease the duration of time that the preterm infant required NCPAP when compared with a gradual wean by cycling off. Our secondary outcomes were the incidence of BPD, the amount of weight gain, the total duration of hospital stay, and the corrected GA when the infant was finally off NCPAP.
Methods
Study design
We conducted a retrospective chart review of infants born between 240 and 326 weeks GA at the Ottawa Hospital General Campus (TOHGC) NICU. Research Ethics Boards at both Children’s Hospital of Eastern Ontario (CHEO) and TOHGC approved this study (Ethics review letter# No.18/77X).
Setting
The TOHGC NICU is a level III NICU that is the primary regional perinatal center for Eastern Ontario, with an active referral network and high-risk delivery service. The unit has a total capacity of 24 infants. There are approximately 800 babies admitted to the NICU each year, approximately 170 of which are born between 240 and 326 weeks GA. In some cases, infants are transferred to the Children’s Hospital of Eastern Ontario (CHEO) before they are finally discharged from the hospital. The CHEO NICU primarily caters to outborn infants who need surgical and other pediatric subspecialist services from its catchment area. The same group of neonatologists manages both NICUs with consistent care protocols. For this study, charts of infants transferred to the CHEO NICU were additionally reviewed for the time period that they were at CHEO. These infants were either discharged directly from CHEO or returned back to TOHGC NICU for further care.
Prior to March 2013, our NICUs would wean infants from NCPAP by cycling between NCPAP and nasal prong trials with increasing lengths of time off of NCPAP. The abrupt wean protocol (Appendix 1) was introduced in the unit with the intention to decrease the number of days on NCPAP and improve patient outcomes. In this protocol, when the infant meets set stability criteria (Fig. 1), NCPAP is taken off and the infant remains in room air or oxygen with nasal prongs with the plan to stay off NCPAP permanently. The majority of the babies were taken off NCPAP the morning after meeting criteria after discussion during the team rounds. A few of the larger and more vigorous infants were taken off NCPAP immediately when they met the stability criteria. If the infant fails without NCPAP support by meeting set failure criteria (Fig. 2), the infant is placed back on NCPAP for a minimum of 48 hours. After that time on NCPAP, the infant is reassessed to see if criteria are again met to reinitiate a second abrupt wean. If the infant fails three attempts at abrupt wean, a gradual wean protocol is initiated (Appendix 2), which is the same as the NCPAP cycling weaning method used before implementation of the new protocol.
Stability criteria.
Failure criteria.
All eligible infants were identified using the NICU admission register at TOHGC NICU. Inborn infants of 240 to 326 weeks GA admitted to TOHGC NICU between March 1, 2012 and February 28, 2013 (cohort one) and between June 1, 2013 and May 31, 2014 (cohort two) were included in the study. The Infants should have required NCPAP support for at least 48 consecutive hours. The exclusion criteria included (1) infants who died or were transferred to another hospital, other than CHEO, before they could be weaned off NCPAP and (2) infants with chromosomal abnormalities or severe congenital anomalies, including congenital heart disease (excluding PDA, small ASD, and small VSD), neurological malformations, chest wall or airway abnormalities, or lung hypoplasia. Infants who were admitted to the NICU in the three months immediately after the implementation of the protocol were excluded from the study, because of possible inconsistent practices immediately after the protocol was introduced and caregivers were educated.
The primary research objective was to evaluate the effect of the abrupt wean on the total number of days on NCPAP among preterm infants born at 240 to 326 weeks GA. The secondary research objectives were to evaluate the effect of the abrupt wean on the incidence of BPD between the two cohorts; BPD was defined as oxygen or non-invasive ventilation requirements at greater than or equal to 36 weeks corrected gestational age [13], the amount of weight gain (g/kg/day), the total duration of hospital stay (days), and the corrected GA when the infant was finally taken off and stayed off NCPAP completely. We also examined the compliance with the abrupt wean protocol as defined by meeting all the set stability criteria before taking the infant off NCPAP.
Data collection
After identification, all patient charts were retrospectively reviewed in detail by the first author (ZP), collecting de-identified data using chart review study forms. Each patient was assigned a unique study number and no identifying data or patient contact details were collected.
Demographic and descriptive data collected in-cluded birth weight, GA (based on best obstetrical assessment), sex, Apgar scores, mode of delivery, antenatal steroids, pregnancy-induced hypertension/preeclampsia, small for gestational age status, and use of surfactant. Adequate antenatal steroids were considered two doses of betamethasone 12–24 hours apart. Small for gestational age was defined as birth weight less than 3rd percentile by sex, as per Fenton growth charts for preterm infants. In addition, we collected data on whether the infant met all the stability criteria for an abrupt wean trial prior to initiation, weight and corrected GA at initiation of abrupt wean, and details regarding abrupt wean failure if relevant. The duration of invasive ventilation and NCPAP days were calculated based on calendar days, even if it lasted for few hours. Other data collected were mode and duration of respiratory support, incidence of BPD, total days in NICU, amount of weight gain until discharge, incidence of sepsis/pneumonia, necrotizing enterocolitis (NEC; defined as modified Bell criteria stage 2A or greater) [14], Intraventricular hemorrhage [15], retinopathy of prematurity (ROP) needing treatment, patent ductus arteriosus (PDA) (confirmed on echocardiogram), discharge on home oxygen, and mortality (defined as death prior to discharge home). These were the most commonly reported variables in other studies, given their potential to affect CPAP duration or the incidence of BPD.
Sample size
In this study, our ultimate analysis aim was to be able to fit multivariable regression models for each outcome, with meaningful covariate adjustment of baseline patient demographics and other clinical variables. Covariate adjustment via multivariable modeling was done to lessen the effects of the confounding bias and to get a more precise estimate of group effect. On the basis of the 15:1 n per predictor rule-of-thumb for modeling a continuous/count outcome (i.e. total days on NCPAP), our initial sample size target was 188, which would allow 10 degrees-of-freedom for covariate adjustment and anticipates up to 20% of observations being excluded from modeling due to missing data ([15×10]/0.8 = 188) [16]. To maximize estimation precision of our main predictor variable of interest (i.e. abrupt wean protocol), we also sought to review a relatively balanced number of infants between cohort one and cohort two.
Statistical analysis
Initially, descriptive statistics was used to characterize the key baseline patient characteristics and study outcomes. We then fit variations of multivariable generalized linear regression models according to the type/distribution of each respective study outcome, with protocol (post vs. pre) as the main predictor variable of interest included in each model. For our primary outcome, number of NCPAP days, a quasi-Poisson model [17] was fitted as some evidence of over-dispersion was found during initial fitting of the data with a Poisson model. From this model, an adjusted rate ratio and 95% confidence interval (95% CI) for the protocol’s association with the outcome was estimated. Multivariable logistic regression was applied for BPD. From this model, an adjusted odds ratio and 95% CI for protocol was estimated. Finally, for three other secondary outcomes (weight gain, number of hospital days, corrected GA when finally off NCPAP), ordinary least squares (linear regressions) were fitted. From these models, adjusted estimates (regression beta coefficients) and 95% CI for protocol were estimated. For all regressions except for one (the BPD model), a common set of covariates was applied. These were: GA, sex, chorioamnionitis, antenatal steroids, surfactant, sepsis/pneumonia, and PDA. In the case of the logistic regression, very low number of events (i.e. BPD cases) limited ability for comprehensive covariate adjustment inappropriate thus only gestational age was chosen as the only covariate in that particular model. All analyses were performed using R version 3.5 [18].
Results
Participants
Cohort one included 81 infants who met the study criteria, out of 159 240 to 326 weeks GA infants admitted during the time period. Cohort two included 89 infants who met the study criteria, out of 195 infants admitted during the time period. Median GA for cohort one was 28.3 weeks; for cohort two, it was 28.4 weeks. The median birth weights for cohort one and two were 1060 and 1090grams, respectively. Table 1 shows the descriptive data; there were no significant differences between the cohorts.
Descriptive Data (Cohort one vs. Cohort two)
Descriptive Data (Cohort one vs. Cohort two)
IQR: interquartile range, PDA: Patent ductus arteriosus, IUGR: intrauterine growth retardation, PIH: Pregnancy induced hypertension, DART: Dexamethasone: A Randomized Trial study, NEC: Necrotizing enterocolitis, IVH: Intraventricular hemorrhage, ROP: retinopathy of prematurity, P: P Value, severe NEC: NEC Stage 2or more, Severe IVH: IVH grade 3 or 4, severe ROP: ROP needing treatment, ‘N’ is different in few of the columns as some of the data was missing.
In terms of the primary study outcome, the median (IQR) days on NCPAP for cohort one was 17.0(7.0, 35.0) days; for the cohort two, it was 11.0 (6.0, 31.0) days (Table 2). Figure 3 shows the spread of the data. For NCPAP days, we found a non-significant adjusted rate ratio of 0.84 (95% CI: 0.69, 1.03, p = 0.09) associated with cohort two (compared to cohort one).
Outcome data (Cohort one vs. Cohort two)
Outcome data (Cohort one vs. Cohort two)
IQR: Interquartile range, NCPAP: Nasal Continuous Positive Airway Pressure, BPD: Bronchopulmonary dysplasia, GA:Gestational age, NICU: Neonatal intensive care unit, P: P value from multivariable linear regression analysis, ‘N’ is different in few of the columns as some of the data was missing.
Primary outcome-Total number of days on NCPAP.
In terms of secondary outcomes, BPD was ob-served in 21.0% of infants in cohort one and 33.7% in cohort two; the adjusted OR for cohort two was 1.36 (95% CI: 0.39, 4.70, p = 0.63). For weight gain, the median (IQR) weight gain was 18.5 (12.0, 22.8) g/d for cohort one, and 21.9 (14.6, 26.5) g/d for cohort two; the adjusted estimate (beta coefficient) for cohort two was 2.21 (95% CI: –0.68, 5.11, p = 0.14). The corrected GA when the infants were consistently on room air, no longer requiring any supplemental oxygen, had a median (IQR) of 32.8 (31.1, 34.7) weeks for cohort one and 32.3 (30.7, 36.0) weeks for cohort two, as some of them required intermittent oxygen by nasal cannula during the NCPAP wean. The median (IQR) duration of hospital (NICU) stay was 45.0 (23.0, 83.0) days for cohort one and 51.0 (30.0, 85.0) days for cohort two; the adjusted estimate for cohort two was 2.50 (95% CI:–4.16, 9.15, p = 0.46). Finally, the corrected GA when being finally taken off NCPAP had a median (IQR) of 32.1 (30.7, 33.4) weeks for cohort one and 31.7 (30.1, 34.1) weeks for cohort two. The adjusted estimate for cohort two was –0.49 (95% CI:–1.23, 0.26, p = 0.20). No statistically significant associations were noted.
Confirmation of compliance to the set stability criteria was limited to what data was recorded in the medical records. Some of the set stability criteria had inadequate documentation. Assessing the compliance with the abrupt wean protocol revealed compliance satisfactory for most of the stability criteria except one. The criteria “tolerates time off NCPAP during care” was poorly documented, so it is unclear whether it was adhered to. Similarly, the failure criteria also had inadequate documentation. Out of the 89 infants in cohort two, 77 infants (86.5%) were taken off NCPAP at set stability criteria as per the protocol. Out of 77 infants who underwent abrupt wean, 25 (32.5%) were successful at the first attempt itself, while the remaining 52 (67.5%) infants had one or more failures. The remaining infants did not undergo an abrupt wean, but instead were gradually weaned based on the individual discretion of the treating physician.
We conducted this retrospective chart review to determine if our abrupt wean protocol of changing the NCPAP weaning practices decreased the total number of days infants were on NCPAP and, secondarily, improved several other outcome parameters. We determined that ceasing NCPAP abruptly at set criteria (when compared with weaning gradually by cycling off), in preterm infants born between 240 and 326 weeks, did not result in any significant difference in the total number of days on NCPAP, rates of BPD, amount of weight gain, total duration of hospital stay, or corrected GA when the infant was finally taken off NCPAP completely.
As noted, our study did not show any significant difference in the number of NCPAP days between the cohorts. This finding is consistent with some earlier studies [11, 20]. However, a comparison of studies is difficult since each study had different NCPAP weaning methods and thresholds. A clinical trial by Eze et al. [19] showed that number of days on NCPAP, success at weaning, the secondary outcomes of BPD, severe ROP (≥stage 3), periventricular leukomalacia, and length of stay were not different between cycling on and off or abrupt removal of NCPAP in infants born < 31 weeks GA. Another prospective study by Jensen et al. [20] showed that there was no difference in the number of NCPAP days after the infants were randomized, but this study compared abrupt wean against pressure wean in infants born < 32 weeks GA. Similarly, Amatya et al. [11] observed that the total days on NCPAP in the abrupt wean group did not differ from a gradual pressure-wean group. On the other hand, Todd et al. [12] in a multicenter trial found that abruptly removing NCPAP in preterm infants < 30weeks GA decreased total duration on NCPAP when compared to incremental cycling off NCPAP either with or without nasal prongs. The infants in this trial had significant differences in baseline characteristics –such as sex and Apgar scores –that could have influenced the study results. Further, other confounding variables like antenatal steroids, PDA, and chorioamnionitis were not reported.
Our study also did not show any significant difference in any of the secondary outcomes. BPD, was observed more frequently among the abrupt wean group, but this finding did not reach the level of significance (p = 0.63). Similarly, in the study by Eze et al. [19], BPD was observed in 35% of abrupt wean group compared with 28% of cycling off NCPAP group; this was not a significant difference. Likewise, Tang et al. [21] also found no significant difference in BPD (p = 0.20) between the gradual wean group and abrupt wean group (13% and 27%, respectively). However, Todd et al. [12] observed significantly less BPD among the abrupt wean group (12.5 versus 42%, p = 0.01).
We were interested in looking at the pattern of weight gain as preterm infants with immature lungs might expend more calories on work of breathing, more so when they are weaned off NCPAP early and abruptly. However, our study found that the method of NCPAP wean did not influence the amount of weight gain. This was consistent with the findings of Tang et al. [21] and Broom et al. [22]. The duration of hospital stay was also similar between our cohorts, as was observed by Tang et al. [21] and Eze et al. [19]. Meanwhile, Todd et al. [12] observed a significantly shorter duration of stay in the NICU among the abrupt wean group.
In a systemic review, Amatya et al. [23] concluded that optimal corrected GA for the successful wean is usually achieved at 32 to 33 weeks, and was inversely related to the GA at birth, with the most premature babies successfully weaning at later corrected GA. The median corrected GA when finally taken off NCPAP was similar between our two cohorts, 32.1 weeks for the cohort one and 31.7 weeks for the cohort two, comparable to the observed value in the systemic review.
There are only a few studies on the NCPAP wea-ning methods in preterm infants and our study adds to the literature. We recognize limitations to our retrospective chart review. There was occasional incomplete data and also the chart review was done by a single person. In addition, this is a relatively small, single center study that limits generalizability. One advantage is that the data was collected over a relatively short time span of 27 months and there were no significant changes in other protocols practiced in the unit. In examining whether babies actually met all stability criteria before they were taken off NCPAP, we found variation in adherence to the new NCPAP weaning protocol. In few cases, there was poor or incomplete documentation of whether all criteria were met and in others it was clear that they were not, and management was subject to the individual neonatologist’s decision. In addition, some of the criteria, for example “tolerates time off NCPAP during care”, could be quite subjective. Similarly, there was poor documentation of failure criteria among cohort two and we had not analyzed the adherence of cohort two to the failure criteria. These limitations could affect the conclusion of our study if, for example, the infants were taken off NCPAP before they had met the set criteria or put back on NCPAP before they had met the failure criteria. We did attempt to eliminate the inconsistency in adherence to protocol by excluding the initial three months after the new protocol implementation from the data collection. In spite of these shortcomings, both cohorts are rather comparable in their baseline demographic profiles and we had reasonable number of subjects in both arms of the study. Future studies should have more objective eligibility and failure criteria with a rigorous method of documentation so that the groups are truly distinct.
Conclusion
There was no statistically significant difference in the number of NCPAP days, rate of BPD, amount of weight gain, total duration of hospital stay, or corrected GA when finally taken off NCPAP among preterm infants born at240 to 326 weeks who were weaned from NCPAP either by cycling on and off or by abrupt discontinuation. As discussed, early weaning off NCPAP can lead to atelectasis, apnea and bradycardia, leading to prolonged use of oxygen, invasive or non-invasive ventilation; however prolonged unnecessary NCPAP has its own risks. Varying weaning methods like abrupt wean from a set NCPAP, gradual wean of the set pressure, cycling on and off NCPAP with or without high flow or low flow nasal cannula were used in the reported studies, in addition to different criteria for the readiness to wean and failure of wean. Given the lack of consensus or best evidence on when and how to best discontinue NCPAP, a large prospective multicenter trial with clear objective criteria for the readiness to wean and failure of wean would be beneficial to determine if there is a best practice for the weaning of NCPAP and improving outcomes in preterm infants.
Footnotes
Acknowledgments
None
Conflict of interest
The authors declare no competing financial interests. No other funding received for this study.
Funding
No funding was received for this study.
Ethics approval and consent to participate
Research Ethics Boards at both the Children’s Hospital of Eastern Ontario (CHEO) (protocol #20180337) and the Ottawa Hospital (protocol #20180480-01H) approved this study (No.18/77X). The authors also confirm that the study was performed in accordance with the Declaration of Helsinki.
Author Contributions
ZP contributed to the conception of the study, collected data, drafted and revised the manuscript, GM conceived the study and revised the manuscript, KT did the analysis and interpretation of the data and revised the manuscript, NP conceived the study and revised the manuscript, SL conceived and supervised the study and revised the manuscript. All the authors have approved the final version of the manuscript for publication.
Appendix 1
Abrupt wean (Phase 4) Algorithm: Discontinuation of non-invasive respiratory support.
Note: Phase 1: Initiation, stabilization and maintenance of invasive ventilation. Phase 2: Active Weaning. Phase 3: Initiation and stabilization of non-invasive support. Phase 4: discontinuation of non-invasive ventilatory support by abrupt wean, and Phase 5: weaning by cycling on and off NCPAP, was used for patients with recurrent failure of Phase 4
Gradual wean (Phase 5) algorithm for patients with recurrent failure of Phase 4
