Seattle-PAP trial: A Study to Evaluate the Efficacy of Seattle-PAP for the Respiratory Support of Premature Neonates

Background: At birth, the majority of neonates born at <30 weeks of gestation require respiratory support to facilitate transition and ensure adequate gas exchange. Although the optimal approach to the initial respiratory management is uncertain, the American Academy of Pediatrics endorses noninvasive respiratory support with nasal continuous positive airway pressure (nCPAP) for premature neonates with respiratory insufficiency. Despite evidence for its use, nCPAP failure, requiring intubation and mechanical ventilation is common. Recently, investigators have described a novel method to deliver bubble nCPAP, termed Seattle-PAP. While preclinical and pilot studies are encouraging with regard to the potential value of Seattle-PAP, a large trial is needed to compare Seattle-PAP directly with the current standard of care for bubble nCPAP (Fisher Paykel-CPAP; FP-CPAP). Methods: We designed a multicenter, non-blinded, randomized controlled trial that will enroll 230 premature infants (220/7 to 296/7 weeks of gestation). Infants will be randomized to receive Seattle-PAP or FP-CPAP. The primary outcome is respiratory failure requiring intubation and mechanical ventilation. Secondary outcomes include measures of short and long-term respiratory morbidity and cost effectiveness. Discussion: This trial will assess whether Seattle-PAP is more efficacious and cost effective than FP-CPAP in real-world practice among premature neonates.

airway pressure (nCPAP). (6) The American Academy of Pediatrics endorsed the use of nCPAP among premature neonates with respiratory distress. (7) This recommendation is based on data indicating lower rates of bronchopulmonary dysplasia (BPD) in infants treated with nCPAP than are observed with more invasive modes of ventilatory support. (8)(9)(10) However, nCPAP failure, requiring intubation and mechanical ventilation in preterm infants, is common, with failure rates exceeding 50% in large clinical trials. (8)(9)(10) In developed countries, CPAP failure is associated with greater morbidity; in developing countries, CPAP failure is associated with greater mortality. (11,12) More recently, bubble nCPAP has re-emerged as a potential strategy to address high nCPAP failure rates. (13) Although data are limited, neonates on bubble nCPAP had lower incidence of respiratory failure (tracheal intubation and mechanical ventilation) than did infants supported on ventilatorderived CPAP. (12,14) Despite evidence that bubble nCPAP may be advantageous for preterm infants, optimal delivery for newborns with respiratory distress is unknown. To address high failure rates associated with CPAP among preterm neonates, investigations of novel strategies to deliver more effective bubble nCPAP are warranted. (15) One potential strategy is with use of Seattle-PAP. (12,13,16,17) In general, bubble nCPAP delivery systems consist of six primary components: sources of air, oxygen, and other breathing gases, a blender to mix these gases together, a gas heater and humidifier, inspiratory and expiratory tubing (breathing circuit), a patient interface (e.g., bi-nasal prongs or nasal mask), and a pressure generator (e.g., bubbler device). The design of a bubbler device can take many forms, but is generally comprised of a tube with its distal end submerged in a body of water. The airway pressures provided by the system are determined largely by the depth of the distal end of the bubbler tube below the surface of the water (5 cm below surface = 5 cm H2O). In conventional bubble nCPAP systems, the bubbler tube points straight down (perpendicular to the surface of the water), which we define as 0°. With Seattle-PAP, the bubbler tube points initially straight down with a transition to a horizontal, then upward sloping section, which we optimized at 135° (Figure 1).
Preclinical evidence demonstrated that this 135° modification provides fluctuations in airway pressures, including lower frequencies pressure oscillations than those created by conventional systems. (17,18) Moreover, a recent study showed that prematurely-born neonates (mean gestational age of 29 weeks) with minimal parenchymal lung disease exhibited lower effort of spontaneous breathing using Seattle-PAP than with conventional bubble nCPAP (Fisher Paykel; FP-CPAP). (13) While the results of preliminary studies are encouraging, safety and efficacy data among smaller, more immature preterm infants are lacking.
If the Seattle-PAP arm proves superior to conventional bubble nCPAP, the likely benefits would be large, with clear applications in both the developed world and in low and middle income countries (LMIC). However, to recommend Seattle-PAP over the current standard of care for bubble nCPAP (FP-CPAP), a large comparative trial is needed. Additionally, since respiratory failure is associated with higher daily costs related to more intensive monitoring and personnel requirements, (19) Seattle-PAP may result in lower overall treatment costs if it prevents respiratory failure. Similarly, since infants failing CPAP have longer lengths of hospital stay than those successfully supported on CPAP, total costs should be lower if use of Seattle-PAP is associated with lower incidence of CPAP failure than is use of FP-CPAP. (20) To estimate the cost effectiveness of Seattle-PAP, a formal economic evaluation ancillary to the proposed randomized controlled trial (RCT) is necessary. (21) The present report describes the rationale and design for an ongoing, randomized controlled trial (RCT) that aims to compare the effectiveness of Seattle-PAP versus conventional FP-CPAP in the prevention of respiratory failure of premature neonates 220/7 -296/7 weeks GA.
Hypothesis: We are testing the hypothesis that premature neonates supported by Seattle-PAP will have lower rates of respiratory failure (need for tracheal intubation and mechanical ventilation) than will neonates supported with conventional bubble CPAP (FP-CPAP).
Aims: The primary objective of this study is to compare the rates of respiratory failure from 72 hours post-delivery to 32 weeks' postmenstrual age (PMA) in neonates born at 220/7 to 296/7 weeks of gestational age (GA) who receive either Seattle-PAP or conventional bubble nCPAP (FP-CPAP).

Methods/design
Study Settings: This is a multicenter, non-blinded, RCT in premature neonates born at 220/7 to 296/7 weeks' GA. 1. Parental consent and/or legal guardian consent given to participate in this research study.
2. Preterm infants delivered at 22 -<30 weeks of completed gestation by best obstetrical estimate.
Exclusion criteria are as follows: 1. Infants with known major congenital cardiac (e.g., transposition of the great arteries), pulmonary (e.g., pulmonary and/or tracheal hypoplasia), or physiological (e.g., anencephaly, omphalocele, congenital diaphragmatic hernia) anomalies. For the purposes of this definition, common preterm cardiac issues such as patent ductus arteriosus (PDA) and patent foramen ovale (PFO) / atrial septal defect (ASD) will NOT be grounds for exclusion.
3. Infants born to mothers who are unable to give informed consent and/or who do not have a legal, surrogate guardian who can provide consent.
Recruitment: A member of the study team will approach parents of potentially eligible infants with threatened preterm delivery between 220/7 and 296/7 weeks' GA to offer study participation and to obtain written informed consent. Consent will be obtained by team members who have been trained in obtaining consent for clinical trials and who are familiar with trial protocol. Whenever possible, consent will be obtained by someone not directly involved in the clinical care of the infant. If obtaining consent is not possible during the antenatal period, as soon as possible following birth, a member of the study team will approach the infant's parents or legal guardians to obtain consent (up to 72 hours post-delivery). Parents or guardians of infants who are not yet eligible but are likely to become eligible (e.g., infants requiring mechanical ventilation who are likely to transition from mechanical ventilation to bubble nCPAP prior to 72 hours) will also be approached. A schedule of enrollment, interventions, and assessments is shown in Table 1. Recruitment began in March, 2017.
Randomization: Eligible infants enrolled are randomly assigned to one of two treatment arms (Seattle-PAP vs. FP-CPAP) ( Figure 2). The allocation sequence is generated using an online, computergenerated randomizer (https:/sealedenvelope.com), with a block size of 6, stratified by gestational age (27 -296/7 and 22 -<27 weeks' gestation). We did not stratify by site, as the treatment of premature neonates is based on shared guidelines throughout the NCH-NRN.(23) Multiple gestations (twins, triplets) are assigned to the same treatment arm.
Intervention: Allocated treatment is applied immediately after randomization. Infants whose condition cannot be maintained with the assigned method of noninvasive respiratory support will be intubated and the originally assigned intervention resumed after extubation. In the NCH-NRN, bubble nCPAP is given through nasal prongs/mask with an initial pressure of 5 to 6 cm H2O. As described above, the primary difference between the groups is the "bubbler" generating the pressure for the circuit (refer to Figure 1).
Blinding: Blinding of the allocated treatment is not feasible, as the mode of respiratory support is apparent to health care professionals and families.
Clinical and respiratory guidelines: Because the duration and type of respiratory support are critical end points, we have taken a number of steps to ensure that respiratory support is applied similarly to both groups: 1) NCH-NRN follows a respiratory algorithm for care (see Supplemental Figure); 2) thresholds for achieving the primary endpoint of respiratory failure are clearly defined (see below). In addition to standardized respiratory support, all aspects of neonatal management and treatment will be in accordance with local guidelines.(23) Internal audits to investigate maintenance and adherence to the guidelines are conducted on a routine basis.
Treatment Failure is defined as any of the following: 1. Tracheal intubation within 72 hours for surfactant administration after initiation of bubble nCPAP and then not extubated by 72 hours.
2. Tracheal intubation of the infant or support with biphasic CPAP (SiPAP) in the neonatal intensive care unit (NICU) after 72 hours and up to 32 weeks GA. As adjudicated by an independent party, tracheal intubation for non-respiratory issues after 72 hours and up to 32 weeks GA (e.g., surgery for retinopathy of prematurity) will not be considered a treatment failure.
3. The infant cannot sustain an oxygen saturation (SpO2) of at least 90%, despite noninvasive respiratory support of up to and including 8 cm H2O bubble nCPAP, and fraction of inspired oxygen (FiO2) greater than 0.40 for more than 1 hour. Primary Outcome measure: The primary outcome is treatment failure (as defined above). Secondary Outcomes: We will capture important secondary outcomes, including short and longerterm respiratory morbidities (Additional file 1: Table S1).
Economic Outcomes: With a birth to hospital discharge time horizon, cost-effectiveness analysis from the perspective of the health-care system perspectives was planned a priori and will be conducted alongside the RCT. From a health care sector perspective, cost-effectiveness analyses include formal health sector costs (medical) costs, such as those paid by a third-party such as government or a private insurer. (24) Sample Size: Using data from other recent, large neonatal randomized trials with similar populations, including the SUPPORT trial and the COIN trial,(9, 10) as well as contemporary data within the NCH-NRN, we estimate that the proportion of patients experiencing respiratory failure in the control arm (FP-CPAP) will be 50%. Based on preclinical work, we estimate the failure rate in the Seattle-PAP arm will be 30%. (13,17) Accounting for two interim analyses, 200 neonates are needed with a two-tailed type I error rate of 0.05 and a power of 80%.(25) Multiples will be randomized as a set to the same study arm, requiring an inflation of the estimate by 1.12 to allow for the design effect due to clustering. (26,27) Thus, the calculated sample size is 115 infants per treatment arm.

Data collection:
Clinical: With the exception of the data related to the screening log, all remaining clinical data will be obtained from the electronic medical records (Additional file 2: Table S2).
Economic: Routinely available costs of inpatient stay will be sourced from the hospital costing units.
Direct medical resource utilization will be ascertained through collection of itemized billing records and UB-04 forms, a uniform billing statement recommended by the National Uniform Billing Committee and utilized for reporting of hospital expenditures by third party payers including the Centers for Medicare and Medicaid Services (CMS). We will convert hospital-reported charges to costs by applying the appropriate CMS cost-center specific ratio of costs to charges. Total hospital costs will be the sum of the product of the number of days in each cost category and the calculated per diem costs. Physician professional fees for the initial hospitalization will be based on CMS reimbursement levels for each day of stay and non-bundled procedures.

Safety
Data and safety monitoring board (DSMB): We have assembled a data and safety monitoring board (DSMB) to protect study subjects and monitor the overall conduct of the trial. Prior to study commencement, the 4-person DSMB agreed to the following activities: 1) review the protocol and amendments; 2) participate in the development, finalization, and approval of the DSMB Charter; 3) recommend discontinuation of the trial for safety reasons; 4) recommend changes to the study protocol (amendments) for safety reasons; 5) evaluate emergent safety information, evaluate any risk, and identify any potential safety concerns for study patients; 6) request additional data not included in the report, if deemed necessary for effective safety monitoring; 7) communicate DSMB findings and recommendations (to stop, continue, or modify the study) to the site-PI (C.B.); 8) provide written minutes on an ongoing basis following scheduled and ad hoc meetings.
Adverse events and their relationship to study, severity, time of experience, expectation, actions taken to resolve the event, and final outcome will be recorded. All serious adverse events (SAEs) will be sent within 48 hours to the DSMB and local IRB. A SAE for this study is any untoward medical occurrence that is believed by the investigators to be causally related to the study intervention and results in any of the following: life-threatening condition (that is, immediate risk of death), severe or permanent disability, and prolonged hospitalization. SAEs occurring after a subject is discontinued from the study will not be reported, unless the investigators feel that the event may have been caused by the study device. All SAEs will be followed until satisfactory resolution is achieved or until the health care provider responsible for the care of the participant deems the event to be chronic or the patient to be stable. All expected and unexpected SAEs, whether or not they are attributable to the study intervention, will be reviewed by the site-PI (C.B.) to determine if there is a reasonable suspected causal relationship to the intervention. If the relationship is reasonable, SAEs will be reported to DSMB and local IRB for consideration.
Interim analysis: Interim analyses were conducted by the DSMB following enrollment of 25 and 110 infants, respectively. The analyses compared the two groups with respect to efficacy, safety, and futility. A Haybittle-Peto stopping guideline was set at P<0.001 for each interim analysis. The interim analyses were completed in June 2017 and January 2018. On the basis of these analyses, and on safety reviews conducted to date, the DSMB recommended that the trial continue without modification.
Duration of study: The projected study duration is 3 years, including 2½ years for subject recruitment.
Training of clinical staff: Comprehensive education and training was undertaken to ensure technical proficiency and protocol compliance at all sites. The site-PI (C.B.), study coordinator (J.N.), and nursing leadership on the study (J.L. and J.B.), ran a 'boot-camp' at the start of the study for all participating sites, with mandatory attendance for physicians, respiratory therapists, and nursing staff to accomplish the following: 1) review differences between Seattle-PAP and the conventional FP-CPAP device; 2) identify areas for device troubleshooting; 3) demarcate study goals and objectives; 4) review safety protocols, procedures, and guidelines for clinical and respiratory care. This effort was intended to optimize reproducibility and consistency. All centers received detailed written instruction on study protocol. The PI and study coordinator were available 24 hours/day to answer any questions or concerns.
Data management, processing, monitoring, and security: Data generated in this study will be appropriately documented and checked for validity and accuracy. Data will be entered into the database, then the data will be matched and checked for validity and accuracy by a second person before being endorsed for analysis. A record of all discrepancies and resolutions will be kept by the study coordinator (J.N.). Outlier data will be investigated. Data will be primarily managed using Research Electronic Data Capture (REDCap) software. (28) REDCap is a secure, web-based application designed to support data capture for clinical studies. All data with identifiers will be stored on firewallprotected secure servers. Study monitoring visits are performed by the sponsor and their representatives after the enrollment of 50, 100, 150, and 200 infants, and upon the closure of the study.

Statistical analysis
Clinical outcomes: Analyses will be performed using Statistical Analysis Software (SAS) Enterprise Guide version 7.15 (SAS Institute Inc., Cary, NC) using an 'as randomized' ('intention-to-treat') principle to compare the primary outcome between treatment arms. Continuous data will be expressed as means with standard deviations or as medians with ranges, whereas categorical variables will be expressed with frequencies and proportions. The primary outcome, and other binary outcomes occurring in at least 5% of patients, will be analyzed using Pearson chi-square tests. Less common binary outcomes will be compared using Fisher exact tests. Risk ratios and risk differences, along with their 95% Wilson score confidence intervals (CI), will also be calculated. Normally distributed, continuous outcomes will be compared using Student's t test, whereas the Wilcoxon ranksum test will be used to compare continuous outcomes with skewed distributions.
In additional analyses, log binomial regression models will be used to evaluate heterogeneity in the effect of treatment with Seattle-PAP on the primary outcome. Treatment effect heterogeneity will be explored across several clinical characteristics that are known to be associated with CPAP treatment failure: gestational age, birth weight, and exposure to antenatal corticosteroids. (15) Treatment effect heterogeneity will be tested by evaluating the significance of interactions between the factors of interest and treatment arm in log binomial regression models that include the factor, treatment arm, and their interaction. Effect estimates within subgroups defined by these factors will be expressed as risk ratios (RRs) with maximum likelihood-based 95 % confidence limits. Regression models will be estimated using generalized estimating equations in order to account for the inherent correlation expected with multiples. (27).
Economic outcomes: To inform whether it is cost-effective to incorporate FP-CPAP or Seattle-PAP into the existing health system, decision analysis will be constructed based on the primary outcome and associated hospital costs. Univariate and probabilistic sensitivity analyses will be conducted to test the impact of uncertainty in data. We will first compare the mean patient-level costs for each treatment arm, without consideration of effectiveness. Because of the anticipated skewed nature of cost data, we will model the logarithm of the mean costs directly, using a generalized linear model with a logarithmic link function and gamma distribution. (24,29) In addition to treatment assignment, covariates will be entered into this model to account for any differences in baseline prognostic indicators (e.g., exposure to antenatal corticosteroids) that are evident despite randomization. The model will take into account clustering among twins.
We will then determine the simultaneous outcome of cost and effect, or value for money, expressed as the incremental cost effectiveness ratio (ICER), which is calculated as the difference in mean cost per patient in the Seattle-PAP and FP-CPAP groups divided by the difference in the mean effect between the 2 groups. We will determine the statistical uncertainty in the joint distribution of costs and effects using nonparametric bootstrapping, in which we will draw 1000 samples of 220 infants, with replacement, from the study data set. (30,31) For each of the 1000 samples, we will calculate the mean cost, mean effect, and ICER. Results of this analysis will be presented using incremental cost-effectiveness plot and cost-effectiveness acceptability curves.(32) Parameter uncertainty, for variables such as price weights, will be estimated using deterministic sensitivity analysis, in which the results are generated again after varying the parameter through its plausible range. More recently, bubble nCPAP has re-emerged as a potential strategy to address high nCPAP failure rates. (13) Although neonates on bubble nCPAP had lower incidence of respiratory failure (tracheal intubation and mechanical ventilation) than did infants supported on ventilator-derived CPAP, optimal delivery of nCPAP to premature neonates remains unknown. (12,14,15). Recognizing the need for a more efficient bubble nCPAP system among preterm infants, (15) investigators designed and developed an alternative device, termed Seattle-PAP. (12,13,17,18) In preclinical studies with juvenile rabbits lavaged to produce pulmonary surfactant deficiency, (12,13,17,18) Seattle-PAP provides oscillations in airway pressures at lower frequencies than are observed with other devices. (17,18) The frequencies of airway pressure oscillations generated by Seattle-PAP are lower than are those generated by conventional bubble nCPAP. (17,18) While conventional bubble CPAP provides stabilization and distension of small airways and alveolar spaces, the range of fluctuations in airway pressure provided by Seattle-PAP may improve recruitment of and ventilation to the low ventilationperfusion compartments of the lung that contribute to hypoxemia. (12) These observations led to a recent study that compared Seattle-PAP and conventional bubble nCPAP (FP-CPAP) among premature infants with an average gestational age of 29 weeks. The authors observed, among 40 neonates, lower effort of breathing with Seattle-PAP than with FP-CPAP; however, the study was not designed to detect important clinical differences between the two devices. (13) To recommend Seattle-PAP over the current standard of care in bubble nCPAP, a larger comparative trial is needed. Thus, the present study is designed to compare the incidence of respiratory failure over the first weeks of life between the two groups. While the present study is not designed to determine differences in longer-term clinical outcomes (BPD), pre-specified secondary analyses are planned.
We acknowledge a number of challenges in the design of the trial. First, the potential benefits and risks of Seattle-PAP versus conventional bubble nCPAP (FP-CPAP) may not be expressed uniformly across infants born at all gestational ages. To that end, we determined a priori to perform subgroup analysis on infants born <27 weeks versus those born 27-<30 weeks of gestation. Second, our recruitment window provides enrollment up to 72 hours of life, irrespective of previous respiratory support (mechanical ventilation, CPAP) and likely over a wide range of acutely impaired lung function.
We recognize that this potential heterogeneity increases the risk of a Type II error. (36) Third, some infants allocated to Seattle-PAP may receive a brief period of FP-CPAP prior to randomization, which conceivably could affect interpretation of the results. We attempted to restrict the impact of this by excluding from the trial infants who have received more than 72 hours of FP-CPAP, which we felt to be the shortest window in which seeking parental consent would be feasible. Finally, blinding of the interventions, which would have been preferable from a study design perspective, was not feasible, due to the complexity of building and operating a system that disguised which device was providing respiratory support to the infant, yet allowed the clinical care team to monitor effective engagement of the infant with the pressure generating system.
The results of the present study will inform the design of larger, multicenter investigations to understand more fully the benefits and risks of bubble nCPAP in low and middle income countries (LMICs). Since CPAP failure in developing countries may result in neonatal death, potential use in LMICs would be even more attractive. (11,14) While previous investigators have described barriers to successful use of bubble CPAP in LMICs (fixation devices are bulky and cover much of the infant's face; interference with parental interaction and feeding; trauma to the nasal skin or septum; need for nursing vigilance to ensure that an adequate seal), others have shown that bubble nCPAP can be applied effectively in resource-limited settings. (14,37,38) The airway pressure oscillations generated with Seattle-PAP, including lower frequencies of pressure oscillations, are modestly different than are those generated with conventional bubble nCPAP. (17,18) Although preclinical and small clinical studies conducted to date have not indicated greater risk of adverse events associated with Seattle-PAP than with conventional bubble nCPAP, (13,17,18) the present trial will provide an opportunity to observe any unanticipated severe adverse events.
Use of CPAP among preterm infants is associated with reduced hospital stay (16), with savings exceeding $10,000 for every six neonates treated with CPAP.(39) Given the interrelatedness of health, availability of resources, and the constraints on healthcare budgets, as well as the substantial resource utilization by preterm neonates, improving health care status is likely to have important cost implications; thus, we will perform a formal economic evaluation ancillary to the proposed RCT with the goal of informing the practice of effective and efficient health care.

Potential impact
Seattle-PAP is a promising new bubble CPAP delivery system that may reduce the need for tracheal