Hazards to avoid in future neonatal studies of nasal high-frequency oscillatory ventilation: lessons from an early terminated trial

Objective To investigate whether nasal high-frequency oscillatory ventilation (nHFOV) started immediately after extubation of mechanically ventilated very low birth weight infants reduces the partial pressure of carbon dioxide at 72 h after extubation in comparison with nasal continuous positive airway pressure. This randomised controlled single-centre trial aimed to include 68 preterm infants at high risk of extubation failure. Results Implementation of the study protocol was feasible. However, from 2015 to 2017, only six patients could be recruited, leading to early termination of the trial. The slow recruitment was due to the introduction of new strategies to avoid endotracheal mechanical ventilation, which reduced the number of eligible infants. Moreover, the included infants failed their extubation more often than anticipated, thereby increasing the required sample size. Based on our single-centre experience, we provide information for study planning and discuss the specific requirements for future trial protocols on nHFOV. The extubation of high-risk infants into nHFOV could well be beneficial, but a multicentric approach is necessary to investigate this hypothesis. Trial Registration Clinicaltrials.gov NCT02340299, on 16 January 2015 Electronic supplementary material The online version of this article (10.1186/s13104-019-4268-2) contains supplementary material, which is available to authorized users.


Introduction
Nasal high-frequency oscillatory ventilation (nHFOV) is a promising mode of non-invasive respiratory support used in preterm infants [1,2]. Notwithstanding, the clinical significance of nHFOV in the treatment of these patients remains uncertain, and high-quality data from randomised controlled trials (RCTs) is limited [3,4]. Short-term crossover studies and an observational study suggested that nHFOV was more effective than nCPAP in improving the exhalation of CO 2 [5][6][7]. To verify this hypothesis, we intended to evaluate whether starting nHFOV immediately after extubation would reduce the partial pressure of carbon dioxide (pCO 2 ) at 72 h after extubation in comparison with nasal continuous positive airway pressure (nCPAP) in an RCT of very low birth weight (VLBW) infants < 32 weeks' gestational age (GA). This RCT started in 2015, but had to be terminated prematurely due to slow patient recruitment.
In the present research note, we reported the limited information available on study feasibility, outcomes and adverse events. More importantly, we aimed to provide information about trial conduct and recruitment efforts, as these information could help future researchers learn from the prior experience of our terminated trial [8].

Study population
The study participants were recruited at a tertiary referral centre admitting about 170 VLBW infants annually. Infants at < 32 0 / 7 weeks' GA with a birth weight < 1500 g were eligible if they had received endotracheal mechanical ventilation for ≥ 120 h, were deemed ready for extubation, and fulfilled the extubation criteria: (1) caffeine treatment according to unit guidelines (2) pCO 2 < 65 mmHg with a pH > 7.2 (3) fraction of inspired oxygen (F i O 2 ) of 25-40% to maintain peripheral oxygen saturation (SpO 2 ) at 90-94% (4)  Infants were excluded if they were participants in another RCT, had been treated with hydrocortisone, or exceeded 28 days of chronological age. Other exclusion criteria were major congenital malformation requiring surgery, duct-dependent congenital heart disease, or neuromuscular disease.

Randomisation
The eligible infants were randomly assigned to receive nHFOV (intervention group) or nCPAP (control group). An independent statistician and a study nurse performed random sequence generation using a 1:1 ratio and variable block sizes. Allocation concealment was ensured by numbered opaque envelopes opened immediately prior to extubation.

Study intervention
After extubation, appropriately sized binasal prongs were applied to the patient (Infant Nasal Prongs, Fisher & Paykel Healthcare Ltd., Auckland, New Zealand) and connected to a heated humidifier (MR850, Fisher & Paykel) and a neonatal ventilator (VN500, Drägerwerk AG, Lübeck, Germany, or Leoni Plus, Heinen & Löwenstein, Bad Ems, Germany) using the manufacturerapproved heated-wire ventilatory circuit. The ventilator remained the same before and after extubation. In the intervention group, nHFOV was provided (frequency 10 Hz; amplitude 20 cm H 2 O; I:E ratio 33:66; P mean 8 cm H 2 O). However, a minimum frequency of 9 Hz, maximum amplitude of 30 cm H 2 O, and maximum P mean of 8 cm H 2 O were applied. In the control group, nCPAP was provided (CPAP level 8 cm H 2 O; flow 7 l/min). The maximum CPAP level was 8 cm H 2 O and the maximum flow 8 l/min. In both groups, the F i O 2 was set to maintain SpO 2 at 90-94%. Within the limits, the physician was free to adjust ventilator settings.

Rescue treatment
Treatment failure of either nHFOV or nCPAP was defined as meeting at least one of the following criteria: (1) a sustained pCO 2 > 80 mmHg and pH < 7.20 confirmed by arterial or capillary blood gas analysis despite maximum ventilator support (2) an F i O 2 > 0.6 to maintain SpO 2 at 90-94% (3) reintubation.
If the nCPAP infants developed treatment failure but did not need immediate reintubation, nHFOV was provided as 'rescue treatment' . In the nHFOV group, any non-invasive rescue treatment could be provided.

Primary and secondary outcomes
The primary outcome was the pCO 2 at 72 h after extubation. The secondary outcomes were pH, partial pressure of oxygen, and base excess at 72 h after extubation, blood gas analysis results at 2 h after extubation, successful extubation within 72 h, airway obstruction due to highly viscous secretions within 72 h, treatment failure within 7 days, reintubation within 7 days, total duration of mechanical ventilation, and total duration of supplemental oxygen until discharge. Other secondary outcomes included rescue treatment and incidences of common adverse effects of prematurity.

Sample size calculation, data monitoring, descriptive statistics
Assuming a variability of the pCO 2 , as previously reported for difficult-to-wean preterm infants in our unit [7], and a treatment failure rate of 22% within 72 h after extubation, we calculated a sample size of 34 patients in each study arm to detect a difference in the pCO 2 of 7 mmHg using a two-sided significance of 0.05 and a power of 0.8. No predefined standards existed for periodical data review or early termination of the trial, and no Data and Safety Monitoring Board (DSMB) was instituted. The categorical data was described in numbers and percentages, the quantitative data as medians and ranges.

Results
The nHFOV trial was halted on 31 December 2017 due to slow recruitment. Within three years, only six patients had been included in the study.
From January 2015 to December 2017, a total of 407 VLBW infants < 32 weeks' GA survived to 120 h of age and were assessed for eligibility. The reasons for their exclusion are detailed in Fig. 1. Most infants were not eligible for the study because they had either never been endotracheally ventilated or had been but for < 120 h.
Among the ventilated patients, 92% (211/229) received surfactant. Patients excluded due to hydrocortisone treatment had a high risk of mortality and morbidity (see Additional file 1). Notably, 39 infants were not included due to 'other' reasons that were mostly not specified in the patients' records. We are aware that these reasons comprised participation in another RCT, failure to assess study eligibility, failure to approach the parents for consent within the window of opportunity, lack of skilled translators, parental disapproval of study participation, failure to comply with the complex extubation criteria, and accidental extubation. Table 1 shows the patient characteristics. Two patients suffered from an intraventricular haemorrhage (IVH) 3°-4° prior to randomisation and happened to be randomised to the nCPAP group. All study patients survived to discharge. In the nHFOV group, three out of four patients were successfully extubated, but one required rescue treatment (P mean was increased to 10 cm H 2 O). In the nCPAP group, both patients had to be reintubated within 72 h. The primary outcome, pCO 2 72 h after extubation, could therefore only be assessed in the nHFOV group and was 65 (52-79) mmHg. Selected secondary outcomes are shown in Table 2. Apart from the pre-existing IVH 3°-4° cases, there were no obvious differences in any adverse outcomes. The small number of study patients in each group precluded any formal statistical comparison.

Discussion
The present RCT failed due to slow patient recruitment. However, we would like to share the experience of our single-centre trial to facilitate the design of future multicentre RCTs about nHFOV.

Feasibility of the trial protocol
Implementation of the study protocol was feasible. The patients included in the trial were at an extremely high risk of re-intubation, with 4/6 (67%) reaching the prespecified criteria of treatment failure and 3/6 (50%) being reintubated. This rate was much higher than anticipated, generated early drop-outs before the primary outcome assessment, and increased the estimated sample size from 34 to 79 patients in each study arm.

Impact of nHFOV on study outcomes
Due to the limited number of study patients, a meaningful outcome assessment was not possible.

Reasons for slow recruitment
The main reason for slow recruitment was new, clinically beneficial treatment strategies to avoid endotracheal mechanical ventilation, such as less-invasive surfactant application (LISA) [9]. By the start of the nHFOV trial in 2015, LISA had just become our routine treatment for VLBW infants with respiratory distress syndrome. Consequently, less patients fulfilled the study eligibility criteria of receiving ≥ 120 h of endotracheal mechanical ventilation and requiring a F i O 2 of 25-40% at extubation. VLBW infants with complicated clinical courses (e.g. IVH 3°-4°, patent ductus arteriosus, abdominal surgery) still required prolonged ventilation and a high F i O 2 . Many of these severely affected patients, however, were not eligible because they had either received hydrocortisone for their first extubation attempt or had not been extubated at any time within the first 28 days of life (Fig. 1). This shows how challenging it is in the era of LISA to target a population of infants at high risk of extubation without including those receiving postnatal steroids. With regard  to the infants for whom the reason for exclusion was not clearly documented, we now estimate that maximising our efforts could have increased the overall recruitment to 10-15 patients in the 2015-2017 period. Nevertheless, we would have still required more than 30 years to include 79 patients in each study arm. The RCT was therefore halted prematurely.

Requirements for future trial protocols
There are several key elements that should be considered in future RCTs of nHFOV. Most importantly, the inclusion criteria should specifically target infants who might benefit from nHFOV. The beneficial effects of nHFOV may be difficult to show in mildly affected infants already stabilised on nCPAP, which may explain why nHFOV was shown to have had no impact on pCO 2 in a recent crossover trial [10]. By contrast, the present pilot RCT showed that targeting high-risk patients would require a multicentric approach. Second, our experience showed that overly complex study inclusion criteria may interfere with recruitment. Pragmatic criteria, for example, the inclusion of all VLBW infants extubated after postnatal steroid treatment, may facilitate the recruitment of high-risk patients in future RCTs.
Third, there is the question of choosing the most appropriate primary outcome. A rise in pCO 2 is a reliable indicator of respiratory fatigue in patients with pulmonary failure. Nobody knows, however, whether the beneficial effects of nHFOV on gas exchange are reflected by the pCO 2 in stable patients as these patients may simply decrease their respiratory efforts to keep the pCO 2 at the previous level. Work of breathing or the pressure-time product would be more sensitive primary outcomes but are difficult to measure during neonatal non-invasive respiratory support [11]. From a clinical point of view, successful extubation or ventilation/perfusion mismatch and bronchopulmonary dysplasia assessed in a standard oxygen reduction test would be more relevant outcomes [12], but these would require larger sample sizes.
Fourth, future trials should assess potential side effects of nHFOV, such as upper airway obstruction due to highly viscous secretions [1]. Notably, a recent in vitro study showed that this effect may occur due to impaired oropharyngeal gas conditioning in the presence of vigorous nHFOV settings [13].
Fifth, future RCTs should institute clear stopping rules and a DSMB to facilitate decisions regarding the continuation, modification, or termination of the trial.
Finally, it should be mentioned that nHFOV may be beneficial in the primary treatment of respiratory distress syndrome [14], but a multicentre RCT of very preterm infants of 24 to 28 weeks' GA would be necessary to investigate this in more detail [15]. NHFOV may also be an effective rescue treatment in infants with nCPAP failure, but again, a multicentric approach would be needed to investigate this hypothesis [16].

Limitations of this RCT
• New treatment strategies to avoid endotracheal ventilation reduced the number of eligible patients. • Complex inclusion criteria interfered with patient recruitment. • The treatment failure rate was higher than expected. • There were no predefined criteria for trial termination and no DSMB. • The trial was halted prematurely due to slow patient recruitment.

Additional file
Additional file 1. Patient characteristics and outcomes of very low birth weight infants < 32 weeks' gestational age ventilated for ≥ 120 h, who were excluded from the study due to hydrocortisone treatment (n = 10).
Authors' contributions HSF, CB, and CC designed the study. HSF collected and analysed the data and drafted the manuscript. CC was the principal investigator. All authors critically revised the manuscript. All authors read and approved the final manuscript.
• fast, convenient online submission

Learn more biomedcentral.com/submissions
Ready to submit your research ? Choose BMC and benefit from:

Ethics approval and consent to participate
The study protocol was approved by the Clinical Ethics Committee of the Charité -Universitätsmedizin Berlin (EA2/011/12). Written informed consent was obtained from all the study patients' legal guardians prior to inclusion in the trial.