Discussion
We compared VL with DL for all intubation encounters in the neonatal intensive care setting. In this study, the primary outcome, the first-pass success rate, did not differ significantly between the two groups, yet there was a trend in favour of VL. This study was planned under the assumption that VL would have a significant impact on the rate of successful first attempts. Based on previous studies and own experience, an improvement of 30% was expected.7 11 However, our results suggest that a potential positive effect of VL is significantly lower than what had been assumed. This study was not sufficiently powered to demonstrate that the observed trend in favour of VL was not random. A recent study investigating video versus DL for neonates and infants scheduled for elective surgery found a difference in first-pass success rates of approximately 10%.13 To substantiate possible beneficial effects of VL, larger sample sizes will be required that enable the detection of smaller, yet clinically significant, beneficial effects of VL compared with DL with adequate power.
Our study nevertheless provides further evidence that VL is a feasible primary approach to tracheal intubation in critically ill term and preterm infants.14 Intubation is a highly complex procedure. The success of intubation at first attempt does not only depend on the intubation method (choice of laryngoscope) but also on a variety of other factors. Huitink and Bouwman summarise the following complexity factors: human factors, experience, location, patient factors, equipment and time pressure.15 This makes it difficult to plan and conduct studies in this area and to interpret and compare study results. Several studies showed that use of VL resulted in an improved first pass success rate when used in educational situations.4 5 7 In our study, the first intubation attempts were mostly performed by residents.6 16 However, not only junior physicians with little intubation experience, but also senior physicians already very experienced in DL had to be trained to perform VL in preparation of this study. This might contribute to the understanding of unexpected and differing study results. Ultimately, it is unclear how much training is required to be able to perform a neonatal intubation (with whatever technique) safely.17 It appears to be critical to have to learn two techniques at the same time.3 17 18 Finally, it remains unclear whether the technique of nasotracheal intubation as performed in this study when compared with orotracheal intubation would affect study outcomes.
The overall frequency of adverse TIAEs, and desaturations and bradycardias did not differ significantly between groups. It was noticeable, however, that oesophageal intubation with concomitant desaturation did never occur in the VL group. Various studies have reported that oesophageal intubation is the most common and severe adverse TIAE.6 7 14 18 This raises the question which TIAEs are important and whether it is more important to prevent certain adverse TIAEs than others. Future studies should classify TIAEs according to their severity. For example, the NAESS (neonatal adverse event severity scale) score could be used as a point of reference.19 Also, the safety of neonatal intubations is difficult to measure and is influenced by many other factors than the choice of laryngoscope, in particular team experience, airway stabilisation measures and premedication.7 20
The median number of intubation attempts and tube malpositioning did not differ between groups and were comparable to other studies.2 16
Though there seemed to be a trend in favour of VL, VL did not significantly improve laryngoscopic view. However, as has been observed in adult studies in anaesthesia, an improved view does not necessarily result in improved first-pass success rates.9 10 The exact reasons for this are still not well understood.
After the first intubation attempt, the randomly allocated method of laryngoscopy was changed frequently in both directions, but ultimately a quarter of the intubation attempts started with DL were ultimately performed with VL. Hence, it appears challenging to randomise on the individual level and to study both techniques simultaneously. Randomisation at the individual level presupposes that both techniques must be trained, kept on hand and mastered ad hoc. The quality of intubations, as a result, may suffer. For future trials, it appears advisable to consider alternative study designs, like a stepped wedge cluster randomised design.
Limitations
The present study is limited in several ways. First, despite allocation concealment and no protocol deviations after allocation concealment, potentially meaningful imbalances in baseline variables cannot be excluded. Second, it is plausible that confounding variables were associated with the primary or secondary outcomes, thus potentially threatening unbiased effect estimation. Third, a larger sample size in a larger effectiveness trial would probably attenuate this limitation. Moreover, a stratified randomisation procedure or alternative trial designs could be used to improve baseline balance of important predictors of the endpoints such as gestational age, weight and prior difficult intubations.