Video versus direct laryngoscopy to improve the success rate of nasotracheal intubations in the neonatal intensive care setting: a randomised controlled trial

Study design and setting

The trial was performed between February 2020 and August 2021 in the DR and NICU at the University Medical Centre Mainz, Germany. This trial was a single-centre, parallel group, RCT. Intubation encounters were randomly assigned to either VL (intervention group) or DL (control group) at first attempt. Intubation encounters were allocated in a 1:1 ratio using a computer-generated allocation sequence with blocks of varying lengths and sequentially numbered opaque envelopes. Randomisation was stratified for the DR and the NICU. Given the nature of the intervention blinding of parents, medical staff and outcome assessors was not feasible.

Participants

Neonates <44^0/7 weeks postmenstrual age in whom tracheal intubation was indicated either in DR or in the NICU were eligible for inclusion in the study. Multiple births were allocated as individuals. Exclusion criteria were as follows: (1) parental decline of participation, (2) equipment failure, (3) intubation outside the DR or the NICU or (4) clinicians’ preference for either of the two methods.

Primary and secondary outcome measures

The primary endpoint was the first-pass success rate, that is, the passage of the tracheal tube through the vocal cords during the first insertion of the laryngoscope, confirmed by three criteria: Auscultation, chest elevation and chest X-ray.

Secondary outcomes were as follows: (1) the frequency of adverse TIAEs. Adverse TIAEs comprised death, cardiopulmonary resuscitation, administration of adrenaline, airway injury/bleeding, chest wall rigidity, vomiting, oesophageal intubation with or without concomitant oxygen desaturation to less than 80%, treatment of arterial hypotension, treatment of pain or discomfort, intraventricular haemorrhage, pneumothorax, Intubation of a main bronchus, difficult bag mask ventilation, use of additional equipment required or change of classification from urgent to emergency intubation and (2) moderate bradycardias (heart rate less than 100/min) and moderate oxygen desaturations to less than 80%; (3) severe bradycardias (heart rate less than 60/min) and severe oxygen desaturations to less than 60%; (4) number of attempts until successful intubation; (5) tracheal tube malpositioning on the first X-ray following intubation; (6) optimal view of the larynx (Cormack-Lehane classification system grade 1)9 10 and (7) frequency of cross-over from VL to DL and vice versa.

Informed consent

Parents were approached by members of the medical team. Study rationale and process were explained in an understandable manner. Followingly, parents received additional written information. Informed consent was given orally and written. Parents could decline participation or end trial participation prematurely without experiencing any disadvantages in clinical care. Different approaches of informed consent were possible: full prospective informed consent, either (1) antenatally, or (2) following admission to the NICU, or (3) using a deferred consent approach. The deferred consent approach was used since tracheal intubations can occur unexpectedly after birth. Neonates, requiring emergent (intubation required to secure airway immediately) or urgent intubation (intubation required to secure airway promptly) within the first 24 hours of life were randomised and allocated to DL or VL without prior parental consent. Parents were then approached within 24 hours following allocation to explain the nature of the trial, their infant’s probable participation and were requested for retrospective consent. If the parents declined to participate in the study, the infant’s data were not included in the analysis.

Trial procedures

Before the start of the study, all residents, fellows and neonatologists were trained in the use of VL and DL. VL was performed using Infantview laryngoscope (ACUTRONIC Medical Systems AG, Switzerland) with Miller blades sizes 0 or 1. DL was performed using Saling micro blade size 00, Miller blades sizes 00, 0 and 1 or Macintosh blades size 0 and 1 (Proact Metal Max+Combi). When VL was used, it was standard practice, for both, the intubating person, and the supervisor, to view on the screen during intubation. All tracheal intubations were performed according to a local standard operating procedure. Except for emergency intubations, premedication was routinely administered in the form of analgesia (fentanyl 2 µg/kg) and sedation (midazolam 0.1 mg/kg) according to the local standard protocol. A muscle relaxant (vecuronium 0.1 mg/kg) was administered if intubation conditions were considered inadequate after analgesia and sedation. As is common in many German academic hospitals, most initial attempts were performed by paediatric residents. If the intubation was not successful at first attempt, crossover was allowed to VL or DL, respectively. Changing the intubating person was also allowed.

Data on infant and provider characteristics, as well as primary and secondary outcomes, were assessed immediately following the intubation encounter in writing using standardised forms that had been previously developed and used.7 After a double-checking, data were transferred to an electronic database.

Sample size estimation and statistical analysis

The sample size was calculated based on success rates previously reported in non-randomised trials.11 To detect an increase in the first-attempt success rate (primary outcome) from 50% in the DL group (control) to 80% in the VL group (intervention) with 80% power and a significance level of 5%, 78 intubation encounters were required. Assuming non-compliance and an expected crossover rate of 5% in each group, the sample size was finally estimated at 90 intubation encounters.

Descriptive statistics were used to present baseline characteristics of infants and intubation encounters with nominal data provided as numbers and percentages and quantitative data as medians and iIQRs. We used intention-to-treat analysis for all endpoints. For binary endpoints, generalised linear mixed models with a fixed effect for study arm and a random effect for neonates were fitted using a logit link and binomial distribution. For count outcomes, we fitted mixed-effect negative binomial regression models with a fixed effect for study arm and a random effect for neonates. Results are provided as OR and rate ratios (RR) with associated 95% CIs, respectively. A two-sided p<0.05 was considered statistically significant. All analyses were performed using Stata V.15 (StataCorp).

Monitoring and regulatory issues

Reporting in this article is based on the CONSORT (Consolidated Standards of Reporting Trials) 2010 Statement for reporting parallel group randomised trials.12

Patient and public involvement

Patient or public representatives were not involved in the planning of this trial. Eligible parents were given detailed verbal and written information about the study.

Sample characteristics

Of 121 intubation encounters assessed for eligibility, 32 (26.4%) were either not randomised (emergencies (n=9), clinicians’ preference for either VL (n=8) or DL (n=2)) or excluded from the analysis (declined consent (n=13)). Finally, 89 intubation encounters (41 in the VL and 48 in the DL group) in 63 patients were analysed. Figure 1 illustrates the flow diagram of patient inclusion and randomisation. Primary outcome data were complete for all intubation encounters included in the study.

Request permission

The characteristics of participating infants are shown in table 1. The main indication for intubation was respiratory failure associated with oxygenation difficulties or respiratory acidosis. Thirty-six intubation encounters occurred in the DR and 53 in the NICU. In this study, intubations were predominantly performed in critical clinical situations. In almost every 10th intubation encounter, the patient could not be adequately stabilised before intubation. The most frequent indication for intubation was oxygenation failure in both groups, followed by tachydyspnoea and surfactant administration which is in line with other studies. No intubations due to accidental extubations occurred. The use of premedication during intubation was similar in both groups, but many infants were intubated without receiving muscle relaxants. There were no technical limitations while using the laryngoscopy devices. The smallest infants weighed 415 g when intubated with VL and 490 g when intubated with DL. Detailed characteristics of intubation encounters are summarised in table 2.

Primary and secondary outcomes

First-attempt success rate was 48.8% (20/41) in the VL group compared with 43.8% (21/48) in the DL group (OR 1.22, 95% CI 0.51 to 2.88).

The frequency of any adverse TIAE was 43.9% (18/41) and 47.9% (23/48) in the VL and DL group, respectively (OR 0.85, 95% CI 0.37 to 1.97). Table 3 summarises the characteristics of adverse TIAEs by intervention arm. While the frequency of adverse TIAE was comparable between the study arms, oesophageal intubation with concomitant desaturation never occurred in VL group, but in 18.8% (9/48) of intubation encounters in the DL group.

Moderate bradycardias (heart rate <100/min) or oxygen desaturations to less than 80% occurred in 29.3% (12/41) and 43.9% (18/41) of intubation encounters in the VL group compared with 31.3% (15/48) and 45.8% (22/48) in der DL group (OR 0.79, 95% CI 0.25 to 2.46 and OR 0.89, 95% CI 0.35 to 2.28). Severe bradycardias (heart rate <60/min) or oxygen desaturations to less than 60% occurred in 9.8% (4/41) and 29.3% (12/41) of intubation encounters in the VL group compared with 8.3% (4/48) and 22.9% (11/48) in der DL group (OR 1.03, 95% CI 0.18 to 5.86 and OR 1.43, 95% CI 0.48 to 4.27).

There was no difference in the mean number of attempts until successful intubation (2.2 attempts in both groups, RR 1.01, 95% CI 0.73 to 1.39).

Laryngoscopic view was optimal in 62.5% (25/40) and 45.8% (22/48) of first intubation attempts in the VL and DL group, respectively (OR 1.97, 95% CI 0.84 to 4.63).

Tube insertion depth at first X-ray was correct in 85.4% (35/41) and 79.2% (38/48) in the VL and DL group, respectively (OR 1.54, 95% CI 0.51 to 4.66).

In the VL group, 14.6% (6/41) of intubations were ultimately performed with DL, while in the DL group, 25.0% (12/48) of intubations were ultimately performed with VL (OR 0.51, 95% CI 0.17 to 1.52).

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.