Video analysis of neonatal intubations using video laryngoscopy: a prospective comparison of clinical practice with resuscitation guidelines

Abstract

Background

The Neonatal Resuscitation Program recommends direct laryngoscopy (DL) as the primary method for neonatal intubation. Video laryngoscopy (VL) is suggested as an option, particularly for training novice operators or for intubating infants with difficult airways. The programme outlines specific steps for intubation, including managing the external environment and techniques for visualising key anatomical landmarks. It is unclear whether the DL method can be effectively applied to VL.

Objectives

To determine the degree of adherence to resuscitation guidelines during intubation using VL, and to examine the relationship between guideline adherence and intubation success.

Methods

In a cohort of newborn infants who were intubated with VL, we simultaneously recorded the view obtained with the video laryngoscope and an external view of the procedure with a GoPro video camera, and synchronised the recordings for analysis. In each set of recordings, we assessed infant and operator positions, interventions during the procedure, and the anatomical landmarks visualised.

Results

We assessed 95 intubation attempts in 57 infants (median corrected gestational age: 28 weeks; median weight: 1160 g). Sixty-six of these attempts (69%) were successful. Operators spent more time attempting to insert the endotracheal tube through a visible glottis than locating it. Sixty-six (69%) attempts were performed with an appropriate lift manoeuvre. The vocal cords were visualised in only 58 (61%) attempts, while the glottis was seen in 85 (89%).

Conclusions

Neonatal intubation using VL differed from the technique recommended in resuscitation guidelines. Revised guidelines considering the use of VL may be warranted.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Newborn resuscitation guidelines recommend direct laryngoscopy (DL) as the primary approach for neonatal intubation, with video laryngoscopy (VL) suggested as an option.

• These guidelines do not distinguish between DL and VL techniques.

• Neonatal intubation using VL has been shown to significantly improve first-attempt success and is likely to become the preferred method.

• It is unclear whether the intubation technique outlined for DL can be effectively applied to VL.

WHAT THIS STUDY ADDS

• This is the first study to examine neonatal intubation technique using synchronised video recordings of both internal laryngoscope views and the external environment.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The approach to neonatal intubation using VL differed from the technique recommended for DL in current resuscitation guidelines. Revised resuscitation guidelines may be warranted.

• While VL allows for prompt visualisation of the glottis, the time taken to insert the endotracheal tube after visualisation was longer than expected.

Background

Reported first-attempt success rates for neonatal intubation using direct laryngoscopy (DL) have remained low (<50%) and unchanged for almost two decades.¹,⁵ Multiple intubation attempts are associated with infant instability and harm.⁶,¹⁰ Neonatal resuscitation guidelines recommend intubating using DL while video laryngoscopy (VL) is considered beneficial for training novice operators and intubating infants with difficult airways.¹¹,¹³ A recent randomised controlled trial demonstrated a greater first-attempt success rate for neonatal intubation using VL compared with DL, across operators of all experience levels.¹⁴ Consequently, it is likely that VL will be increasingly adopted as the primary method for neonatal intubation.

The Neonatal Resuscitation Program (NRP) serves as the training and education standard for healthcare professionals managing newborn infants in over 130 countries worldwide. It outlines specific steps for intubation, including infant positioning, techniques for visualising key anatomical landmarks and procedural timing. The most recent guidelines outline the procedure using DL.¹¹ We hypothesised that VL may require a different approach than the conventional practices recommended for DL. We investigated the proportion of infants intubated using VL in adherence to NRP guidelines¹¹ and evaluated the relationship between these recommended steps and intubation success.

Methods

Settings and practice

This study was conducted at the National Maternity Hospital (NMH), Dublin, Ireland, a university maternity hospital with approximately 7000 births per year. The study protocol was approved by the local Research Ethics Committee before the first infant was enrolled (EC25.2021) and was authorised for prospective and deferred consent.

Participants

Infants of any gestation who underwent endotracheal intubation in the delivery room (DR) or neonatal intensive care unit (NICU) with VL were eligible for inclusion. Infants with known upper airway anomalies were excluded. Infants could be included more than once.

Study procedure and equipment

We studied infants who were intubated with VL from September 2021. We recorded the view acquired with a C-MAC video laryngoscope (Karl Storz, Tuttlingen, Germany) and straight Miller blade (MIL size #0 or #1). Simultaneously, we recorded intubation attempts with a GoPro camera (Hero9, San Mateo, California, USA) fixed to an incubator or resuscitaire (Fisher & Paykel Healthcare, Auckland, New Zealand; or GE Healthcare, Wauwatosa, Wisconsin, USA) using a clamp mount with an adjustable neck (GoPro, Jaws Flex Clamp, San Mateo, California, USA). The external camera’s field of view included the infant’s entire body, underlying mattress and immediate surroundings (figure 1).

Figure 1. A still image from a rendered video illustrating synchronised views during intubation: one from the video laryngoscope and the other from the GoPro video camera, capturing internal and external perspectives simultaneously.

Video analysis

We used iMovie software (Mac OS X, Apple, California) to merge both recordings into a single video. The laryngoscope footage was scaled down and overlaid onto the corresponding GoPro video, resulting in one combined recording (figure 1). We synchronised the videos to run simultaneously, from laryngoscope blade insertion, offering real-time perspectives of both the internal and the external environment.

Outcomes

Operators were trained to perform intubation according to NRP guidelines,¹¹ and the study outcomes were selected based on these steps. Each video was assessed against this standard using a predefined data collection form. We collected the following outcomes and compared successful intubations with unsuccessful attempts: location of the C-MAC monitor (positioned either to the right or the left of the operator; note that when placed to the right, the monitor extends more centrally across the resuscitaire), bed height, infant alignment (head and neck in line with the body), head stabilisation by the operator and the laryngoscope hold (operator’s thumb resting on the upper surface of the handle). We noted the presence of a gastric tube, whether cricoid pressure was applied with the assistant’s thumb and index finger, and if a correct lift manoeuvre was performed (the operator lifts the laryngoscope in the direction the handle is pointing). We also observed if there was a stylet in the endotracheal tube (ETT) and whether the ETT was introduced from the right side of the mouth or straight down the blade’s open channel. The visualisation of anatomical landmarks, including the epiglottis, glottis and vocal cords, was recorded.

To objectively classify the laryngeal view, we used a modified version of the four-grade Cormack-Lehane (CL) scoring system as described by Yentis and Lee,¹⁵ illustrated in figure 2. We classified grade I and IIa as views suitable for intubation, indicating an unrestricted view of the glottis. We observed whether a suitable view was achieved at any point and if cricoid pressure was applied at the time of the best view. At the time of ETT introduction, we noted if the blade tip was in the vallecula, if the glottis was visible, if the view was suitable and the location of the glottis on the monitor. To record the glottic location at the time of ETT introduction, we included a diagram of the monitor in the data collection form, dividing it into four quadrants (online supplemental figure 1). The quadrant containing the majority of the glottis was recorded, and if it was bisected by the midline, this was noted. We recorded if the view was obscured by glare from secretions, fogging or blood, if suction was applied and whether the view improved after suctioning. Challenges such as laryngospasm or pulmonary haemorrhage were noted. We collected the time from (a) blade insertion to glottic visualisation, (b) glottic visualisation to ETT insertion through the glottis or abandoned attempt and (c) duration of the attempt, noting whether the attempt was successful.

Figure 2. Description of the Yentis and Lee grading system¹⁵ with corresponding views captured during video laryngoscopy for illustration.

Evaluation of video recordings

Two reviewers (CNC, AC) independently evaluated the videos using a data collection form. In cases of inter-observer disagreements, the relevant video was reviewed together, and any variances were resolved by discussion. If consensus was not reached, a third reviewer (EOC) screened the video, and a decision was made based on the majority opinion.

Intubation procedure

Premedication was administered as per local practice. Operators viewed the airway indirectly on the monitor. An intubation attempt was defined as the insertion of the VL blade into the infant’s mouth, regardless of whether an ETT was introduced. The duration of each attempt was timed from the moment the blade was inserted until it was removed. A colorimetric exhaled carbon dioxide (CO₂) detector (Pedicap, Covidien, Colorado, USA) was used to determine the success of an attempt. All intubation attempts, successful and unsuccessful, in a single intubation procedure were recorded and included for analysis. The guidance provided by supervisors was not standardised.

Statistical analysis

The reported first-attempt success rate using DL is <50%,¹,³ with a mean of two attempts per patient. We planned to recruit a convenience sample of 50 infants and 100 intubation attempts. We analysed data with SPSS V.27 statistical software (IBM, Armonk, New York, USA). We presented categorical variables as proportions and numerical variables as mean (SD) when normally distributed and median (IQR) when the distribution was skewed. Outcome variables were analysed using the Student t-test for parametric comparisons and Mann-Whitney U for non-parametric comparisons of continuous variables, and the χ² test for categorical variables. Data were clustered by intubation attempt. A p value of <0.05 was considered statistically significant. We used the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for reporting observational studies.¹⁶

Results

Seventy-four infants were intubated using VL over a 12-month period from September 2021 to 2022. Sixteen eligible infants were not enrolled in the study (research team not contacted or video recording error), while the parents of one infant declined retrospective consent for the use of their infant’s data. The final analysis included data from 57 infants (median (IQR) corrected gestational age 28 (26–34) weeks and weight at intubation 1160 (720–2450) g), who underwent 95 intubation attempts (table 1). Overall, 82 (86%) intubation attempts took place in the NICU. Senior paediatric trainees with more than 6 months of tertiary neonatal experience performed 44 (46%) intubation attempts. Sixty six of 95 (69%) intubation attempts were successful.

Table 1. Intubation characteristics.

Characteristic
Infant demographics	N=57
Gestational age (weeks)*	27 (25–34)
Birth weight (g)*	1160 (688–2485)
Corrected gestational age at intubation (weeks)*	28 (26–34)
Weight at intubation (g)*	1160 (720–2450)
Female sex	25 (44)
Intubation characteristics	N=95
Location of intubation
Delivery room	4 (4)
Operating theatre	9 (10)
NICU	82 (86)
Pre-oxygenation	82 (86)
Pre-medication	81 (85)
Role of clinician
Junior paediatric trainee†	1 (1)
Senior paediatric trainee‡	44 (46)
Neonatology trainee	29 (31)
Neonatologist	21 (22)

^*Median (IQR).

^†< 6 months tertiary neonatal experience.

^‡ > 6 months tertiary neonatal experience.

NICU, neonatal intensive care unit.

Operators spent more time attempting to insert the ETT through a visible glottis than locating it (table 2). They took a median (IQR) of 22 (12–38) s to locate the glottis and 32 (20–53) s to advance the ETT through the glottis. Sixteen (17%) of 95 intubation attempts were completed within the recommended 30 s timeframe. The infant was centrally aligned in 67 (71%) attempts, their head was stabilised by the operator in 61 (64%) and the table height was optimal in a minority of attempts. Sixty-six (69%) attempts were performed with an appropriate lift manoeuvre, and the laryngoscope was held as prescribed in 77 (81%). While cricoid pressure was applied in 86 (91%) attempts, it was applied in accordance with the NRP in only 15 (16%).

Table 2. Outcomes.

Outcome	All attempts N=95	Successful N=66	Unsuccessful N=29	P value
Duration of attempt(s)*	64 (40–95)	61 (41–89)	70 (36–108)	0.42
Time from blade insertion to glottic visualisation(s)*	22 (12–38)†	23 (13–38)‡	17 (12–37)§	0.70
Time from glottic visualisation to intubation or abandoned attempt(s)*	32 (20–53)†	31 (19–49)‡	45 (22–79)§	0.16
Correct table height	41 (43)	26 (39)	15 (52)	0.26
Monitor on operators right	90 (95)	62 (94)	28 (97)	0.6
Video laryngoscope hold as per NRP	77 (81)	52 (78)	25 (86)	0.40
Infant centrally aligned	67 (71)	46 (70)	21 (72)	0.79
Infants head stabilised by operator only	61 (64)	42 (64)	19 (65)	0.86
Infants head stabilised by operator and instructor	25 (26)	17 (26)	8 (28)	0.85
Gastric tube in situ	51 (54)	32 (48)	19 (59)	0.13
Cricoid pressure applied	86 (91)	62 (94)	24 (83)	0.09
Cricoid pressure applied as per NRP	15 (16)	9 (14)	6 (21)	0.39
Correct lift manoeuvre	66 (69)	54 (82)	12 (41)	<0.001
Stylet in ETT	73 (77)	59 (89)	14/16 (88)¶	0.83
ETT introduced from right side of mouth	7 (7)	6 (9)	1/16 (6)¶	0.72
ETT introduced down blade’s open channel	75 (79)	60 (91)	15/16 (94)¶	0.72
At any point
Epiglottis visualised	83 (87)	58 (88)	25 (86)	0.82
Glottis visualised	85 (89)	65 (98)	20 (69)	<0.001
Vocal cords visualised	58 (61)	47 (72)	11 (38)	0.002
Glottis visualised and in midline	67 (71)	52 (78)	15 (52)	<0.001
Suitable view for intubation	78 (82)	63 (95)	15 (52)	<0.001
Cricoid pressure at time of best view	67 (71)	50 (76)	17 (59)	0.92
At ETT introduction
Blade tip in vallecula	41 (43)	27 (41)	14 (48)	0.5
Glottis visible	62 (65)	53 (80)	9 (31)	<0.001
Suitable view for intubation	62 (65)	53 (80)	9 (31)	<0.001
Glottic location on monitor at ETT introduction
Not visible	20 (21)	13 (20)	7 (24)	0.63
Left upper quadrant	1 (1)	1 (2)	0 (0)	0.51
Right upper quadrant	19 (20)	17 (26)	2 (7)	0.03
Upper midline	39 (41)	32 (48)	7 (24)	0.03
Centre	3 (3)	3 (4)	0 (0)	0.24
No ETT introduced	13 (14)	0 (0)	13 (45)	<0.001
View obscured by glare (secretions, fogging or blood)	19 (20)	9 (14)	10 (34)	0.02
Suction during attempt	49 (52)	35 (53)	14 (48)	0.67
Warranted	37 (39)	29 (44)	8 (28)	0.13
View improved	34 (36)	26 (39)	8 (28)	0.27
Laryngospasm or pulmonary haemorrhage	11 (12)	3 (5)	8 (28)	0.001

P values compare successful and unsuccessful attempts. Number (%) unless otherwise stated.

^*Median (IQR).

^†Data available for 85 intubation attempts. Glottis not visualised at any point in 10 attempts.

^‡Data available for 65 successful attempts. Glottis not visualised in 1 successful attempt.

^§Data available for 20 unsuccessful attempts. Glottis not visualised in 9 unsuccessful attempts.

^¶ETT introduced in 16 out of 29 unsuccessful attempts.

ETT, endotracheal tube; NRP, Neonatal Resuscitation Program.

The view frequently changed during intubation attempts. A suitable view was achieved in 78 (82%) attempts; however, once the ETT was introduced, an appropriate view was maintained in only 62 (65%) attempts. The view deteriorated at the point of ETT introduction in 27 (28%) attempts: 13 by paediatric trainees, 9 by neonatal fellows and 5 by consultants. The vocal cords were visualised in 58 (61%) attempts, while the glottis was seen more frequently in 85 (89%) attempts. In a minority of attempts, the ETT was inserted from the right side of the infant’s mouth; more commonly, it was inserted down the blade’s open channel. In 41 (43%) attempts, the blade tip was positioned in the vallecula and the glottic opening was commonly located in an upper portion of the monitor (59 (62%) attempts). The operator’s view was obscured by glare in 19 (20%) attempts, and laryngospasm or pulmonary haemorrhage were encountered in 11 (12%).

No differences were observed between successful and unsuccessful attempts in infant alignment, laryngoscope hold, application of cricoid pressure or whether the ETT was inserted straight down the groove of the blade (table 2). Compared with unsuccessful attempts, operators in successful intubations were more likely to perform the correct lift manoeuvre (54/66 (82%) vs 12/29 (41%), p<0.001), achieve a suitable view (63/66 (95%) vs 15/29 (52%), p<0.001) and visualise the vocal cords (47/66 (72%) vs 11/29 (38%), p=0.002) and glottis in the midline (52/66 (78%) vs 15/29 (52%), p<0.001). The duration of the intubation attempts was comparable in both successful and unsuccessful attempts.

Discussion

This study explores the technique of neonatal intubation using VL and compares clinical practice with resuscitation guidelines. We observed that several steps outlined in the NRP, as taught to clinicians locally, were not followed. These guidelines were primarily developed for DL and may need adjustments when applied to VL. Our study builds on previous observational research that analysed VL recordings of the neonatal airway in isolation, initially focusing on unsuccessful intubations¹⁷ and more recently examining the relationship between VL technique and intubation success.¹⁸ External video recordings have also been used to evaluate different aspects of DR management.¹⁹,²² To our knowledge, this study is the first to synchronise airway views with the external environment during neonatal intubations.

Video laryngoscopy is a valuable tool that facilitates interactive supervision during intubation²³ and increases first-attempt success rates for practitioners at all experience levels.^{14 24 25} Despite VL being available in three-quarters of neonatal units in Ireland and the UK, it is used as a first-line device by only 7%²⁶ and 31%,²⁷ respectively. This is similar to the rates of first-line use documented in a subset of intubations collected as part of the NEAR4NEOS multi-centre registry for neonatal intubation.⁵ Following the recent study by Geraghty et al,¹⁴ the use of VL in neonatal intubation is likely to increase. To support this transition, a better understanding of VL techniques in neonatal intubation is important.

The optimal positioning of the glottis on the video laryngoscope’s monitor remains uncertain. While we hypothesised that positioning the glottis in a central ‘bulls-eye’ location may be relevant, this was observed in only 4% of attempts. The correct lift manoeuvre was executed in only 69% of attempts. Video laryngoscopy provides a full view of the glottis without requiring alignment of the oral and tracheal axes,^{28 29} which is typically obtained with the lift manoeuvre during DL. The enhanced visualisation provided by VL may paradoxically complicate ETT passage, as suggested by the extended time operators spent directing the ETT through the glottis after locating it.

To avoid obstructing the view during DL, the NRP recommends introducing the ETT from the side of the mouth, outside the blade. We observed that in most attempts, operators inserted the ETT down the blade’s open channel, as this approach does not obstruct the view with VL. This technique, however, may limit the amount of space the operator has to angle the ETT towards the glottis if an adequate lift is not performed. Although a stylet was used in the majority of intubation attempts, caution is required if it introduces curvature to the ETT, particularly when advancing it through the blade’s straight channel.

We found that in 29% of all intubation attempts, the infant was misaligned. Infant misalignment suggests that the intubation team may have been overly focused on the monitor, indicating a potential gap in situational awareness during VL. We noted that in most intubation attempts, the best view was achieved with cricoid pressure, which commonly deviated from the recommended technique. Further investigation is needed to develop more effective practice. While we acknowledge that the NRP outlines the DL procedure, we suggest emphasising the importance of ensuring the team member applying cricoid pressure has an unobstructed view of the monitor during VL to control the extent and direction of the pressure.

We observed that the vocal cords were not visualised at any point in 28% of successful intubations, although the glottis was frequently seen. This finding is noteworthy as guidelines typically recommend introducing the ETT after visualising the cords.¹¹ We suggest that operators prioritise visualising the glottis instead, as it was identified in 98% of successful intubations. In our study, only 17% of intubation attempts were completed within the 30 s timeframe recommended by the NRP.¹¹ This is consistent with previous studies,^{14 18 24} suggesting that achieving intubation within this timeframe may not be realistic.

Strengths and limitations

We developed a novel approach to assess neonatal intubation by synchronising internal airway views with external video recordings. We captured intubations immediately after birth in the DR and emergently in the NICU, including cases with and without pre-medication. Two clinicians independently reviewed the videos, with a third reviewer available to resolve any disagreements, thereby strengthening the validity of the results.

This study was performed at a single centre with only one type of video laryngoscope. The number of unsuccessful intubation attempts was small, which was unexpected given the anticipated first-attempt success rate. Video analysis could not be blinded to intubation outcomes, which may have introduced observer bias into the comparative analysis. Each intubation attempt was analysed independently, without considering the destabilising effects of any preceding unsuccessful attempts. We also used an airway grading system originally developed for DL, which may not be directly applicable to VL.

Conclusion

Neonatal intubation performed in clinical practice differed from resuscitation guideline recommendations. Video laryngoscopy may require a different approach compared with what is conventionally taught and recommended for DL. With the anticipated rise in the availability and use of VL in neonatology, there is a need for resuscitation guidelines and educational resources that can effectively adapt to these evolving practices.

Data availability statement

Data are available upon reasonable request.